US20190317812A1 - Guaranteed quality of service in cloud computing environments - Google Patents
Guaranteed quality of service in cloud computing environments Download PDFInfo
- Publication number
- US20190317812A1 US20190317812A1 US16/385,442 US201916385442A US2019317812A1 US 20190317812 A1 US20190317812 A1 US 20190317812A1 US 201916385442 A US201916385442 A US 201916385442A US 2019317812 A1 US2019317812 A1 US 2019317812A1
- Authority
- US
- United States
- Prior art keywords
- compute node
- workload
- data
- utilization levels
- compute
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Abandoned
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5003—Managing SLA; Interaction between SLA and QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5003—Managing SLA; Interaction between SLA and QoS
- H04L41/5019—Ensuring fulfilment of SLA
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
- H04L43/0805—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability
- H04L43/0817—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability by checking functioning
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H04L67/32—
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/60—Scheduling or organising the servicing of application requests, e.g. requests for application data transmissions using the analysis and optimisation of the required network resources
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/22—Indexing; Data structures therefor; Storage structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
Abstract
Systems, methods, apparatuses, and computer-readable media for guaranteed quality of service (QoS) in cloud computing environments. A workload related to an immutable log describing a transaction may be received. A determination is made based on the immutable log that a first compute node stores at least one data element to process the transaction. Utilization levels of computing resources of the first compute node may be determined. Utilization levels of links connecting the first compute node to the fabric may be determined. A determination may be made, based on the utilization levels, that processing the workload on the first compute node satisfies one or more QoS parameters specified in a service level agreement (SLA). The workload may be scheduled for processing on the first compute node based on the determination that processing the workload on the first compute node satisfies the one or more QoS parameters specified in the SLA.
Description
- This application claims the benefit of U.S. Provisional Application No. 62/658,060, filed Apr. 16, 2018. This application also claims the benefit of U.S. Provisional Application No. 62/663,477, filed Apr. 27, 2018. The aforementioned provisional applications are incorporated by reference herein in their entirety.
- Embodiments herein generally relate to cloud computing, and more specifically, to providing guaranteed quality of service in cloud computing environments.
- Cloud computing services generally provide computing resources to a plurality of different customers. One challenge to cloud computing providers is the ability to ensure that Quality of Service (QoS) offered to customers meets guaranteed levels of QoS (e.g., in a service level agreement). Often, the vast amount of data flowing through the cloud computing environment poses significant challenges to delivering QoS.
- Embodiments disclosed herein provide systems, methods, articles of manufacture, and computer-readable media for guaranteed quality of service (QoS) in cloud computing environments. A workload comprising an immutable log describing a transaction may be received. A determination is made based on the immutable log that a first compute node stores at least one data element to process the transaction. Utilization levels of computing resources of the first compute node may be determined. Utilization levels of links connecting the first compute node to the fabric may be determined. A determination may be made, based on the utilization levels, that processing the workload on the first compute node satisfies one or more QoS parameters specified in a service level agreement (SLA). The workload may be scheduled for processing on the first compute node based on the determination that processing the workload on the first compute node satisfies the one or more QoS parameters specified in the SLA.
-
FIG. 1 depicts hardware components of a data transmission network. -
FIG. 2 depicts an example network including an example set of devices communicating with each other over an exchange system. -
FIG. 3 illustrates an embodiment of an example system. -
FIG. 4 illustrates an embodiment of exemplary logic and data flows through a computing architecture. -
FIG. 5 illustrates an embodiment of a first logic flow. -
FIG. 6 illustrates an embodiment of a neural network. -
FIG. 7 illustrates an embodiment of an exemplary computing architecture. -
FIG. 8 illustrates an embodiment of an example system. -
FIG. 9 illustrates an embodiment of an example system. -
FIG. 10 illustrates an embodiment of an example system. -
FIG. 11 illustrates an embodiment of a second logic flow. -
FIGS. 12A-12B depict embodiments of a database system. -
FIGS. 12C-12D depicts embodiments of data structures in the database system. -
FIG. 13 depicts an embodiment of changes to a derived database object over time periods. -
FIG. 14 illustrates an embodiment of a third logic flow. - Embodiments disclosed herein provide techniques to support predictable quality of service (QoS) guarantees in cloud computing platforms. Massive amounts of data may be stored in a mass storage system using a hierarchy of caches to provide predictable performance. Performance may be defined using service level agreements (SLAs) which provide for a predictable amount of jitter. Embodiments disclosed herein may provide predictable performance using a non-blocking matrix switch that enables the transfer of large amounts of data with guaranteed jitter and predictability (as may be defined in the SLAs).
- Programs may operate against the data, which may be stored using a flat model that is broken into small workloads that can be divided between the various hardware compute nodes. Because of the massive amount of data that can be stored, ongoing and/or recurring calculations, such as generating monthly financial statements, can be calculated based on the totality of data rather than persisting intermediary reports (e.g., using October end-of-month statements as the starting point for November operations). Some calculations may be made with limited accuracy, such that a user can request an answer within a short period of time (e.g., one second), at a certain accuracy (e.g., 99%), or both at a very high cost.
- Generally, a workload to be processed may be received by the cloud computing platform which includes a plurality of hardware compute nodes communicably coupled via a fabric. The workload may be related to an immutable log for one or more transactions. A scheduler may determine where to place the workload for processing based on one or more heuristics that will meet the parameters defined by the SLA. For example, if a first compute node stores data needed to process the workload, the scheduler may determine to place the workload on the first compute node. As another example, if the first compute node is not suitable to accept the workload (e.g., because the computing resources of the first compute node are being utilized to process other workloads), the scheduler may place the workload on a second compute node that is proximate to the first compute node, thereby facilitating faster access to the needed data that is stored on the first compute node. As yet another example, if the communications link between the first and second compute nodes is saturated (and/or used at a level that exceeds a utilization threshold), the scheduler may place the workload on a third compute node that is proximate to the first compute node, where the communications link between the first and third compute nodes is not overutilized. Doing so allows the workload to be processed in a manner which satisfies the guarantees specified in the SLA.
-
FIG. 1 is a block diagram that provides an illustration of the hardware components of adata transmission network 100, according to embodiments of the present technology.Data transmission network 100 is a specialized computer system that may be used for processing large amounts of data where a large number of computer processing cycles are required. -
Data transmission network 100 may also includecomputing environment 114.Computing environment 114 may be a specialized computer or other machine that processes the data received within thedata transmission network 100.Data transmission network 100 also includes one ormore network devices 102.Network devices 102 may include client devices that are capable of communicating withcomputing environment 114. For example,network devices 102 may send data to thecomputing environment 114 to be processed, may send signals to thecomputing environment 114 to control different aspects of the computing environment or the data it is processing, among other reasons.Network devices 102 may interact with thecomputing environment 114 through a number of ways, such as, for example, over one ormore networks 108. As shown inFIG. 1 ,computing environment 114 may include one or more other systems. For example,computing environment 114 may include adatabase system 118 and/or acommunications grid 120. - In other embodiments,
network devices 102 may provide a large amount of data, either all at once or streaming over a period of time to thecomputing environment 114 vianetworks 108. For example,network devices 102 may include network computers, sensors, databases, or other devices that may transmit or otherwise provide data to computingenvironment 114. For example,network devices 102 may include local area network devices, such as routers, hubs, switches, or other computer networking devices. These devices may provide a variety of stored or generated data, such as network data or data specific to the network devices themselves.Network devices 102 may also include sensors that monitor their environment or other devices to collect data regarding that environment or those devices, andsuch network devices 102 may provide data they collect over time.Network devices 102 may also include devices within the internet of things (IoT), such as devices within a home automation network. Some of these devices may be referred to as edge devices, and may involve edge computing circuitry. Data may be transmitted by network devices directly tocomputing environment 114 or to network-attached data stores, such as network-attacheddata stores 110 for storage so that the data may be retrieved later by thecomputing environment 114 or other portions ofdata transmission network 100. -
Data transmission network 100 may also include one or more network-attacheddata stores 110. Network-attacheddata stores 110 are used to store data to be processed by thecomputing environment 114 as well as any intermediate or final data generated by the computing system in non-volatile memory. However, in certain embodiments, the configuration of thecomputing environment 114 allows its operations to be performed such that intermediate and final data results can be stored solely in volatile memory (e.g., RAM), without a requirement that intermediate or final data results be stored to non-volatile types of memory (e.g., disk). This can be useful in certain situations, such as when thecomputing environment 114 receives ad hoc queries from a user and when responses, which are generated by processing large amounts of data, need to be generated on-the-fly. In this non-limiting situation, thecomputing environment 114 may be configured to retain the processed information within memory so that responses can be generated for the user at different levels of detail as well as allow a user to interactively query against this information. - Network-attached
data stores 110 may store a variety of different types of data organized in a variety of different ways and from a variety of different sources. For example, network-attached data storage may include storage other than primary storage located withincomputing environment 114 that is directly accessible by processors located therein. Network-attached data storage may include secondary, tertiary or auxiliary storage, such as large hard drives, servers, virtual memory, among other types. Storage devices may include portable or non-portable storage devices, optical storage devices, and various other mediums capable of storing, containing data. A machine-readable storage medium or computer-readable storage medium may include a non-transitory medium in which data can be stored and that does not include carrier waves and/or transitory electronic signals. Examples of a non-transitory medium may include, for example, a magnetic disk or tape, optical storage media such as compact disk or digital versatile disk, flash memory, memory or memory devices. A computer-program product may include code and/or machine-executable instructions that may represent a procedure, a function, a subprogram, a program, a routine, a subroutine, a module, a software package, a class, or any combination of instructions, data structures, or program statements. A code segment may be coupled to another code segment or a hardware circuit by passing and/or receiving information, data, arguments, parameters, or memory contents. Information, arguments, parameters, data, etc. may be passed, forwarded, or transmitted via any suitable means including memory sharing, message passing, token passing, network transmission, among others. Furthermore, the data stores may hold a variety of different types of data. For example, network-attacheddata stores 110 may hold unstructured (e.g., raw) data, such as manufacturing data (e.g., a database containing records identifying products being manufactured with parameter data for each product, such as colors and models) or product sales databases (e.g., a database containing individual data records identifying details of individual product sales). - The unstructured data may be presented to the
computing environment 114 in different forms such as a flat file or a conglomerate of data records, and may have data values and accompanying time stamps. Thecomputing environment 114 may be used to analyze the unstructured data in a variety of ways to determine the best way to structure (e.g., hierarchically) that data, such that the structured data is tailored to a type of further analysis that a user wishes to perform on the data. For example, after being processed, the unstructured time stamped data may be aggregated by time (e.g., into daily time period units) to generate time series data and/or structured hierarchically according to one or more dimensions (e.g., parameters, attributes, and/or variables). For example, data may be stored in a hierarchical data structure, such as a ROLAP OR MOLAP database, or may be stored in another tabular form, such as in a flat-hierarchy form. -
Data transmission network 100 may also include one or more server farms 106.Computing environment 114 may route select communications or data to the one or more severfarms 106 or one or more servers within the server farms.Server farms 106 can be configured to provide information in a predetermined manner. For example,server farms 106 may access data to transmit in response to a communication.Server farms 106 may be separately housed from each other device withindata transmission network 100, such ascomputing environment 114, and/or may be part of a device or system. -
Server farms 106 may host a variety of different types of data processing as part ofdata transmission network 100.Server farms 106 may receive a variety of different data from network devices, from computingenvironment 114, from cloud network 116, or from other sources. The data may have been obtained or collected from one or more sensors, as inputs from a control database, or may have been received as inputs from an external system or device.Server farms 106 may assist in processing the data by turning raw data into processed data based on one or more rules implemented by the server farms. For example, sensor data may be analyzed to determine changes in an environment over time or in real-time. -
Data transmission network 100 may also include one or more cloud networks 116. Cloud network 116 may include a cloud infrastructure system that provides cloud services. In certain embodiments, services provided by the cloud network 116 may include a host of services that are made available to users of the cloud infrastructure system on demand. Cloud network 116 is shown inFIG. 1 as being connected to computing environment 114 (and therefore havingcomputing environment 114 as its client or user), but cloud network 116 may be connected to or utilized by any of the devices inFIG. 1 . Services provided by the cloud network can dynamically scale to meet the needs of its users. The cloud network 116 may comprise one or more computers, servers, and/or systems. In some embodiments, the computers, servers, and/or systems that make up the cloud network 116 are different from the user's own on-premises computers, servers, and/or systems. For example, the cloud network 116 may host an application, and a user may, via a communication network such as the Internet, on demand, order and use the application. - While each device, server and system in
FIG. 1 is shown as a single device, it will be appreciated that multiple devices may instead be used. For example, a set of network devices can be used to transmit various communications from a single user, orremote server 140 may include a server stack. As another example, data may be processed as part ofcomputing environment 114. - Each communication within data transmission network 100 (e.g., between client devices, between a device and
connection management system 150, betweenservers 106 andcomputing environment 114 or between a server and a device) may occur over one ormore networks 108.Networks 108 may include one or more of a variety of different types of networks, including a wireless network, a wired network, or a combination of a wired and wireless network. Examples of suitable networks include the Internet, a personal area network, a local area network (LAN), a wide area network (WAN), or a wireless local area network (WLAN). A wireless network may include a wireless interface or combination of wireless interfaces. As an example, a network in the one ormore networks 108 may include a short-range communication channel, such as a Bluetooth or a Bluetooth Low Energy channel. A wired network may include a wired interface. The wired and/or wireless networks may be implemented using routers, access points, bridges, gateways, or the like, to connect devices in thenetwork 108, as will be further described with respect toFIG. 2 . The one ormore networks 108 can be incorporated entirely within or can include an intranet, an extranet, or a combination thereof. In one embodiment, communications between two or more systems and/or devices can be achieved by a secure communications protocol, such as secure sockets layer (SSL) or transport layer security (TLS). In addition, data and/or transactional details may be encrypted. - Some aspects may utilize the Internet of Things (IoT), where things (e.g., machines, devices, phones, sensors) can be connected to networks and the data from these things can be collected and processed within the things and/or external to the things. For example, the IoT can include sensors in many different devices, and high value analytics can be applied to identify hidden relationships and drive increased efficiencies. This can apply to both big data analytics and real-time analytics. This will be described further below with respect to
FIG. 2 . - As noted,
computing environment 114 may include acommunications grid 120 and a transmissionnetwork database system 118.Communications grid 120 may be a grid-based computing system for processing large amounts of data. The transmissionnetwork database system 118 may be for managing, storing, and retrieving large amounts of data that are distributed to and stored in the one or more network-attacheddata stores 110 or other data stores that reside at different locations within the transmissionnetwork database system 118. The compute nodes in the grid-basedcomputing system 120 and the transmissionnetwork database system 118 may share the same processor hardware, such as processors that are located withincomputing environment 114. -
FIG. 2 illustrates an example network including an example set of devices communicating with each other over an exchange system and via a network, according to embodiments of the present technology. As noted, each communication withindata transmission network 100 may occur over one or more networks.System 200 includes anetwork device 204 configured to communicate with a variety of types of client devices, forexample client devices 230, over a variety of types of communication channels. - As shown in
FIG. 2 ,network device 204 can transmit a communication over a network (e.g., a cellular network via a base station 210). The communication can be routed to another network device, such as network devices 205-209, viabase station 210. The communication can also be routed tocomputing environment 214 viabase station 210. For example,network device 204 may collect data either from its surrounding environment or from other network devices (such as network devices 205-209) and transmit that data tocomputing environment 214. - Although network devices 204-209 are shown in
FIG. 2 as a mobile phone, laptop computer, tablet computer, temperature sensor, motion sensor, and audio sensor respectively, the network devices may be or include sensors that are sensitive to detecting aspects of their environment. For example, the network devices may include sensors such as water sensors, power sensors, electrical current sensors, chemical sensors, optical sensors, pressure sensors, geographic or position sensors (e.g., GPS), velocity sensors, acceleration sensors, flow rate sensors, among others. Examples of characteristics that may be sensed include force, torque, load, strain, position, temperature, air pressure, fluid flow, chemical properties, resistance, electromagnetic fields, radiation, irradiance, proximity, acoustics, moisture, distance, speed, vibrations, acceleration, electrical potential, electrical current, among others. The sensors may be mounted to various components used as part of a variety of different types of systems (e.g., a financial operation). The network devices may detect and record data related to the environment that it monitors, and transmit that data tocomputing environment 214. - In another example, another type of system that may include various sensors that collect data to be processed and/or transmitted to a computing environment according to certain embodiments includes a home automation or similar automated network in a different environment, such as an office space, school, public space, sports venue, or a variety of other locations. Network devices in such an automated network may include network devices that allow a user to access, control, and/or configure various home appliances located within the user's home (e.g., a television, radio, light, fan, humidifier, sensor, microwave, iron, and/or the like), or outside of the user's home (e.g., exterior motion sensors, exterior lighting, garage door openers, sprinkler systems, or the like). For example, network device 202 may include a home automation switch that may be coupled with a home appliance. In another embodiment, a network device can allow a user to access, control, and/or configure devices, such as office-related devices (e.g., copy machine, printer, or fax machine), audio and/or video related devices (e.g., a receiver, a speaker, a projector, a DVD player, or a television), media-playback devices (e.g., a compact disc player, a CD player, or the like), computing devices (e.g., a home computer, a laptop computer, a tablet, a personal digital assistant (PDA), a computing device, or a wearable device), lighting devices (e.g., a lamp or recessed lighting), devices associated with a security system, devices associated with an alarm system, devices that can be operated in an automobile (e.g., radio devices, navigation devices), and/or the like. Data may be collected from such various sensors in raw form, or data may be processed by the sensors to create parameters or other data either developed by the sensors based on the raw data or assigned to the system by a client or other controlling device.
- In another example, another type of system that may include various sensors that collect data to be processed and/or transmitted to a computing environment according to certain embodiments includes a power or energy grid. A variety of different network devices may be included in an energy grid, such as various devices within one or more power plants, energy farms (e.g., wind farm, solar farm, among others) energy storage facilities, factories, homes and businesses of consumers, among others. One or more of such devices may include one or more sensors that detect energy gain or loss, electrical input or output or loss, and a variety of other efficiencies. These sensors may collect data to inform users of how the energy grid, and individual devices within the grid, may be functioning and how they may be made more efficient.
- Network device sensors may also perform processing on data it collects before transmitting the data to the
computing environment 214, or before deciding whether to transmit data to thecomputing environment 214. For example, network devices may determine whether data collected meets certain rules, for example by comparing data or values computed from the data and comparing that data to one or more thresholds. The network device may use this data and/or comparisons to determine if the data should be transmitted to thecomputing environment 214 for further use or processing. -
Computing environment 214 may includemachines environment 214 is shown inFIG. 2 as having two machines, 220 and 240,computing environment 214 may have only one machine or may have more than two machines. The machines that make upcomputing environment 214 may include specialized computers, servers, or other machines that are configured to individually and/or collectively process large amounts of data. Thecomputing environment 214 may also include storage devices that include one or more databases of structured data, such as data organized in one or more hierarchies, or unstructured data. The databases may communicate with the processing devices withincomputing environment 214 to distribute data to them. Since network devices may transmit data tocomputing environment 214, that data may be received by thecomputing environment 214 and subsequently stored within those storage devices. Data used by computingenvironment 214 may also be stored indata stores 235, which may also be a part of or connected to computingenvironment 214. -
Computing environment 214 can communicate with various devices via one ormore routers 225 or other inter-network or intra-network connection components. For example,computing environment 214 may communicate withdevices 230 via one ormore routers 225.Computing environment 214 may collect, analyze and/or store data from or pertaining to communications, client device operations, client rules, and/or user-associated actions stored at one ormore data stores 235. Such data may influence communication routing to the devices withincomputing environment 214, how data is stored or processed withincomputing environment 214, among other actions. - Notably, various other devices can further be used to influence communication routing and/or processing between devices within
computing environment 214 and with devices outside ofcomputing environment 214. For example, as shown inFIG. 2 ,computing environment 214 may include amachine 240, such as a web server. Thus,computing environment 214 can retrieve data of interest, such as client information (e.g., product information, client rules, etc.), technical product details, news, current or predicted weather, and so on. - In addition to
computing environment 214 collecting data (e.g., as received from network devices, such as sensors, and client devices or other sources) to be processed as part of a big data analytics project, it may also receive data in real time as part of a streaming analytics environment. As noted, data may be collected using a variety of sources as communicated via different kinds of networks or locally. Such data may be received on a real-time streaming basis. For example, network devices may receive data periodically from network device sensors as the sensors continuously sense, monitor and track changes in their environments. Devices withincomputing environment 214 may also perform pre-analysis on data it receives to determine if the data received should be processed as part of an ongoing project. The data received and collected by computingenvironment 214, no matter what the source or method or timing of receipt, may be processed over a period of time for a client to determine results data based on the client's needs and rules. -
FIG. 3 illustrates a conceptual overview of asystem 300 that may generally be representative of a distributed cloud-based computing system or another type of computing network in that one or more techniques described herein may be implemented according to various embodiments. As shown inFIG. 3 ,system 300 may generally include computing resources (CRs) 302-y, where y is any positive integer, to compute information and data. Thecomputing resources 302 may include resources of multiple types, such as—for example—processors, co-processors, fully-programmable gate arrays (FPGAs), memory, networking equipment, circuit boards, storage, and other computing equipment. The embodiments are not limited to these examples. - The
computing resources 302 may be included as part of a computer, such as a server, server farm, blade server, a server sled, or any other type of server or computing device, and may be within one ormore racks 304. In embodiments, theracks 304 may be part of one or more data centers 308 and may be coupled with each other via various networking equipment. For example, theracks 304 within a data center 308 may be coupled with each via afabric 303. Thefabric 303 may include a combination of electrical and/or optical signaling media, and high bandwidth interconnects, such as Gigabit Ethernet, 10 Gigabit Ethernet, 100 Gigabit Ethernet, InfiniB and, Peripheral Component Interconnect (PCI) Express (PCIe), and so forth. Further, thefabric 303 may include switching infrastructure, such as switches, routers, gateways, and so forth. Thefabric 303 is configured such that anyrack 304 may send signals to (and receive signals from) eachother racks 304 within a data center 308 to communicate data and information. In embodiments, thefabric 303 may be coupled withnetworking infrastructure 305 such that it enables communication of signals between racks of one data center 308 withracks 304 of another data center 308 to communicate data and information. For example,FIG. 3 illustratesracks 304 of data center 308-1 coupled withracks 304 of data center 308-2 viafabrics 303,networking infrastructure 305, and the cloud-based infrastructure 307. The cloud-based infrastructure 307 illustratively includes acloud controller 309. Although depicted as a component of the cloud-based infrastructure 307, thefabrics 303,racks 304,networking infrastructure 305, and data centers 308 may each include an instance of thecontroller 309. Thecontroller 309 may comprise software, hardware, and/or a combination of software and hardware. - In embodiments, the
networking infrastructure 305 includes networking equipment, such as routers, firewalls, switches, gateways, cabling, and so forth to communicate data and information between a data center 308 and with the cloud-based infrastructure 307 and another data center 308. For example, thenetworking infrastructure 305 may include edge access routers, edge access switches, and edge firewalls capable of communicating with core routers, core switches, and core firewalls of the cloud-based infrastructure 307. The core networking equipment of the cloud-based infrastructure 307 may couple with edge networking equipment of another data center 308 to enable communication between data centers 308. Note that embodiments are not limited in this manner, and thenetworking infrastructure 305 and/or cloud-based infrastructure 307 may include other networking equipment, servers, relays, interconnects, and so forth to enable communication between a data center 308 and other data centers 308. - In one example, the
system 300 may be a distributed cloud-based computing system to provide a financial service platform. Thesystem 300 may process data and information, such as financial data and financial information, to provide financial services, for example. The financial services include, but are not limited to, investment and asset management services, active equity management services, active quantitative equity services, cash fund services, alternatives services, currency management services, index investing services, electronic trading services, multi-asset services, investment research services, investment trading services, accounting services, custody services, fund administration services, outsourcing services, performance measurement services, portfolio analysis services, data analytics services, investment analytics services, benchmark/indices/indicator services, D-as-a-Service (DaaS) services, and so forth. Embodiments are not limited to these examples. To provide these financial services, thecontroller 309 may dynamically pool or compose a plurality of thecomputing resources 302 together within a data center 308 and/or among data centers 308 in the cloud via the cloud-based infrastructure 307. In one example, computingresources 302 of data center 308-1 may be composed with computingresources 302 of data center 308-2 to process data and information, e.g., a workload, to provide a financial service. Once the workload completes, thecontroller 309 may decompose the composedcomputing resources 302 and make thecomputing resources 302 available to process another workload. Note that in embodiments, thesystem 300 may enable multiple instances of pooled or composedcomputing resources 302 to provide data and information processing in parallel and embodiments are not limited in this manner. - In embodiments,
system 300 may be coupled with one or more other systems, such as investment trading systems, banking systems, regulatory systems, risk management systems, performance systems, accounting system, data warehouse systems, financial institution system, and so forth. These other systems may be coupled withsystem 300 via networking, such as thenetworking infrastructure 305 and the cloud-based infrastructure. Embodiments are not limited in this manner. - The
controller 309 may be used to provide access to resources in thesystem 300. For example, thecontroller 309 may control access to thefabrics 303,racks 304,networking infrastructure 305, and data centers 308. Furthermore, when a workload is received for processing in thesystem 300, thecontroller 309 may select one ormore computing resources 302 to process the workload. For example, a workload may be defined by one or more entries in an immutable log. The immutable log may store data that are related to one or more transactions to be processed by thesystem 300. For example, the immutable log may specify an account identifier of an account, an asset (e.g., a stock, ETF, mutual fund, etc.) identifier subject to the transaction, a transaction type (e.g., buy, sell, etc.), and any other data describing the transaction. To process the transaction, thesystem 300 may generally need to determine the current price of the asset, process the transaction according to the determined price, update the account balances, and update the position holdings of the account in the asset. - The
system 300 may process transactions according to one or more service level agreements (SLAs). The SLAs may define quality of service (QoS) parameters that must be fulfilled by thesystem 300 when processing transactions. For example, the QoS parameters may specify guaranteed latency, jitter, bandwidth, transaction processing times, and the like. More generally, however, because thesystem 300 processes large numbers of transactions for a large number of clients, thesystem 300 must attempt to process each transaction as quickly as possible. Therefore, thecontroller 309 is configured to schedule received workloads in thesystem 300 in a manner that satisfies the QoS guarantees in the SLAs. - In some embodiments, the
controller 309 may consider theCRs 302 as an n by m matrix of compute nodes, where n and m are any positive integer. For example, each compute node (physical and/or virtual) may be considered a point in the matrix, and the connections between each point in the matrix correspond to a network link in thefabric 303. Doing so allows thecontroller 309 to consider the matrix when scheduling workloads for processing in thesystem 300 such that the processing conforms with the QoS guarantees in the SLA for a given client. - Generally, to schedule workloads, the
controller 309 may consider a plurality of heuristics. For example, thecontroller 309 may consider utilization levels of thecomputing resources 302. For example, for a given compute node and/or virtual machine executing on a compute node, thecontroller 309 may determine processor (CPU) utilization, memory utilization, storage utilization, network I/O utilization, and utilization of any other resource of the compute node. Generally, thecontroller 309 may selectCRs 302 that have the lowest levels of utilization. Furthermore, thecontroller 309 may consider the utilization of the network links of thefabric 303 connected to eachCR 302. For example, thecontroller 309 may consider the bandwidth utilization, throughput, latency, jitter, and any other aspect of the links of thefabric 303. More generally, thecontroller 309 may select nodes having the links with the lowest levels of utilization. Furthermore, thecontroller 309 may analyze the workload to determine what data stored by thesystem 300 is needed to process the transaction. For example, CRs 302-13 may store the current price of a stock that is specified in an immutable log for the transaction, while CRs 302-1 may store stale price data of the stock (e.g., last month's price of the stock). Therefore, thecontroller 309 may determine to place the workload on or near CRs 302-13 to provide faster access to the pricing data needed to process the trade. Similarly, thecontroller 309 may determine that the workload need not be placed near CRs 302-1, as CRs 302-1 include pricing data that will not be used to process the transaction. Further still, thecontroller 309 may determine whichCRs 302 have the greatest number of links to the needed data. Doing so may ensure that alternate routes to the data exist if one or more other links become saturated. - Therefore, to schedule a workload in compliance with the QoS guarantees, the
controller 309 may determine whichCRs 302 have the lowest levels of resource utilization, whichCRs 302 were least recently used, whichCRs 302 have network links have the lowest levels of utilization, whichCRs 302 store data that is most frequently accessed to process transactions, whichCRs 302 include and/or are located nearest to the data required to process the transaction, and/or whichCRs 302 have the most links in thefabric 303 to access the data required to process the transaction. Furthermore, thecontroller 309 may consider the impact that processing the workload will have on thesystem 300. For example, if placing the workload on CRs 302-5 would saturate the links to CRs 302-5 in thefabric 303, thecontroller 309 may refrain from placing the workload on CRs 302-5. - In some embodiments, the
controller 309 may determine whether the utilization of theCRs 302 and/or the links in thefabric 303 exceed a respective threshold. For example, if the current and/or estimated use of the processors of a compute node is 80% and a processor use threshold is 75%, thecontroller 309 may determine to forego deploying a workload (and/or a portion thereof) to the compute node. As another example, if 70% of the memory of a compute node is currently utilized (and/or estimated to be utilized while processing the workload), and the memory use threshold is 60%, thecontroller 309 may forego deploying a workload (and/or a portion thereof) to the compute node. As another example, a compute node may have 10 network links in thefabric 303 to the data needed to process a transaction. If deploying the workload to the compute node would saturate all 10 links, thecontroller 309 may determine to forego deploying a workload (and/or a portion thereof) to the compute node. Instead, thecontroller 309 may determine to deploy the workload to a compute node that has more links to the needed data and/or links that will not be saturated (and/or utilized beyond a threshold utilization level) by processing the workload. As another example, if a network switch of the compute node is utilized beyond a threshold utilization, thecontroller 309 may determine to forego deploying a workload to the compute node. If, however, the network switch is not utilized beyond the threshold, thecontroller 309 may deploy the workload (and/or a portion thereof) to the compute node. As another example, thecontroller 309 may estimate an amount of time required to process the workload on theCRs 302 in light of the resource and/or fabric utilizations and determine whether the estimated time exceeds a guaranteed processing time in the SLA. If the estimated time to process the workload does not exceed the guaranteed processing time specified in the SLA, thecontroller 309 may deploy the workload (and/or a portion thereof) to theCRs 302. - In some embodiments, the
controller 309 may maintain alog 310 describing each received workload and/or transaction. Thelog 310 may include entries specifying where each transaction is deployed for processing, what data the transaction accessed, and the results of processing each transaction (e.g., whether the amount of time required to process the transaction satisfied and/or violated the QoS guarantees). Thelog 310 may comprise a model which may process a transaction to produce a generate distribution specifying how the processing of the transaction will access data and/or communicate data within thesystem 300. For example, an ETF may comprise a plurality of stocks. Thelog 310 may specify that processing trades of the ETF may require access to data describing each stock in the ETF. Therefore, thecontroller 309 may determine to generate a grouping of the CRs 302 (e.g., one or more compute nodes) to process the ETF transaction. The grouping ofCRs 302 may generally include the shortest number of paths in thefabric 303 to the data needed to process the transaction (e.g., by selectingCRs 302 that are nearest to the data describing each stock in the ETF). However, the grouping ofCRs 302 may be defined by a radius that ensures the links in thefabric 303 will not become oversaturated when processing the ETF transaction. - In some embodiments, the
controller 309 may compute a score for each of theCRs 302 and select one or more of theCRs 302 having the highest score to process a workload (or a portion thereof). The score may generally reflect the mostsuitable CRs 302 for processing the workload in light of the QoS requirements. For example, each score may be compared to threshold. If the score computed for a givenCR 302 does not exceed the threshold, thecontroller 309 may determine to not deploy the workload (or a portion thereof) to theCR 302. If the score exceeds the threshold, thecontroller 309 may deploy the workload (or a portion thereof) to theCR 302. Thecontroller 309 may use any suitable function to compute a score for the workload, where the function considers one or more of the current and forecast utilization of theCRs 309, the current and forecast utilization of the links of thefabric 303 to eachCR 309, the distance (e.g., a number of network hops) of eachCR 309 to the data needed to the workload, and the like. - In some embodiments, the
controller 309 may break up the workload into smaller workloads (or subunits) and schedule each subunit of the workload on one ormore CRs 302 to ensure that each subunit is processed in a manner that satisfies the QoS guarantees. Thecontroller 309 may generally schedule each subunit according to the heuristics described above to ensure that each subunit is scheduled for processing according to the QoS guarantees. Furthermore, in some embodiments, thecontroller 309 may consider a grouping ofCRs 302 when determining to deploy a workload. In such embodiments, thecontroller 309 considers whether the current and/or planned utilization of the groupedCRs 302 exceeds a threshold when deploying workloads. Similarly, thecontroller 309 may consider the number of links in thefabric 303 of the groupedCRs 302 as well as the use of the links in thefabric 303 when determining to deploy a workload to the groupedCRs 302. -
FIG. 4 depicts a block diagram describing exemplary logic and data flows through a computing architecture. -
Data 402 may be received in adata storage 404. Thedata 402 may include discrete units of data and/or one or more data streams (e.g., communication channels that repeatedly provide one or more data units at a given rate). Thedata 402 may include, for example, data relating to an individual user (e.g., a financial client), aggregate data (e.g., reflecting conditions in a market, such as a financial market), cancelations of previously-received data, corrections of previously-received data, etc. In some cases, cancelations and/or corrections may be received that cancels or corrects other data items that have not yet been received (e.g., due to the order in which the data was transmitted or batched, network conditions, data losses, etc.). - The
data storage 404 may include hardware, software, or a combination of hardware and software suitable for storing data. Thedata storage 404 may include one or more data structures, such as thelog 310 or a database. The data structures may be configured to store and organize the data, and/or to facilitate retrieval of the data. In some embodiments, the entries in thelog 310 may be organized chronologically (e.g., in the order in which thedata 402 was received by thedata storage 404, in a time-stamp order of thedata 402, etc.). In some embodiments, thelog 310 is a persistent and/or immutable log which allows individual data records to be written, but not to be directly deleted or changed. In some embodiments, theimmutable log 310 specifies one or more transactions for processing. Thecontroller 309 may schedule the transactions specified in theimmutable log 310 for processing in compliance with QoS guarantees as described above. - In some embodiments, the
data 402 stored in thedata storage 404 may be subjected to one ormore filters 408. Thefilters 408 may include data governance filters which, for example, match one or more rules against thedata 402 and selectively pass thedata 402 to other components in the architecture. - The
data storage 404 and/or the filter(s) 408 may provide information to amachine learning model 410, such as an artificial neural network (ANN). Theunderlying model 410 may be configured to learn associations from patterns in thedata 402, to predict future trends based on historical data observations, and to provide insights into why thedata 402 appears the way that it does. - A
library 414 of microservices 412-i may make use the data (e.g., the raw data stored in thedata storage 404, the filtered data as presented by thefilters 408, information output from themachine learning model 410, or various combinations of these types of data). Each microservice 412-i may represent an atomic computing unit configured to perform a defined task (e.g., computing a value for a financial variable for certain subsets of the data 402). The microservices 412-i may be used individually, or variously combined into macroservices 416-i. The macroservices 416-i may represent more complex operations in which the outputs of various microservices 412-i are combined or otherwise used to perform a specified task. - For instance, one macroservice 416-1 may use the outputs of various microservices 412-i to generate a report 418 (such as a financial report, disclosure form, etc.). In another example, an entity (such as a financial regulator) may issue a
request 420 via a macroservice 416-2, and the microservice 416-2 may perform various operations to comply with the request (e.g., calling on another macroservice 416-1 to generate a report responsive to therequest 420, correctingdata 402 in thedata storage 404, etc.). In some embodiments, macroservices 416-i may be combined to form other macroservices 416-i. - The microservices 412-i and/or the macroservices 416-i may be exposed to a third party (e.g., by use of an application programming interface, or “API”). In some cases, a single entity may provide the microservices 412-i and the macroservices 416-i. In other cases, one entity may provide the
library 414 of microservices 412-i, and another entity may use themicroservices 414 to generate their own customized macroservices 416-i. - The
machine learning model 410 may be generated and/or refined via a machine learning process, such as the one depicted in the flow chart ofFIG. 4 . Machine learning is a branch of artificial intelligence that relates to mathematical models that can learn from, categorize, and make predictions about data. Such mathematical models, which can be referred to as machine-learning models, can classify input data among two or more classes; cluster input data among two or more groups; predict a result based on input data; identify patterns or trends in input data; identify a distribution of input data in a space; or any combination of these. Examples of machine-learning models can include (i) neural networks; (ii) decision trees, such as classification trees and regression trees; (iii) classifiers, such as Naïve bias classifiers, logistic regression classifiers, ridge regression classifiers, random forest classifiers, least absolute shrinkage and selector (LASSO) classifiers, and support vector machines; (iv) clusterers, such as k-means clusterers, mean-shift clusterers, and spectral clusterers; (v) factorizers, such as factorization machines, principal component analyzers and kernel principal component analyzers; and (vi) ensembles or other combinations of machine-learning models. In some examples, neural networks can include deep neural networks, feed-forward neural networks, recurrent neural networks, convolutional neural networks, radial basis function (RBF) neural networks, echo state neural networks, long short-term memory neural networks, bi-directional recurrent neural networks, gated neural networks, hierarchical recurrent neural networks, stochastic neural networks, modular neural networks, spiking neural networks, dynamic neural networks, cascading neural networks, neuro-fuzzy neural networks, or any combination of these. - Different machine-learning
models 410 may be used interchangeably to perform a task. Examples of tasks that can be performed at least partially using machine-learning models include various types of scoring; workload placement; transaction analysis; bioinformatics; cheminformatics; software engineering; fraud detection; customer segmentation; generating online recommendations; adaptive websites; determining customer lifetime value; search engines; placing advertisements in real time or near real time; classifying DNA sequences; affective computing; performing natural language processing and understanding; object recognition and computer vision; robotic locomotion; playing games; optimization and metaheuristics; detecting network intrusions; medical diagnosis and monitoring; or predicting when an asset, such as a machine, will need maintenance. - Machine-learning models can be constructed through an at least partially automated (e.g., with little or no human involvement) process called training. During training, input data can be iteratively supplied to a machine-learning model to enable the machine-learning model to identify patterns related to the input data or to identify relationships between the input data and output data. With training, the machine-learning model can be transformed from an untrained state to a trained state. Input data can be split into one or more training sets and one or more validation sets, and the training process may be repeated multiple times. The splitting may follow a k-fold cross-validation rule, a leave-one-out-rule, a leave-p-out rule, or a holdout rule. An overview of training and using a machine-learning model is described below with respect to the flow chart of
FIG. 5 . - In
block 502, training data is received. In some examples, the training data is received from a remote database or a local database, constructed from various subsets of data, or input by a user. The training data can be used in its raw form for training a machine-learning model or pre-processed into another form, which can then be used for training the machine-learning model. For example, the raw form of the training data can be smoothed, truncated, aggregated, clustered, or otherwise manipulated into another form, which can then be used for training the machine-learning model. In one example, the training data comprises a transaction log maintained by thecontroller 309 that describes each of a plurality of transactions scheduled for processing by thecontroller 309. - In
block 504, a machine-learning model is trained using the training data. The machine-learning model can be trained in a supervised, unsupervised, or semi-supervised manner. In supervised training, each input in the training data is correlated to a desired output. This desired output may be a scalar, a vector, or a different type of data structure such as text or an image. This may enable the machine-learning model to learn a mapping between the inputs and desired outputs. In unsupervised training, the training data includes inputs, but not desired outputs, so that the machine-learning model has to find structure in the inputs on its own. In semi-supervised training, only some of the inputs in the training data are correlated to desired outputs. - In
block 506, the machine-learning model is evaluated. For example, an evaluation dataset can be obtained, for example, via user input or from a database. The evaluation dataset can include inputs correlated to desired outputs. The inputs can be provided to the machine-learning model and the outputs from the machine-learning model can be compared to the desired outputs. If the outputs from the machine-learning model closely correspond with the desired outputs, the machine-learning model may have a high degree of accuracy. For example, if 90% or more of the outputs from the machine-learning model are the same as the desired outputs in the evaluation dataset, the machine-learning model may have a high degree of accuracy. Otherwise, the machine-learning model may have a low degree of accuracy. The 90% number is an example only. A realistic and desirable accuracy percentage is dependent on the problem and the data. - In some examples, if the machine-learning model has an inadequate degree of accuracy for a particular task, the process can return to block 504, where the machine-learning model can be further trained using additional training data or otherwise modified to improve accuracy. If the machine-learning model has an adequate degree of accuracy for the particular task, the process can continue to block 508.
- In
block 508, new data is received. In some examples, the new data is received from a remote database or a local database, constructed from various subsets of data, or input by a user. The new data may be unknown to the machine-learning model. For example, the machine-learning model may not have previously processed or analyzed the new data. The new data may comprise a new transaction for scheduling by thecontroller 309. - In
block 510, the trained machine-learning model is used to analyze the new data and provide a result. For example, the new data, such as the workload (and/or the immutable log describing the transactions of the workload) can be provided as input to the trained machine-learning model. The trained machine-learning model can analyze the new data and provide a result that includes a classification of the new data into a particular class, a clustering of the new data into a particular group, a prediction based on the new data, or any combination of these. For example, the trained model may output one ormore CRs 302 to process one or more portions of the workload. Thecontroller 309 may then schedule the one or more portions of the workload on theCRs 302 outputted by the model. - In
block 512, the result is post-processed. For example, the result can be added to, multiplied with, or otherwise combined with other data as part of a workload. As another example, the result can be transformed from a first format, such as a time series format, into another format, such as a count series format. Any number and combination of operations can be performed on the result during post-processing. - A more specific example of a machine-learning model is the
neural network 600 shown inFIG. 6 . Theneural network 600 is represented as multiple layers of interconnected neurons, such asneuron 608, that can exchange data between one another. The layers include aninput layer 602 for receiving input data, ahidden layer 604, and anoutput layer 606 for providing a result. Thehidden layer 604 is referred to as hidden because it may not be directly observable or have its input directly accessible during the normal functioning of theneural network 600. Although theneural network 600 is shown as having a specific number of layers and neurons for exemplary purposes, theneural network 600 can have any number and combination of layers, and each layer can have any number and combination of neurons. - The neurons and connections between the neurons can have numeric weights, which can be tuned during training. For example, training data can be provided to the
input layer 602 of theneural network 600, and theneural network 600 can use the training data to tune one or more numeric weights of theneural network 600. - In some examples, the
neural network 600 can be trained using backpropagation. Backpropagation can include determining a gradient of a particular numeric weight based on a difference between an actual output of theneural network 600 and a desired output of theneural network 600. Based on the gradient, one or more numeric weights of theneural network 600 can be updated to reduce the difference, thereby increasing the accuracy of theneural network 600. This process can be repeated multiple times to train theneural network 600. For example, this process can be repeated hundreds or thousands of times to train theneural network 600. - In some examples, the
neural network 600 is a feed-forward neural network. In a feed-forward neural network, every neuron only propagates an output value to a subsequent layer of theneural network 600. For example, data may only move one direction (forward) from one neuron to the next neuron in a feed-forward neural network. - In other examples, the
neural network 600 is a recurrent neural network. A recurrent neural network can include one or more feedback loops, allowing data to propagate in both forward and backward through theneural network 600. This can allow for information to persist within the recurrent neural network. For example, a recurrent neural network can determine an output based at least partially on information that the recurrent neural network has seen before, giving the recurrent neural network the ability to use previous input to inform the output. - In some examples, the
neural network 600 operates by receiving a vector of numbers from one layer; transforming the vector of numbers into a new vector of numbers using a matrix of numeric weights, a nonlinearity, or both; and providing the new vector of numbers to a subsequent layer of theneural network 600. Each subsequent layer of theneural network 600 can repeat this process until theneural network 600 outputs a final result at theoutput layer 606. For example, theneural network 600 can receive a vector of numbers as an input at theinput layer 602. Theneural network 600 can multiply the vector of numbers by a matrix of numeric weights to determine a weighted vector. The matrix of numeric weights can be tuned during the training of theneural network 600. Theneural network 600 can transform the weighted vector using a nonlinearity, such as a sigmoid tangent or the hyperbolic tangent. In some examples, the nonlinearity can include a rectified linear unit, which can be expressed using the following equation: -
y=max(x, 0 )Equation 1 - In
Equation 1, y is the output and x is an input value from the weighted vector. The transformed output can be supplied to a subsequent layer, such as thehidden layer 604, of theneural network 600. The subsequent layer of theneural network 600 can receive the transformed output, multiply the transformed output by a matrix of numeric weights and a nonlinearity, and provide the result to yet another layer of theneural network 600. This process continues until theneural network 600 outputs a final result at theoutput layer 606. - Other examples of the present disclosure may include any number and combination of machine-learning models having any number and combination of characteristics. The machine-learning model(s) can be trained in a supervised, semi-supervised, or unsupervised manner, or any combination of these. The machine-learning model(s) can be implemented using a single computing device or multiple computing devices, such as the communications system discussed herein.
- Implementing some examples of the present disclosure at least in part by using machine-learning models can reduce the total number of processing iterations, time, memory, electrical power, or any combination of these consumed by a computing device when analyzing data. For example, a neural network may more readily identify patterns in data than other approaches. This may enable the neural network to analyze the data using fewer processing cycles and less memory than other approaches, while obtaining a similar or greater level of accuracy.
- The methods, systems, and functionality described herein may be embodied as instructions on a computer readable medium or as part of a computing architecture.
FIG. 7 illustrates an embodiment of anexemplary computing architecture 700 suitable for implementing various embodiments described herein. In one embodiment, thecomputing architecture 700 may comprise or be implemented as part of an electronic device, such as acomputer 701. The embodiments are not limited in this context. - As used in this application, the terms “system” and “component” are intended to refer to a computer-related entity, either hardware, a combination of hardware and software, software, or software in execution, examples of which are provided by the
exemplary computing architecture 700. For example, a component can be, but is not limited to being, a process executing on a processor, a processor, a hard disk drive, multiple storage drives (of optical and/or magnetic storage medium), an object, an executable, a thread of execution, a program, and/or a computer. By way of illustration, both an application running on a server and the server can be a component. One or more components can reside within a process and/or thread of execution, and a component can be localized on one computer and/or distributed between two or more computers. Further, components may be communicatively coupled to each other by various types of communications media to coordinate operations. The coordination may involve the uni-directional or bi-directional exchange of information. For instance, the components may communicate information in the form of signals communicated over the communications media. The information can be implemented as signals allocated to various signal lines. In such allocations, each message is a signal. Further embodiments, however, may alternatively employ data messages. Such data messages may be sent across various connections. Exemplary connections include parallel interfaces, serial interfaces, and bus interfaces. - The
computing architecture 700 includes various common computing elements, such as one or more processors, multi-core processors, co-processors, memory units, chipsets, controllers, peripherals, interfaces, oscillators, timing devices, video cards, audio cards, multimedia input/output (I/O) components, power supplies, and so forth. The embodiments, however, are not limited to implementation by thecomputing architecture 700. - As shown in
FIG. 7 , thecomputing architecture 700 comprises aprocessing unit 702, asystem memory 704 and achipset 706. Theprocessing unit 702 can be any of various commercially available processors, including without limitation an AMD® Athlon®, Duron® and Opteron® processors; ARM® application, embedded and secure processors; IBM® and Motorola® DragonBall® and PowerPC® processors; IBM and Sony® Cell processors; Intel® Celeron®, Core (2) Duo®, Core i9™, Core m3™, vPro™, Itanium®, Pentium®, Xeon®, and XScale® processors; and similar processors. Dual microprocessors, multi-core processors, and other multi-processor architectures may also be employed as theprocessing unit 702. - In some embodiments, the
processing unit 702 couples with thechipset 706 via a highspeedserial link 703 and couples with thesystem memory 704 via a highspeedserial link 705. In other embodiments, theprocessing unit 702 may couple with thechipset 706 and possibly other processor units via a system bus and may couple with thesystem memory 704 via thechipset 706. In further embodiments, theprocessing unit 702 and the chipset may reside in a System-On-Chip (SoC) package. - The
chipset 706 provides an interface for system components including, but not limited to, thesystem memory 704 to theprocessing unit 702. Thechipset 706 may couple with any of several types of bus structure that may further interconnect to a memory bus (with or without a memory controller), a peripheral bus, and a local bus using any of a variety of commercially available bus architectures.Interface adapters chipset 706 via a slot architecture. Example slot architectures may include without limitation Accelerated Graphics Port (AGP), Card Bus, (Extended) Industry Standard Architecture ((E)ISA), Micro Channel Architecture (MCA), NuBus, Peripheral Component Interconnect (Extended) (PCI(X)), PCI Express, Personal Computer Memory Card International Association (PCMCIA), and the like. - The
computing architecture 700 may comprise or implement various articles of manufacture. An article of manufacture may comprise a computer-readable storage medium to store logic. Examples of a computer-readable storage medium may include any tangible media capable of storing electronic data, including volatile memory or non-volatile memory, removable or non-removable memory, erasable or non-erasable memory, writeable or re-writeable memory, and so forth. Examples of logic may include executable computer program instructions implemented using any suitable type of code, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, object-oriented code, visual code, and the like. Embodiments may also be at least partly implemented as instructions contained in or on a non-transitory computer-readable medium, which may be read and executed by one or more processors to enable performance of the operations described herein. - The
system memory 704 may include various types of computer-readable storage media in the form of one or more higher speed memory units, such as read-only memory (ROM), random-access memory (RAM), dynamic RAM (DRAM), Double-Data-Rate DRAM (DDRAM), synchronous DRAM (SDRAM), static RAM (SRAM), programmable ROM (PROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, polymer memory such as ferroelectric polymer memory, ovonic memory, phase change or ferroelectric memory, silicon-oxide-nitride-oxide-silicon (SONOS) memory, magnetic or optical cards, an array of devices such as Redundant Array of Independent Disks (RAID) drives, solid state memory devices (e.g., USB memory, solid state drives (SSD) and any other type of storage media suitable for storing information. In the illustrated embodiment shown inFIG. 7 , thesystem memory 704 can includenon-volatile memory 708 and/orvolatile memory 710. A basic input/output system (BIOS) can be stored in thenon-volatile memory 708. Thecontroller 309 and/or thelog 310 may be stored in thememory - The
computing architecture 700 may include various types of computer-readable storage media in the form of one or more lower speed memory units, including an internal (or external) hard disk drive (HDD) 712, a magnetic floppy disk drive (FDD) 714 to read from or write to a removablemagnetic disk 716, and anoptical disk drive 718 to read from or write to a removable optical disk 720 (e.g., a CD-ROM or DVD). TheHDD 712,FDD 714 andoptical disk drive 720 can be connected to thesystem bus 706 by anHDD interface 722, anFDD interface 724 and anoptical drive interface 726, respectively. TheHDD interface 722 for external drive implementations can include at least one or both of Universal Serial Bus (USB) and IEEE 694 interface technologies. - The drives and associated computer-readable media provide volatile and/or nonvolatile storage of data, data structures, computer-executable instructions, and so forth. For example, a number of program modules can be stored in the drives and
memory units operating system 728, one ormore application programs 730,other program modules 732, andprogram data 734. In one embodiment, the one ormore application programs 730,other program modules 732, andprogram data 734 can include, for example, the various applications and/or components described herein, such as thecontroller 309 and theimmutable log 310. - A user may enter commands and information into the
computer 701 through one or more wire/wireless input devices, for example, akeyboard 736 and a pointing device, such as amouse 738. Other input devices may include microphones, infra-red (IR) remote controls, radio-frequency (RF) remote controls, game pads, stylus pens, card readers, dongles, finger print readers, gloves, graphics tablets, joysticks, keyboards, retina readers, touch screens (e.g., capacitive, resistive, etc.), trackballs, trackpads, sensors, styluses, and the like. These and other input devices are often connected to theprocessing unit 702 through aninput device interface 740 that is coupled to thechipset 706, but can be connected by other interfaces such as a parallel port, IEEE 694 serial port, a game port, a USB port, an IR interface, and so forth. - A
monitor 742 or other type of display device is also connected to thechipset 706 via an interface, such as avideo adaptor 728. Themonitor 742 may be internal or external to thecomputer 701. In addition to themonitor 742, a computer typically includes other peripheral output devices, such as speakers, printers, and so forth. - The
computer 701 may operate in a networked environment using logical connections via wire and/or wireless communications to one or more remote computers, such as aremote computer 744. Theremote computer 744 can be a workstation, a server computer, a router, a personal computer, portable computer, microprocessor-based entertainment appliance, a peer device or other common network node, and typically includes many of or all the elements described relative to thecomputer 701, although, for purposes of brevity, only a memory/storage device 746 is illustrated. The logical connections depicted include wire/wireless connectivity to a local area network (LAN) 748 and/or larger networks, for example, a wide area network (WAN) 750. Such LAN and WAN networking environments are commonplace in offices and companies, and facilitate enterprise-wide computer networks, such as intranets, all which may connect to a global communications network, for example, the Internet. - When used in a LAN networking environment, the
computer 701 is connected to theLAN 748 through a wire and/or wireless communication network interface oradaptor 752. Theadaptor 752 can facilitate wire and/or wireless communications to theLAN 748, which may also include a wireless access point disposed thereon for communicating with the wireless functionality of theadaptor 752. - When used in a WAN networking environment, the
computer 701 can include amodem 754, or is connected to a communications server on theWAN 750, or has other means for establishing communications over theWAN 750, such as by way of the Internet. Themodem 754, which can be internal or external and a wire and/or wireless device, connects to thechipset 706 via theinput device interface 740. In a networked environment, program modules depicted relative to thecomputer 701, or portions thereof, can be stored in the remote memory/storage device 746. It will be appreciated that the network connections shown are exemplary and other means of establishing a communications link between the computers can be used. - The
computer 701 is operable to communicate with wire and wireless devices or entities using theIEEE 802 family of standards, such as wireless devices operatively disposed in wireless communication (e.g., IEEE 802.13 over-the-air modulation techniques). This includes at least Wi-Fi (or Wireless Fidelity), WiMax, and Bluetooth™ wireless technologies, among others. Thus, the communication can be a predefined structure as with a conventional network or simply an ad hoc communication between at least two devices. Wi-Fi networks use radio technologies called IEEE 802.13x (a, b, g, n, etc.) to provide secure, reliable, fast wireless connectivity. A Wi-Fi network can be used to connect computers to each other, to the Internet, and to wire networks (which use IEEE 802.3-related media and functions). - Some aspects may utilize the Internet of Things (IoT), where things (e.g., machines, devices, phones, sensors) can be connected to networks and the data from these things can be collected and processed within the things and/or external to the things. For example, with the IoT, sensors may be deployed in many different devices, and high-value analytics can be applied to identify hidden relationships and drive increased efficiencies. This can apply to both Big Data analytics and realtime (streaming) analytics.
- Some systems may use Hadoop®, an open-source framework for storing and analyzing big data in a distributed computing environment. Apache™ Hadoop® is an open-source software framework for distributed computing. For example, some grid systems may be implemented as a multi-node Hadoop® cluster, as understood by a person of skill in the art. Some systems may use cloud computing, which can enable ubiquitous, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.
-
FIG. 8 illustrates an example of asystem 800 that may generally be representative of a distributed cloud-based computing system or another type of computing network in that one or more techniques described herein may be implemented according to various embodiments. As shown inFIG. 8 ,system 800 may generally be a rack-based system including of a number of racks 804-n, where n may be any positive integer. Each of theracks 804 may be configured to house computingresources 802 to process data and information. Moreover, theracks 804 may be coupled with each via afabric 803, which may be similar to or the same asfabric 303 as similarly discussed above. Theracks 804 may be located within the same data center and other data centers coupled via thefabric 803 and cloud-based networking infrastructure. Thefabric 803 may include a combination of electrical and/or optical signaling media, and high bandwidth interconnects, such as Gigabit Ethernet, 10 Gigabit Ethernet, 100 Gigabit Ethernet, InfiniBand, Peripheral Component Interconnect (PCI) Express (PCIe), PCIe 1.0, PCIe 2.0, PCIe 3.0, PCIe 4.0, PCIe 5.0, and so forth. In one example, thefabric 803 may include networking hardware to support communication of data and information in accordance with the PCIe 4.0 and provides 16 gigatransfers per second (GT/s) bit rate. As will be discussed in more detail below, these high data rates enables computing resources between theracks 803 and across the cloud to be pooled together to provide processing and memory capabilities. Although not pictured, eachrack 804 may inclue instances of thecontroller 309 and/or thelog 310. More generally, thesystem 800 may include instances of thecontroller 309 and/or thelog 310 to schedule workloads for processing on the racks 804-1 and/or any component thereof. - In embodiments, a
rack 804 includescomputing resources 802, which may include processingresources 822 andmemory resources 820. Theprocessing resources 822 include one ormore processors 850 having processing circuitry to process information and data. The one ormore processors 850 may be a single core processor or multi-core processor. In some embodiments, the one ormore processors 850 may each be a multi-chip package (MCP), a system on chip (SoC) package, and so forth including other circuitry and components, such as memory 840, amemory controller 842, and one ormore interfaces 844. Embodiments are not limited in this manner. In one embodiment, thecomputing resources 802 are representative of thecomputing resources 302 of thesystem 300, and thecontroller 309 may schedule workloads for processing in thesystme 800 as described above. - In embodiments, the memory 840 is a local memory, e.g., coupled and/or on the same die or package as the one or
more processors 850. In one example, the memory 840 is cache memory and stores information and data for processor cores of theprocessors 850. More specifically, the memory 840 may store copies of data that is frequently used by a processor core stored in “main” memory, such asmemory 830 of thesame rack 804 ormemory 830 of adifferent rack 804. In embodiments, the memory 840 local to a processor core, e.g., on the same die or package, may be relatively small in size compared to “main” memory. For example,memory 830 may be 4, 8, 16, kilobytes (KB) or megabytes (MB) in size, while “main” memory may be on the order of gigabytes (GB) in size, e.g., 2, 4, 8, 16 GBs. - In embodiments, the
memory 830 and memory 840 may be configured as a hierarchy of one or more cache levels (L1, L2, L3, etc.). Memory 840 may be on the same die or package as theprocessor 850 and may be part of higher level cache (L1, or L2).Memory 830, which may be within thesame rack 804 but not on the die with the processing cores, may be lower level cache, e.g., L3 cache. Whilememory 830 of adifferent rack 804 then theprocessor 850 utilizing it, may be used as even lower level cache, e.g., L4 or L5 cache. In these embodiments, the higher level cache may store information and data that is accessed more frequently than information and data store in a low-level cache. Moreover, the information and data may be moved among the different levels ofmemory 840 and 830 based on a change in use and/or access. For example, as information and data are used more frequently, it may be moved/copied from a lower level cache, e.g., L3, L4, or L5, to a higher level cache, e.g., L1 or L2 Similarly, as information and data stored in lower level cache is used less frequently, it may be copied or moved to a higher level cache. - In one example, the first rack 804-1 may include processing resources 822-1 having one or more processors 850-1 and local memory 840-1, which may be part of the same die or package as processors 850-1, and may be an L1 or L2 cache for the processor 850-1. Further, memory 830-1 of the same rack 804-1 may be considered or configured a lower level cache for the processor 850-1, e.g., L3. The processor 850-1 may also use memory 830-n, where n may be any positive integer other than 1 in this example, as even lower level cache, L4 or L5 cache.
- In embodiments, the
processing resources 822 may include amemory controller 842 or a digital circuit to manage the flow of data between memory 840,memory 830, andprocessors 850. Thememory controller 842 may be part of the same die or package as theprocessors 850 or integrated on another chip. Thememory controller 842 may control read and writes to memory 840, which may be local or on the same die as thememory controller 842. In some embodiments, thememory controller 842 may be coupled with thememory resources 820 via one ormore interfaces memory controller 842 may operate in conjunction with memory controller 832 of thememory resources 820. Thememory controllers 832 and 842 may operate in conjunction with each other to perform read/write operations to store data and information inmemory 830, for example. In embodiments, the information, data, and memory requests may be communicated between thememory controllers 832 and 842 via theinterfaces interfaces fabric 803 coupling theprocessing resources 822 and thememory resources 820 within thesame rack 804. Theinterfaces resources 802, e.g., utilizingPCIe 4. However, Embodiments are not limited in this manner. - In some embodiments, the
computing resources 802, including theprocessing resources 822 and thememory resources 820, of onerack 804 may be coupled with and utilizecomputing resources 802 of anotherrack 804 via one ormore switches 850, which may be part of thefabric 803. In embodiments, theswitches 850 may be electrical and/or optically coupled with each other and enable communication via packet switching. In embodiments, theswitches 802 include circuitry to extend the PCIe fabric from within arack 804 to one or moreother racks 804 and computingresources 802 therein. More specifically, aswitch 850 may include circuitry and logic such thatcomputing resources 802 of onerack 804 can share input/output (I/O) and memory functionality withcomputing resources 802 of anotherrack 804 utilizing single root I/O virtualization (SR-IOV) or multifunction virtualization. Embodiments are not limited in this manner. - Like
system 300,system 800 may enable the combination ofcomputing resources 802 to process a workload, job, and/or task. For example, thecontroller 309 may receive a request to process one or more workloads, such as those related to performing financial calculations, determine thecomputing resources 802 required to perform the one or more workloads as described above, and combine thecomputing resources 802 to perform the one or more workloads, e.g. generate composedcomputing resources 802. In embodiments, thecontroller 309 may determine the number ofcomputing resources 802 to combine based on requirements, such as a service level agreement (SLA), or another prioritizing scheme. As mentioned, the computing resources may be combined within arack 804, acrossracks 804 within the same data center, and acrossracks 804 in different data centers via thefabric 803 and cloud-based infrastructure. -
FIG. 9 illustrates another example of asystem 900, which may be similar to or the same assystem 800, and/or any other system discussed herein.FIG. 9 illustrates a number of computing resources, includingmemory resources 920 andprocessing resources 922, coupled with each via a plurality of switches 950-g, where g may be any positive integer. In embodiments, the computing resources may be coupled via fabric and the cloud-base infrastructure. In the illustrated example, one or more of the computing resources may be located within the same data center, while one or more other computing resources may be located in a different data center. - As discussed, the
processing resources 922 may include one or more processors, processing packages, and processing cores, and thememory resources 920 may include one or more memory chips or banks of memory to store information and data for thesystem 900. Thesystem 900 may include any number of processing resources 922-m and memory resources 920-p, where m and p may be any positive integers (same or different). These computing resources may be pooled and grouped together to perform one or more workloads. Thesystem 900 includes thecontroller 309, which may be one or more servers and devices capable of coordinate various aspects of thesystem 900 including receiving workloads, determining computing resources for the workloads, causing the resources to perform work on the workloads, grouping computing resources together to perform the work on the workloads, notifying a user and/or user system that the workloads completed, and other coordinating tasks. For example, thecontroller 309 may coordinate and cause processing of multiple workloads at time, which may include determine priority levels for the workloads and so forth. In embodiments, thecontroller 309 may operating in accordance with one or more SLAs and or user configurations. For example, an SLA and/or user setting may specify which workloads and/or which types of workloads have higher priorities than other types of workloads. Embodiments are not limited in this manner. - In the illustrated example of
FIG. 9 , thecontroller 309 has generated groupedresources 931 that includes processors from processing resources 922-1 and memory from memory resources 920-1. In embodiments, the groupedresources 931 may represent a composed node, and the computing resources may be utilized to perform one or more workloads. Note that although the box around the groupedresources 931 includes a portion of switch 950-1, the traffic including memory read/write requests, data, and information, communicated between the processing resources 922-1 and memory resources 920-1 may pass through switch 950-1 and/or any other of the switches 950-g based on the networking configuration. Further, in embodiments, the processing resources 922-1 and memory resources 920-1 may be incorporated in the same rack, and the traffic may pass through switch 950-1. In other instances, the processing resources 922-1 and memory resources 920-1 may be incorporated into different racks, and the traffic may pass through switch 950-1 and/or any other switches 950-g. As similarly discussed above, theswitches 950, the fabric, and the cloud-basedinfrastructure 907 support high band width communications, e.g.,PCIe 4 operating at approximately 16 GT/s or 8 GB/s total for 4 lanes). -
FIG. 10 illustrates another example of asystem 1000, which may be similar to or the same assystem 300,system 800,system 900, and/or any other system discussed herein.FIG. 10 illustrates a number of computing resources, includingmemory resources 1020 andprocessing resources 1022, coupled with each via a plurality of switches 1050-g, where g may be any positive integer. In embodiments, the computing resources may be coupled via fabric and the cloud-base infrastructure 1007. In the illustrated example, one or more of the computing resources may be located within the same data center, while one or more other computing resources may be located in a different data center. - In the illustrated example,
system 1000 may be the same assystem 1000, however, may be in a different configuration. In this example, thecontroller 309 has generated groupedresources 1031 that includes processors from processing resources 1022-1, a memory from memory resources 1020-1, and memory from memory resources 1020-2. In embodiments, the groupedresources 1031 may represent a composed node, and the computing resources may be utilized to perform one or more workloads, such as processing transactions defined by one or more records (or entries) in animmutable log 310. As similarly discussed, traffic communicated between the computing resources of the groupedresources 1031 may be communicated through one or more switches, including switch 1050-1 and switch 1050-2. However, embodiments are not limited in this manner, and the traffic may be communicated to other switches 1050-g (and networking equipment). In this example, thecontroller 309 may receive a request to perform one or more workloads to process data and information and based on a priority level, generated the groupedresources 1031. Moreover, embodiments are not limited to these examples. One or more computing resources may grouped to generate a composed node from any one of a plurality of racks that may be part of one or more data centers and coupled via the cloud-basedinfrastructure 1007, for example. -
FIG. 11 illustrates an example of aprocessing flow 1100 that may be representative of some or all the operations executed by one or more embodiments described herein. For example, theprocessing flow 1100 may illustrate operations performed by thecontroller 309 to schedule one or more workloads for processing on a cloud-based distributed system. However, embodiments are not limited in this manner, and one or more other components may perform operations to enable and support the operations discussed in thisprocessing flow 1100. - At
block 1102, theprocessing flow 1100 includes receiving a request to process or one or more workloads by a cloud-based computing system. The request may include information and data used to perform the workload, such as financial information, and may be received from one or more other systems. In embodiments, the request may indicate processing that needs to be done on the information and data to generate a result. For example, the workload may be defined by an immutable log specifying details of a transaction to be processed (e.g., the purchase and/or sale of a stock). - At
block 1104, theprocessing flow 1100 includes thecontroller 309 determining one or more computing resources to be utilized to process the workload. The one or more resources may include processing resources, such as processors and/or processing cores, and memory resources, such as memory. In embodiments, the controller may determine which resources to process based on one or more criteria including, a priority level for the workload, computing resources available, location of computing resources, processing/memory capabilities of the computing resources, processing requirements for the workload, an SLA associated with the requester of the workload, and so forth. For example and in some embodiments, the controller may determine one or more computing resources to process the workload based on the computing resources being within the same data center. In other instances, the controller may determine group computing resources that are located within different data centers. Embodiments are not limited to this example. As stated, thecontroller 309 may consider the utilization levels of the computing resources (e.g., compute nodes), the utilization levels of the fabric links to each computing resource, the number of links from the computing resource to data needed to process the transaction, and/or thelog 310 when selecting one or more computing resources to process the workload. - At
block 1106, thecontroller 309 may allocate and group the one or more computing resources to process the workload. For example, thecontroller 309 may identify and provide information to the computing resources (and controlling software) to allocate the resources for the workload. In one example, the computing resources may be controlled by an operating system and the entire computing resource may be allocated to process the workload. In another example, the computing resources may be part of a virtual environment and may be controlled via a virtual machine monitor, such as a hypervisor, that operates virtual machines in the virtual environment. In another example, the computing resources may be controlled by a docker engine, and the workload may be processed in a container operating. In both the virtual machines and docker system, the computing resources may be shared among a plurality of workloads. Embodiments are not limited in this manner. Thecontroller 309 may generally allocate and group resources based on the utilization levels of the grouped computing resources, the utilization levels of the fabric links to each group of computing resources, the number of links from the group of computing resources to data needed to process the transaction, and/or thelog 310. - At
block 1108, theprocessing flow 1100 includes processing and/or causing the workload to be processed by the computing resources. For example, the transaction defined by the immutable log may be processed. The computing resources may communicate between each other via a fabric and/or cloud-based networking infrastructure which includes one or more high speed interconnects, such PCIe-4. As previously discussed, these high-speed interconnects enable the computing resources to be located within the same data center and among other data centers while maintaining high-speed connectivity between themselves. For example, a processing resource may be coupled with a memory resource in a different data center and still be able to utilize the memory resource as “main” memory, cache memory, and/or in a memory hierarchy as previously discussed. - In embodiments, the
processing 1100 includes determining that the workload is complete, e.g., done being processed atblock 1110. For example, the computing resources and/or controlling software may send a notification to the controller indicating that the workload has completed. Atblock 1112, thecontroller 309 may release the computing resource, e.g., make them available to process other workloads. Further and atblock 1114, the controller may notify the requesting system that the workload is complete including providing any results for the workload. -
FIGS. 12A-12B depict embodiments of adatabase system 1200 andFIGS. 12C-12D depicts embodiments of data structures in thedatabase system 1200. In particular,FIG. 12A illustrates a layer diagram of adatabase system 1200 that interacts with computing node(s) 1210. The computing node(s) 1210 may be local or remotely located computers, servers, workstations, or the like such as thecomputer 700 illustrated inFIG. 7 . In many embodiments, the computing node(s) 1210 obtain and forward raw event data to thedatabase system 1200 or consume data, derived data, queried data, summarized data, reports, and/or the like from thedatabase system 1200. - The
database system 1200 may comprise a combination of hardware and code to receive and store raw event data as log object(s) 1260 in a persistent log and to derive or compute derived data based on the raw event data to store in database object(s) 1250 in a persistent database. In many embodiments, thedatabase system 1200 captures a representation of the financial environment within which financial calculations and decisions are made and permanently stores that representation, at least for a period of time, to facilitate derivation and/or summarization of financial data. The summarization and/or derivation adds business intelligence to the data to form financial information to provide to authorized users or consumers in the form of reports, tables, lists, SQL databases, graph databases, relational databases, and/or any other data structure. Thedatabase system 1200 may maintain persistent records of derivations and summarizations in the form of database objects 1250. The derivations include, for instance, computations, corrections, and/or cancelations of the raw event data in the log object(s) 1260. The derived data may include, for instance, trades, settlements, and holdings such as stock holdings, stock trades, stock buys, stock sells, mutual fund holdings, mutual fund trades, mutual fund buys, mutual fund sells, commodity holdings, commodity trades, commodity buys, commodity sells, net asset values, and/or the like. Thedatabase system 1200 may include immutable log objects, such as the log objects 1260, describing transactions scheduled for processing by thecontroller 309. - In many embodiments, the
database system 1200 may maintain not only persistent records of the computations, summarizations, corrections, and/or cancelations but also persistent records of the logic or code to perform the computations, summarizations, corrections, and/or cancelations such that thedatabase system 1200 or another system can perform such derivations on the raw event data included in the log object(s)s 1260 to recreate the data derived from such computations, summarizations, corrections, and/or cancelations. The logic or code may comprise, for instance, the logic to perform the derivations, code in the host environment (e.g. software environment) that affect the logic to perform the derivations, code of a virtual machine within which thedatabase system 1200 performs derivations, and/or the like. - The
database system 1200 may comprise service layer(s) 1220, translation layer(s) 1230, and physical layer(s) 1240. The service layer(s) 1220 may comprise producer(s) 1222 and consumer(s) 1224. The producer(s) 1222 may comprise one or more application programming interfaces (APIs) that receive, index, encrypt, and store raw event data from the computing node(s) 1210 as log object(s) 1260 in a persistent log. The consumer(s) 1224 may query data and derived data on behalf of the computing node(s) 1210 and present the results of the query in a format requested by the computing node(s) 1210 or a format optimized for usage by the computing node(s) 1210. In other words, the service layer(s) 1220 may provide an interface to access the data in the database object(s) 1250 and/or the log object(s) 1260 in any data structure. For example, a consumer may request a report via a comma-separated values file format, an online analytical processing (OLAP) format or an online transactional processing (OLTP) format. The producer(s) 1222 and consumer(s) 1224 may each comprise one or more of the microservices and/or macroservices such as the microservices 412-1 through 412-N and the macroservices 416-1 through 416-N illustrated inFIG. 4 to perform the services provided to the computing node(s) 1210. One or more microservices and/or macroservices operating on behalf of the consumer(s) 1224 may generate the data structure and populate fields of the data structure with pointers to the corresponding raw event data and derived data prior to transmitting the data structure to the computing node(s) 1210. - In some embodiments, the service layer(s) 1220 may perform services such as estimating a trade settlement, estimating a net asset value, estimating a stock holding, estimating a stock value, estimating a mutual fund value, estimating a commodity value, and/or the like. In further embodiments, the service layer(s) 1220 may perform services such as estimating a trade settlement, estimating a net asset value, estimating a stock holding, estimating a stock value, estimating a mutual fund value, estimating a commodity value, and/or the like as of a specified time and within a specific time frame.
- In many embodiments, in addition to limiting access to data to authorized users or processes, the
database system 1200 may affinitize data to one or more geographical locations and/or anti-affinitize the data to one or more geographical locations. In other words, some financial data should or must remain within certain geographical areas and should not or must not enter other geographical areas. A service layer(s) 1220 service may provide at least one mechanism to enforcing geographical locations on data in the financial information provide to the computing node(s) 1210. - The translation layer(s) 1230 may include one or more layers to facilitate the generation of representations of the data in multiple formats or any format for which an API can generate by provision of predetermined indices and indices that the
translation layer 1230 generates on-the-fly. For instance, the raw event data in the log object(s) 1260 may include a unique index or pointer that uniquely identifies the location of the log object(s) 1260 in the persistent log. Similarly, the database object(s) may include a unique index or pointer that uniquely identifies the location of the database object(s) 1250 in the persistent database. Furthermore, the database object(s) 1250 may include predetermined sets of indices to support common or frequently-requested queries or derivations. The translation layer(s) 1230 may generate indices during execution of less frequently requested queries or derivations. - In one embodiment, generation of indices to raw event data may comprise generation of indices to a current holding of stock shares, a buy of stock shares, a sell of stock shares, a correction of raw event data such as trade data or settlement data, a cancelation of raw event data such as trade data or settlement data, and/or the like. In another embodiment, generation of indices to raw event data may comprise generation of indices to timestamps to indicate that time at which the transaction settled, a price of a share of the stock at the time of settlement of the transaction, an entity that placed the order, an entity that fulfilled the order, the number of shares of the stock that transferred at the settlement, the funds transferred at settlement, the entity that transferred the funds, the entity that received the funds, and/or the like.
- The physical layer(s) 1240 may support the random-access of data such as raw event data in the log object(s) 1260 and derived data in the data object(s) 1250 to support the translation(s)
layer 1230 and services layer(s) 1220 services. In some embodiments, the physical layer(s) 1240 provide “byte access”. In other words, the physical layer(s) 1240 provide access at a level of granularity of a byte or 8 bits to increase the efficiency of access to the data without necessarily implementing queuing or caching schemes to reduce inefficiencies involved with levels of granularity that are greater than a byte. In many of the embodiments, the physical layer(s) 1240 maintains a list of the log object(s) 1260 and a list of the database object(s) 1250. In some embodiments, the list of database object(s) 1250 reside in a relational database such as a database with an Apache Kafka architecture. -
FIG. 12B illustrates an embodiment of asystem 1201 such as the physical layer(s) 1240 illustrated inFIG. 12A . Thesystem 1201 may provide random access to large stores of data including the database object(s) 1250 and the log object(s) 1260 to the hosts 1211-1 through 1211-N viaswitch 1202. The hosts 1211-1 through 1211-N may perform the services layer(s) 1220 and translation layer(s) 1230 operations. Theswitch 1202 may comprise one or more switches such as an Avago/PLX PEX3090 family switch to provide random-access to the drives 1212-1 through 1212-N for the hosts 1211-1 through 1211-N via a fabric such as PCIe (Peripheral Component Interconnect Express). In one embodiment, theswitch 1202 can support up to fifty hosts 1211-1 through 1211-N. In other embodiments, the fabric may include Ethernet, other conductor-based buses or optical buses such as Fiber Channel, Infiniband, Omni-path, and/or the like. - In many embodiments, the
database system 1200 executing on the hosts 1211-1 through 1211-N, is optimized for a low granularity of random access, such as byte access, to data residing on the drives 1212-1 through 1212-N. For example, at a byte level of granularity, thedatabase system 1200 may, in many instances, access the data and only the data of interest since the byte is a common level of granularity with which many computer systems operate. In contrast, hard disk drive may have a level of granularity of on block, which is 4 kilobytes (KB). As a result, if thedatabase system 1200 stores data on a hard disk drive and requires access to one byte of data, the hard disk drive will read and return 4 KB. If thedatabase system 1200 requires another byte of data, thedatabase system 1200 may have to read another 4 KB of data to obtain the one byte. - The
switch 1202 may provide random access to the drives 1212-1 through 1212-N by virtualizing routing between the hosts 1211-1 through 1211-N and the drives 1212-1 through 1212-N. Bus architectures such as PCIe are designed to interconnect one host with one device. Theswitch 1202 may implement connections such as Tunneled Window Connections (TWCs) that allow multiple hosts 1211-1 through 1211-N to communicate with multiple drives 1212-1 through 1212-N via a PCIe bus. In other words, theswitch 1202 may offer multi-route input/output (I/O) virtualization (MRIOV) to facilitate direct access by any host 1211-1 through 1211-N to the content on any drive 1212-1 through 1212-N. For instance, theswitch 1202 may intercept an incoming packet and emulate another device to hide the host or drive status and make the host or drive look like a target device so the hosts 1211-1 through 1211-N may each directly access any of the drives 1212-1 through 1212-N. - In several embodiments, the drives 1212-1 through 1212-N may comprise 3D XPoint® NVMe (non-volatile memory) solid state drives (SSDs) that offer byte level access. In combination with the MRIOV or similar arrangement, the bank of drives 1212-1 through 1212-N can be as accessible as memory such as DDR dynamic random-access memory and 3D XPoint® NVMe (non-volatile memory). In other embodiments, the drives 1212-1 through 1212-N may comprise other SSDs, flash drives, optical drives, hard drives, or the like.
-
FIG. 12C illustrates an embodiment of alog object 1271 such as the log object(s) 1260 inFIG. 12A for a persistent log such as thelog 310. Thelog object 1271 may persist raw event data. In many embodiments, the raw event data receives anorder 1272 that identifies the order in which the raw data arrives at the log in relation to other raw event data that arrives at the log. Thelog object 1271 encompasses theorder 1272 and the raw event data asdata 1273. In several embodiments, a cryptographic hash such as SHA-1 encrypts the content of thelog object 1271. - The
data 1273 may include a timestamp to identify a time of the occurrence of the event, a hash of the raw event data to uniquely identify the raw event data, a context to describe the event, and possibly other data. For example, an event may involve the purchase of shares of a stock. The raw event data may include a timestamp to indicate that time at which the transaction settled, a price of a share of the stock at the time of settlement of the transaction, an entity that placed the order, an entity that fulfilled the order, the number of shares of the stock that transferred at the settlement, the funds transferred at settlement, the entity that transferred the funds, the entity that received the funds, and/or the like. The hash of the raw event data may operate as a pointer to uniquely identify thelog object 1271 for the purposes of generation of report, databases, and/or the like by the translation layer(s) 1230 and the services layer(s) such as the translation layer(s) 12030 and the services layer(s) 1220 depicted inFIG. 12A . -
FIG. 12D illustrates an embodiment of a database objects 1280 such as the database object(s) 1250 inFIG. 12A for a persistent database. The database objects 1280 may persist derived data such as computed data, summarized data, corrected data, and canceled data. The database objects 1280 may comprise afirst database 1281, or child table, coupled with asecond database 1282, or parent table, via a relation such as a foreign key relation. In the context of relational databases, a foreign key is a field (or collection of fields) in one table that uniquely identifies a row of another table or the same table. In other words, the foreign key is defined in a second table, but it refers to the primary key or a unique key in the first table. For example, a table called Corrections and Cancelations has a primary key called correction_id. Another table called Correction and Cancelation Details has a foreign key which references correction_id to uniquely identify the relationship between both tables. - The table containing the foreign key is called the child table, and the table containing the candidate key is called the referenced or parent table. In database relational modeling and implementation, a unique key is a set of zero or more attributes, the value(s) of which are guaranteed to be unique for each tuple (row) in a relation. Note that the
database system 1200 inFIG. 12A can implement the database objects 1280 in other database structures and are not limited to a relational database. - In some embodiments, the number of instances of the
first database 1281 may differ and may be less than the number of instances of thesecond database 1282. The first database may include an original creator identifier (ID) (Kid) 1283, atimestamp 1284, queries 1285, and possibly other fields. Theoriginal creator ID 1283 may refer to the creator of an original record that resides in a log object such as thelog object 1271 in the persistent log or immutable log. Thetimestamp 1284 may be a pointer to a timestamp in the log object that refers to the time of creation of the raw event data and thequeries 1285 may include one or more columns that include indices or pointers for data in the log object or other database objects 1280. For example, the log object may include a raw data event that describes a current holding of stock shares, a buy of stock shares, a sell of stock shares, a correction of raw event data, a cancelation of raw event data, and/or the like. The database objects 1280 may create a corrections table to identify corrections in the persistent log and perform the corrections in the database objects 1280. Thereafter, the reports or other queries related to the stock holdings can find the corrections data in thedatabase object 1280 and avoid having to perform the incorrect or canceled trades included in the log. - The
second database 1282 may provide details related to raw event data in a log such as a correction associated with the raw event data or derived data based on the raw event data. Thesecond database 1282 may include a deleter ID (Did) 1290, a corrector ID (Cid) 1291, data (1392), and possibly other fields. Thedeleter ID 1290 may include an index for an event that deletes or cancels an event in the raw event data of the log. Thecorrector ID 1291 may include an index for the event that changes or corrects the raw event data of the log or adds derived data related to the raw event data and thedata 1292 may include data that is the corrected data or derived data associated with the log object. Many embodiments may include corrections and cancelations that result from, e.g., the differences in time between a trade and settlement of that trade. -
FIG. 13 illustrates anembodiment 1300 of a corrections and cancelations table in a database object such as the database objects 1280 depicted inFIG. 12D and how the table changes over a period of time fromtime 1 throughtime 6. Thedatabase object 1310 describes the first database 1381 that includes arelation 1305 such ass a foreign key relation to a deriveddatabase object 1320 attime 1 such as thesecond database 1282 illustrated inFIG. 12D . The deriveddatabase 1320 may maintain corrections and cancelations for a specific stock of a specific fund such as a mutual fund. The corrections and cancelations table 1320 may track the cancelations and corrections for this stock and fund because the process of trading stock by this fund may involve corrections and cancelations over periods of time that have an impact on one or more financial aspects of the fund such as tax liabilities, net asset values (NAVs), and/or the like. - At
time 1, the deriveddatabase object 1320 may include a first-row entry 1312 that indicates that the trade is still valid because the deleter id (Did) is set to infinity, the creator ID (Cid) is anevent number 100, the timestamp (Ts) of the event is 1 and the trade (V) is a buy of 10 shares of the stock. At thetime 2, a second-row entry 1322 is added to the deriveddatabase object 1320 to indicate that the trade is still valid because the deleter id (Did) is set to infinity, the creator ID (Cid) is anevent number 110, the timestamp (Ts) of the event is 2 and the trade (V) is a buy of 20 shares of the stock. -
Event 120 performs a correction attime 3. In particular, at thetime 3, the first-row entry 1332 is modified in the deriveddatabase object 1320 to indicate that the trade is not valid as of theevent 120 because the deleter id (Did) is changed from infinity to theevent 120. A third-row entry 1334 is added to the deriveddatabase object 1320 to indicate that the trade is still valid because the deleter id (Did) is set to infinity, the creator ID (Cid) is theevent number 120 that is the same event that canceled the trade in the first-row entry 1332, the timestamp (Ts) of the event is 1, and the trade (V) is a buy of 8 shares of the stock. Note that setting the timestamp (Ts) to 1 indicates that the added trade replaces the original trade, which was canceled, at the time associated with Ts equal to 1. -
Time 4 illustrates another embodiment of a cancelation and correction. Theevent 130 cancels the trade from theevent 120 by amending the deleter ID in thethird row entry 1342 to include theevent number 130 and by adding afourth row entry 1344 to the deriveddatabase object 1320 to indicate that the trade is still valid because the deleter id (Did) is set to infinity, the creator ID (Cid) is anevent number 130, the timestamp (Ts) of the event is 1 and the trade (V) is a buy of 5 shares of the stock. Again, setting the Ts equal to 1 indicates that this trade is the replacement trade for the trade that previously occurred at the time Ts equal to 1. -
Time 5 illustrates an embodiment of to reintroduce a succession of canceled events and to cancel those events while introducing a new event to replace the event at the time of Ts equal to 1 via cancelation of the fourth row and addition of the fifth, sixth, andseventh rows 1352. Theevent 140 cancels the fourth row by inclusion of theevent number 140 in the delete ID field. Theevent 140 also adds the fifth and sixth rows but also deletes these rows by inclusion of theevent number 140 in the delete ID fields. In particular, the fifth row reintroduces and cancels the buy of 10 shares at the time Ts equal to 1 and the sixth row reintroduces and cancels the buy of 8 shares at the time Ts equal to 1. The event adds the trade at the seventh row with the indication that the trade is still valid because the deleter id (Did) is set to infinity, the creator ID (Cid) is anevent number 140, the timestamp (Ts) of the event is 1, and the trade (V) is a buy of 7 shares of the stock. -
Time 6 illustrates an embodiment of an undelete function in the deriveddatabase object 1320 at the fourth, fifth, sixth, andseventh rows 1362. At the fourth row, theevent 150 undeletes the trade byevent 130 by changing the delete ID in the fourth row fromevent 140 to infinity. Theevent 150 also cancels or reconfirms the deletion of the trades in the fifth and sixth rows by changing the delete IDs in the fifth and sixth rows fromevent 140 to 150. Then, at the seventh row in the deriveddatabase object 1320 attime 6, theevent 150 changes the delete ID from infinity toevent 150 to cancel the trade of buying 7 shares of the stock. - In several embodiments, database objects such as the derived
database object 1320 may include ladders to describe a jump from, e.g.,time 1 totime 6, to avoid repetitions of calculations involved with making corrections and calculations over a period of time. If, for instance, a consumer such the consumer(s) 1224 inFIG. 12A does not need to know the detail of the corrections that occurred betweentime 1 andtime 6, the ladder may provide a record that describes indicates a ladder fromtime 1 totime 6 and indicates the resulting buy of shares for that time period. In such embodiments, the ladder may only be valid for calculations or reports that begin on or after thetime 6 because the corrections and calculations could have affected values of other trades or other financial information during the period of time betweentimes -
FIG. 14 illustrates an example of aprocessing flow 1400 that may be representative of some or all the operations executed by one or more embodiments described herein. For example, theprocessing flow 1400 may illustrate operations performed by thecontroller 309 to schedule one or more workloads for processing on a cloud-based distributed system. However, embodiments are not limited in this manner, and one or more other components may perform operations to enable and support the operations discussed in thisprocessing flow 1400. - As shown, at
block 1402, thecontroller 309 may analyze animmutable log 310 for a transaction. As stated, one or more entries of theimmutable log 310 may specify an account identifier, asset identifier (e.g., a stock ticker), a transaction type, and any other parameter for the transaction. For example, theimmutable log 310 may specify to purchase a specified amount of a stock. Atblock 1404, thecontroller 309 may determine one or more data elements required to process the transaction. For example, thecontroller 309 may determine that information describing the account and/or the stock is needed to process the transaction. Atblock 1310, thecontroller 309 may determine the locations of the data elements determined atblock 1404. For example, the account information may be stored on a first compute node, while the current price of the stock is located on a second compute node. - At
block 1408, thecontroller 309 may identify one or more compute nodes proximate to the determined locations of the data. For example, thecontroller 309 may identify the first compute node, the second compute node, and a plurality of other compute nodes within a predefined distance (e.g., a number of network hops) of the first and/or second compute nodes. Atblock 1410, thecontroller 309 may determine the utilization levels of each resource (e.g., CPU, RAM, storage I/O, network I/O, etc.) of the compute nodes identified atblock 1408. Thecontroller 309 may further estimate an amount of time the compute node may require to process the workload (and/or a portion thereof). Atblock 1412, thecontroller 309 may determine the utilization levels of each network link to thefabric 302 for the compute nodes identified atblock 1408. Thecontroller 309 may further determine other attributes of each network link, such as latency, jitter, etc. Atblock 1414, thecontroller 309 may compute a score for each node identified atblock 1408. For example, using thelog 310, thecontroller 309 may compute a score reflecting the suitability of each node identified atblock 1408 to process the workload in accordance with the QoS parameters of the SLA for the requesting client. - At
block 1416, thecontroller 309 selects one or more nodes identified atblock 1408 that satisfy the QoS parameters. For example, thecontroller 309 may select a predefined number of nodes having the highest scores computed atblock 1414. In addition and/or alternatively, thecontroller 309 may select the nodes having the closest proximity to the data needed to process the transaction. In addition and/or alternatively, thecontroller 309 may select the nodes having the lowest resource utilization levels. In addition and/or alternatively, thecontroller 309 may select the nodes having the lowest network link utilization levels. In addition and/or alternatively, thecontroller 309 may select the nodes having the greatest number of network links to the needed data. In addition and/or alternatively, thecontroller 309 may select the least recently used nodes. In addition and/or alternatively, thecontroller 309 may select the compute nodes determined to process the workload (and/or one or more portions thereof) in the least amount of time (and/or within amounts of time specified in the SLA). Once selected, thecontroller 309 may deploy the workload (and/or one or more portions thereof) to each node selected by thecontroller 309 to process the workload. - Some embodiments may be described using the expression “one embodiment” or “an embodiment” along with their derivatives. These terms mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment. Further, some embodiments may be described using the expression “coupled” and “connected” along with their derivatives. These terms are not necessarily intended as synonyms for each other. For example, some embodiments may be described using the terms “connected” and/or “coupled” to indicate that two or more elements are in direct physical or electrical contact with each other. The term “coupled,” however, may also mean that two or more elements are not in direct contact with each other, but yet still co-operate or interact with each other.
- It is emphasized that the Abstract of the Disclosure is provided to allow a reader to quickly ascertain the nature of the technical disclosure. It is submitted with the understanding that it will not be used to interpret or limit the scope or meaning of the claims. In addition, in the preceding Detailed Description, it can be seen that various features are grouped together in a single embodiment for the purpose of streamlining the disclosure. This method of disclosure is not to be interpreted as reflecting an intention that the claimed embodiments require more features than are expressly recited in each claim. Rather, as the following claims reflect, inventive subject matter lies in less than all features of a single disclosed embodiment. Thus the following claims are hereby incorporated into the Detailed Description, with each claim standing on its own as a separate embodiment. In the appended claims, the terms “including” and “in which” are used as the plain-English equivalents of the respective terms “including” and “wherein,” respectively. Moreover, the terms “first,” “second,” “third,” and so forth, are used merely as labels, and are not intended to impose numerical requirements on their objects.
- What has been described above includes examples of the disclosed architecture. It is, of course, not possible to describe every conceivable combination of components and/or methodologies, but one of ordinary skill in the art may recognize that many further combinations and permutations are possible. Accordingly, the novel architecture is intended to embrace all such alterations, modifications and variations that fall within the spirit and scope of the appended claims.
Claims (20)
1. A method, comprising:
receiving a workload related to an immutable log describing a transaction for processing by at least one compute node of a plurality of compute nodes communicably coupled by a fabric in a cloud computing environment;
determining, based on the immutable log describing the transaction, that a first compute node stores at least one data element required to process the transaction;
determining utilization levels of a plurality of computing resources of the first compute node;
determining utilization levels of a plurality of links connecting the first compute node to the fabric;
determining, based on the utilization levels of the computing resources and the links, that processing the workload on the first compute node satisfies one or more Quality of Service (QoS) parameters specified in a service level agreement (SLA); and
scheduling the workload for processing on the first compute node based on the determination that processing the workload on the first compute node satisfies the one or more QoS parameters specified in the SLA.
2. The method of claim 1 , wherein the immutable log specifies an account identifier, an asset, and a transaction type for trading the asset.
3. The method of claim 2 , wherein the at least one data element required to process the transaction comprises a current price of the asset, wherein the transaction type comprises one or more of buying the asset and selling the asset, wherein processing the transaction comprises one or more of estimating a trade settlement, estimating a net asset value, estimating a stock holding, estimating a stock value, estimating a mutual fund value, and estimating a commodity value.
4. The method of claim 1 , further comprising:
receiving a second workload related to a second transaction described by the immutable log, wherein the first and second transactions are described by different entries in the immutable log;
determining, based on the immutable log describing, that the first compute node stores at least one data element required to process the second transaction;
determining utilization levels of the plurality of computing resources of the first compute node;
determining utilization levels of the plurality of links connecting the first compute node to the fabric;
determining, based on the utilization levels of the computing resources and the links of the first compute node, that processing the second workload on the first compute node violates the one or more QoS parameters of the SLA; and
refraining from scheduling the second workload on the first compute node.
5. The method of claim 4 , further comprising:
identifying a second compute node of the plurality of compute nodes proximate to the first compute node;
determining utilization levels of a plurality of computing resources of the second compute node;
determining utilization levels of a plurality of links connecting the second compute node to the fabric;
determining, based on the utilization levels of the computing resources and the links of the second compute node, that processing the second workload on the second compute node satisfies the one or more QoS parameters of the SLA; and
scheduling the second workload on the second compute node.
6. The method of claim 1 , further comprising:
prior to scheduling the workload for processing on the first compute node, dividing the workload into a plurality of units;
determining a grouping of at least two of the plurality of compute nodes, the at least two of the plurality of compute nodes comprising at least the first compute node and a second compute node proximate to the first compute node; and
determining utilization levels of the plurality of computing resources of each compute node in the grouping;
determining utilization levels of a plurality of links connecting each compute node in the grouping to the fabric;
determining, based on the utilization levels of the computing resources and the links of the compute nodes in the grouping, that scheduling the plurality of units of the workload to the compute nodes in the grouping satisfies the one or more QoS parameters of the SLA; and
scheduling the plurality of units of the workload to the compute nodes in the grouping.
7. The method of claim 1 , wherein the QoS parameters comprise: a latency, a jitter, a workload execution time, a throughput, and a bandwidth.
8. A non-transitory computer-readable medium storing instructions executable by a processor circuit cause the processor circuit to:
receive a workload related to an immutable log describing a transaction for processing by at least one compute node of a plurality of compute nodes communicably coupled by a fabric in a cloud computing environment;
determine, based on the immutable log describing the transaction, that a first compute node stores at least one data element required to process the transaction;
determine utilization levels of a plurality of computing resources of the first compute node;
determine utilization levels of a plurality of links connecting the first compute node to the fabric;
determine, based on the utilization levels of the computing resources and the links, that processing the workload on the first compute node satisfies one or more Quality of Service (QoS) parameters specified in a service level agreement (SLA); and
schedule the workload for processing on the first compute node based on the determination that processing the workload on the first compute node satisfies the one or more QoS parameters specified in the SLA.
9. The computer-readable medium of claim 8 , wherein the immutable log specifies an account identifier, an asset, and a transaction type for trading the asset.
10. The computer-readable medium of claim 9 , wherein the at least one data element required to process the transaction comprises a current price of the asset, wherein the transaction type comprises one or more of buying the asset and selling the asset, wherein processing the transaction comprises one or more of estimating a trade settlement, estimating a net asset value, estimating a stock holding, estimating a stock value, estimating a mutual fund value, and estimating a commodity value.
11. The computer-readable medium of claim 8 , the instructions executable by the processor circuit to cause the processor circuit to:
receive a second workload related to a second transaction described by the immutable log, wherein the first and second transactions are described by different entries in the immutable log;
determine, based on the immutable log describing, that the first compute node stores at least one data element required to process the second transaction;
determine utilization levels of the plurality of computing resources of the first compute node;
determine utilization levels of the plurality of links connecting the first compute node to the fabric;
determine, based on the utilization levels of the computing resources and the links of the first compute node, that processing the second workload on the first compute node violates the one or more QoS parameters of the SLA; and
refrain from scheduling the second workload on the first compute node.
12. The computer-readable medium of claim 11 , the instructions executable by the processor circuit to cause the processor circuit to:
identify a second compute node of the plurality of compute nodes proximate to the first compute node;
determine utilization levels of a plurality of computing resources of the second compute node;
determine utilization levels of a plurality of links connecting the second compute node to the fabric;
determine, based on the utilization levels of the computing resources and the links of the second compute node, that processing the second workload on the second compute node satisfies the one or more QoS parameters of the SLA; and
schedule the second workload on the second compute node.
13. The computer-readable medium of claim 8 , the instructions executable by the processor circuit to cause the processor circuit to:
prior to scheduling the workload for processing on the first compute node, divide the workload into a plurality of units;
determine a grouping of at least two of the plurality of compute nodes, the at least two of the plurality of compute nodes comprising at least the first compute node and a second compute node proximate to the first compute node; and
determine utilization levels of the plurality of computing resources of each compute node in the grouping;
determine utilization levels of a plurality of links connecting each compute node in the grouping to the fabric;
determine, based on the utilization levels of the computing resources and the links of the compute nodes in the grouping, that scheduling the plurality of units of the workload to the compute nodes in the grouping satisfies the one or more QoS parameters of the SLA; and
schedule the plurality of units of the workload to the compute nodes in the grouping.
14. The computer-readable medium of claim 8 , wherein the QoS parameters comprise: a latency, a jitter, a workload execution time, a throughput, and a bandwidth.
15. An apparatus, comprising:
a processor circuit; and
a memory storing instructions executable by the processor circuit to cause the processor circuit to:
receive a workload related to an immutable log associated with a transaction to process by at least one compute node of a plurality of compute nodes communicably coupled by a fabric in a cloud computing environment;
determine, based on the immutable log associated with the transaction, that a first compute node stores at least one data element required to process the transaction;
determine utilization levels of a plurality of computing resources of the first compute node;
determine utilization levels of a plurality of links connecting the first compute node to the fabric;
determine that the utilization levels of the computing resources and the links satisfy one or more Quality of Service (QoS) parameters specified in a service level agreement (SLA); and
schedule the workload to process on the first compute node based on the determination that the utilization levels of the computing resources and the links satisfy the one or more QoS parameters specified in the SLA.
16. The apparatus of claim 15 , wherein the immutable log specifies an account identifier, an asset, and a transaction type to trade the asset.
17. The apparatus of claim 16 , wherein the at least one data element required to process the transaction comprises a current price of the asset, wherein the transaction type comprises one or more of a purchase of the asset and a sale of the asset, wherein the first compute node is to process the transaction based on one or more of an estimated trade settlement, an estimated net asset value, an estimated stock holding, an estimated stock value, an estimated mutual fund value, and an estimated commodity value.
18. The apparatus of claim 15 , the memory storing instructions executable by the processor circuit to cause the processor circuit to:
receive a second workload related to a second transaction described by the immutable log, wherein the first and second transactions are described by different entries in the immutable log;
determine, based on the immutable log describing, that the first compute node stores at least one data element required to process the second transaction;
determine utilization levels of the plurality of computing resources of the first compute node;
determine utilization levels of the plurality of links connecting the first compute node to the fabric;
determine that the utilization levels of the computing resources and the links of the first compute node violate the one or more QoS parameters of the SLA; and
refrain from scheduling the second workload on the first compute node.
19. The apparatus of claim 18 , the memory storing instructions executable by the processor circuit to cause the processor circuit to:
identify a second compute node of the plurality of compute nodes proximate to the first compute node;
determine utilization levels of a plurality of computing resources of the second compute node;
determine utilization levels of a plurality of links connecting the second compute node to the fabric;
determine that the utilization levels of the computing resources and the links of the second compute node satisfy the one or more QoS parameters of the SLA; and
schedule the second workload on the second compute node.
20. The apparatus of claim 15 , wherein the QoS parameters comprise: a latency, a jitter, a workload execution time, a throughput, and a bandwidth, the memory storing instructions executable by the processor circuit to cause the processor circuit to:
prior to scheduling the workload for processing on the first compute node, divide the workload into a plurality of units;
determine a grouping of at least two of the plurality of compute nodes, the at least two of the plurality of compute nodes to comprise at least the first compute node and a second compute node proximate to the first compute node; and
determine utilization levels of the plurality of computing resources of each compute node in the grouping;
determine utilization levels of a plurality of links connecting each compute node in the grouping to the fabric;
determine that the utilization levels of the computing resources and the links of the compute nodes in the grouping satisfy the one or more QoS parameters of the SLA; and
schedule the plurality of units of the workload to the compute nodes in the grouping.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US16/385,442 US20190317812A1 (en) | 2018-04-16 | 2019-04-16 | Guaranteed quality of service in cloud computing environments |
US17/700,980 US11599393B2 (en) | 2018-04-16 | 2022-03-22 | Guaranteed quality of service in cloud computing environments |
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201862658060P | 2018-04-16 | 2018-04-16 | |
US201862663477P | 2018-04-27 | 2018-04-27 | |
US16/385,442 US20190317812A1 (en) | 2018-04-16 | 2019-04-16 | Guaranteed quality of service in cloud computing environments |
Related Child Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US17/700,980 Continuation US11599393B2 (en) | 2018-04-16 | 2022-03-22 | Guaranteed quality of service in cloud computing environments |
Publications (1)
Publication Number | Publication Date |
---|---|
US20190317812A1 true US20190317812A1 (en) | 2019-10-17 |
Family
ID=68161618
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US16/385,442 Abandoned US20190317812A1 (en) | 2018-04-16 | 2019-04-16 | Guaranteed quality of service in cloud computing environments |
US17/700,980 Active US11599393B2 (en) | 2018-04-16 | 2022-03-22 | Guaranteed quality of service in cloud computing environments |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US17/700,980 Active US11599393B2 (en) | 2018-04-16 | 2022-03-22 | Guaranteed quality of service in cloud computing environments |
Country Status (1)
Country | Link |
---|---|
US (2) | US20190317812A1 (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20200007414A1 (en) * | 2019-09-13 | 2020-01-02 | Intel Corporation | Multi-access edge computing (mec) service contract formation and workload execution |
CN111104220A (en) * | 2019-12-06 | 2020-05-05 | 北京浪潮数据技术有限公司 | Arm architecture-based server configuration method, system and related equipment |
US20200218579A1 (en) * | 2019-01-08 | 2020-07-09 | Hewlett Packard Enterprise Development Lp | Selecting a cloud service provider |
US10963301B2 (en) * | 2019-07-17 | 2021-03-30 | Google Llc | Scheduling operations on a computation graph |
CN113946438A (en) * | 2021-09-03 | 2022-01-18 | 西安电子科技大学 | Online migration scheduling method, device, equipment and system based on container integration |
EP3958484A3 (en) * | 2020-08-21 | 2022-04-06 | Sap Se | Interface for processing sensor data with hyperscale services |
US11463919B2 (en) * | 2018-04-09 | 2022-10-04 | Huawei Technologies Co., Ltd. | Communications method and apparatus |
US20220327566A1 (en) * | 2021-04-08 | 2022-10-13 | Bank Of America Corporation | System for intelligent and adaptive real time valuation engine using lstm neural networks and multi variant regression analysis |
US20230021278A1 (en) * | 2021-07-12 | 2023-01-19 | Cisco Technology, Inc. | Circuit-Style Network with Co-Routed Bidirectional Network Paths |
WO2023056332A1 (en) * | 2021-09-29 | 2023-04-06 | Commscope Technologies Llc | Device management system for radio access network with interface to hyperscale services portal |
US11645247B2 (en) | 2020-08-21 | 2023-05-09 | Sap Se | Ingestion of master data from multiple applications |
US11720397B2 (en) * | 2019-07-08 | 2023-08-08 | Fujitsu Limited | Information processing method and apparatus to migrate tasks between resources |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20230252503A1 (en) * | 2022-02-09 | 2023-08-10 | Amperity, Inc. | Multi-stage prediction with fitted rescaling model |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8346909B2 (en) * | 2004-01-22 | 2013-01-01 | International Business Machines Corporation | Method for supporting transaction and parallel application workloads across multiple domains based on service level agreements |
US7573819B2 (en) * | 2004-04-01 | 2009-08-11 | Verizon Services Corp. | Methods and apparatus for controlling bandwidth and service in a communications system |
US9521089B2 (en) * | 2014-08-30 | 2016-12-13 | International Business Machines Corporation | Multi-layer QoS management in a distributed computing environment |
US9513967B2 (en) * | 2014-09-18 | 2016-12-06 | International Business Machines Corporation | Data-aware workload scheduling and execution in heterogeneous environments |
US20160321751A1 (en) * | 2015-04-28 | 2016-11-03 | Domus Tower, Inc. | Real-time settlement of securities trades over append-only ledgers |
US10237201B2 (en) * | 2016-05-30 | 2019-03-19 | Dell Products, L.P. | Quality of service (QoS) based device for allocating compute workloads to hosts providing storage and network services in software-based data center |
-
2019
- 2019-04-16 US US16/385,442 patent/US20190317812A1/en not_active Abandoned
-
2022
- 2022-03-22 US US17/700,980 patent/US11599393B2/en active Active
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11463919B2 (en) * | 2018-04-09 | 2022-10-04 | Huawei Technologies Co., Ltd. | Communications method and apparatus |
US11765624B2 (en) * | 2018-04-09 | 2023-09-19 | Huawei Technologies Co., Ltd. | Communications method and apparatus |
US20220417802A1 (en) * | 2018-04-09 | 2022-12-29 | Huawei Technologies Co., Ltd. | Communications method and apparatus |
US20200218579A1 (en) * | 2019-01-08 | 2020-07-09 | Hewlett Packard Enterprise Development Lp | Selecting a cloud service provider |
US11720397B2 (en) * | 2019-07-08 | 2023-08-08 | Fujitsu Limited | Information processing method and apparatus to migrate tasks between resources |
US10963301B2 (en) * | 2019-07-17 | 2021-03-30 | Google Llc | Scheduling operations on a computation graph |
US11755367B2 (en) | 2019-07-17 | 2023-09-12 | Google Llc | Scheduling operations on a computation graph |
US20200007414A1 (en) * | 2019-09-13 | 2020-01-02 | Intel Corporation | Multi-access edge computing (mec) service contract formation and workload execution |
US11924060B2 (en) * | 2019-09-13 | 2024-03-05 | Intel Corporation | Multi-access edge computing (MEC) service contract formation and workload execution |
CN111104220A (en) * | 2019-12-06 | 2020-05-05 | 北京浪潮数据技术有限公司 | Arm architecture-based server configuration method, system and related equipment |
EP3958484A3 (en) * | 2020-08-21 | 2022-04-06 | Sap Se | Interface for processing sensor data with hyperscale services |
US11645247B2 (en) | 2020-08-21 | 2023-05-09 | Sap Se | Ingestion of master data from multiple applications |
US11726846B2 (en) | 2020-08-21 | 2023-08-15 | Sap Se | Interface for processing sensor data with hyperscale services |
US20220327566A1 (en) * | 2021-04-08 | 2022-10-13 | Bank Of America Corporation | System for intelligent and adaptive real time valuation engine using lstm neural networks and multi variant regression analysis |
US20230021278A1 (en) * | 2021-07-12 | 2023-01-19 | Cisco Technology, Inc. | Circuit-Style Network with Co-Routed Bidirectional Network Paths |
CN113946438A (en) * | 2021-09-03 | 2022-01-18 | 西安电子科技大学 | Online migration scheduling method, device, equipment and system based on container integration |
WO2023056332A1 (en) * | 2021-09-29 | 2023-04-06 | Commscope Technologies Llc | Device management system for radio access network with interface to hyperscale services portal |
Also Published As
Publication number | Publication date |
---|---|
US20220283860A1 (en) | 2022-09-08 |
US11599393B2 (en) | 2023-03-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11599393B2 (en) | Guaranteed quality of service in cloud computing environments | |
US11138376B2 (en) | Techniques for information ranking and retrieval | |
US11693917B2 (en) | Computational model optimizations | |
Liu et al. | A speculative approach to spatial‐temporal efficiency with multi‐objective optimization in a heterogeneous cloud environment | |
Junaid et al. | A hybrid model for load balancing in cloud using file type formatting | |
AU2016328959B2 (en) | Updating attribute data structures to indicate trends in attribute data provided to automated modeling systems | |
Corizzo et al. | DENCAST: distributed density-based clustering for multi-target regression | |
US11875408B2 (en) | Techniques for accurate evaluation of a financial portfolio | |
Xu et al. | esDNN: deep neural network based multivariate workload prediction in cloud computing environments | |
Hamrouni et al. | A survey of dynamic replication and replica selection strategies based on data mining techniques in data grids | |
US11094015B2 (en) | Data access and processing system | |
Prabhu et al. | Big data analytics | |
Agrawal et al. | Adaptive real‐time anomaly detection in cloud infrastructures | |
Paul Martin et al. | CREW: Cost and Reliability aware Eagle‐Whale optimiser for service placement in Fog | |
US20220027400A1 (en) | Techniques for information ranking and retrieval | |
Casimiro et al. | A probabilistic model checking approach to self-adapting machine learning systems | |
Wang et al. | Model and algorithm for heterogeneous scheduling integrated with energy-efficiency awareness | |
Ni et al. | Robust factorization machines for credit default prediction | |
Sun et al. | Spatial-temporal attention network for crime prediction with adaptive graph learning | |
Singh et al. | Cloud based evaluation of databases for stock market data | |
Xie et al. | A prediction model of cloud security situation based on evolutionary functional network | |
Hamrouni et al. | Review on data replication strategies in single vs. interconnected cloud systems: Focus on data correlation‐aware strategies | |
Satu et al. | Modeling online customer purchase intention behavior applying different feature engineering and classification techniques | |
Akerkar et al. | Analytics and big data | |
US11861688B1 (en) | Recovery-aware content management |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
STPP | Information on status: patent application and granting procedure in general |
Free format text: DOCKETED NEW CASE - READY FOR EXAMINATION |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: RESPONSE TO NON-FINAL OFFICE ACTION ENTERED AND FORWARDED TO EXAMINER |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: FINAL REJECTION MAILED |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: DOCKETED NEW CASE - READY FOR EXAMINATION |
|
STPP | Information on status: patent application and granting procedure in general |
Free format text: NOTICE OF ALLOWANCE MAILED -- APPLICATION RECEIVED IN OFFICE OF PUBLICATIONS |
|
STCB | Information on status: application discontinuation |
Free format text: ABANDONED -- FAILURE TO PAY ISSUE FEE |