CN106357452A - High-availability framework system with single-point heterogeneous data storage function and realizing method thereof - Google Patents
High-availability framework system with single-point heterogeneous data storage function and realizing method thereof Download PDFInfo
- Publication number
- CN106357452A CN106357452A CN201610872590.6A CN201610872590A CN106357452A CN 106357452 A CN106357452 A CN 106357452A CN 201610872590 A CN201610872590 A CN 201610872590A CN 106357452 A CN106357452 A CN 106357452A
- Authority
- CN
- China
- Prior art keywords
- data
- write
- queue
- log
- write operation
- 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.)
- Granted
Links
Classifications
-
- 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
- H04L67/1095—Replication or mirroring of data, e.g. scheduling or transport for data synchronisation between network nodes
-
- 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
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/069—Management of faults, events, alarms or notifications using logs of notifications; Post-processing of notifications
-
- 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
- H04L67/04—Protocols specially adapted for terminals or networks with limited capabilities; specially adapted for terminal portability
-
- 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
- H04L67/08—Protocols specially adapted for terminal emulation, e.g. Telnet
-
- 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
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
-
- 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
- H04L67/1097—Protocols in which an application is distributed across nodes in the network for distributed storage of data in networks, e.g. transport arrangements for network file system [NFS], storage area networks [SAN] or network attached storage [NAS]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L69/00—Network arrangements, protocols or services independent of the application payload and not provided for in the other groups of this subclass
- H04L69/16—Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Computer Security & Cryptography (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a high-availability framework system with a single-point heterogeneous data storage function. The high-availability framework system comprises a client type base ti-a59-client, a heterogeneous data conversion and storage service module ti-a59-builder, a data write operation log service module ti-a59-log-srv, a data queue framework ti-queue, a message queue service module MQ Server, a single-point deployment type heterogeneous data storage server Single Server DataStore, a write request log service data storage module Write Log DataStore, and a write request control log data storage module Ctrl Log DataStore. The invention also discloses a method for realizing the system. The high-availability framework system has the advantages that the property of high-availability support of single-point deployment storage of the existing framework is realized, the stronger cross-platform deployment capability is realized, multiple data storage types are supported, the data synchronizing between heterogeneous data is supported, and the important properties of data backup and data restoration functions under the condition of support node faults are supported.
Description
Technical field
The invention belongs to computer realm, it is related to a kind of High Availabitity framework of data storage, more particularly to single-point isomery
The High Availabitity frame system of data storage；Moreover, it relates to the High Availabitity frame system of this single-point isomeric data storage
Implementation method.
Background technology
Prior art is to be disposed the High Availabitity framework (see Fig. 1) of data storage by a single-point that external team develops.
In Fig. 1, each module declaration is as follows:
1.dyno: existing single-point stores High Availabitity framework.
The client class libraries of 2.dyno-client: available frame.
The data storage agent service of 3.dyno-proxy: available frame.
The isomorphism data storage service of 4.single server datastore: single-point deployment.
5.rack a/b: single-point disposes storage server, by dyno-proxy and single server datastore group
Become.This framework can allow can only single-point deployment isomorphism data storage accomplish clustered deploy(ment), by the client of this framework
(dyno-client) (write request), the agency service of framework when data write operation is carried out to a certain node
(dyno-proxy) write operation can be synchronized to other nodes in cluster and complete data syn-chronization write (sync write
request).During write operation, individual node breaks down, and client and agency service can select other enabled nodes to complete to count
According to write.Individual node fault does not interfere with the data write of other nodes, thus the High Availabitity realizing one point data storage is write
Enter function.
Equally, rely on the digital independent node selection function of available frame, digital independent behaviour is initiated by framework client
When making (read request), this client can find out an enabled node from cluster, and carries out digital independent from this node,
If the data that node need not be wanted, agency service can forward read the request to data place node (forward read
request).During read operation, individual node breaks down, and client and agency service can be skipped this node and select the next one can
Carry out digital independent with node, achieve the High Availabitity read functions of one point data storage in this way.
Available frame presently, there are several shortcomings as follows:
1. available frame only supports the isomorphism data syn-chronization of two main flow nosql data bases, does not support that isomeric data is synchronous,
Isomeric data is not just synchronously included at the beginning of design and is realized target by available frame.
2. available frame does not support that fault data recovers function, and available frame is can in order to solve the height of one point data storage
With problem, fault recovery is not listed in as Core Feature and is realized in list.Can only manually mode save from other by hand at present
Point copy data recovers data to malfunctioning node, and the reset mode of each data storage is different, the complexity of data recovery
Property has some idea of.
3. available frame is only supported to be deployed in linux/unix operating system, and professional platform independence is slightly weak.All structures of framework
Script uses shell-command, limits it and cannot be carried out building on windows platform and compile.
Content of the invention
The technical problem to be solved in the present invention is to provide a kind of High Availabitity frame system of single-point isomeric data storage, can
Support more data storage class, support isomeric data between data syn-chronization, also can supporting node fault when data backup
Data recovers function.For this reason, the present invention also provides the implementation method of this system.
For solving above-mentioned technical problem, the present invention provides a kind of High Availabitity frame system of single-point isomeric data storage, bag
Include:
Client class libraries ti-a59-client, provides the High Availabitity read-write operation of single-point isomeric data；
M IS and storage services module ti-a59-builder, for receiving by client class libraries ti-a59-
The write data that client submits to, is converted to the form of isomeric data storage support and is written in one point data storage；
Data write operation log service module ti-a59-log-srv, for the data recovery after memory node fault with newly
Increase data initialization during node；
Data queue framework ti-queue, provides high availability to support for data write operation；
Message Queuing Services module mq server, is route, duplication and the persistence work(of systems with data write request
Energy；
The isomeric data storage server single server datastore of single-point deployment, the depositing of responsible isomeric data
Storage, and data write is provided and reads interface；
Write request log services data memory module write log datastore, preserves data write operation log services
The data write operation that module ti-a59-log-srv receives, for the data syn-chronization after fault or during newly-increased a59 memory node；
Write request controls daily record data memory module ctrl log datastore, preserves the number of current a59 memory node
According to write operation execution journal it is ensured that same data write operation will not be performed a plurality of times.
As currently preferred technical scheme, described client class libraries ti-a59-client provides data write operation meeting
Data is compressed and is distributed to each a59 memory node via data queue framework ti-queue, all nodes all can receive
Identical data simultaneously completes data write；It is based on single-point that described client class libraries ti-a59-client provides data reading operation
The primary api of data storage, being sent read request by the load-balancing algorithm that tcp/http proxy acts on behalf of can to any one
A59 memory node completes digital independent.
As currently preferred technical scheme, described tcp/http proxy be provide high availability, load balancing with
And the agent software based on tcp and http application, realize the High Availabitity of a59 memory node data is read by coordinating ha software
Extract operation；Described tcp/http proxy software is realized using haproxy+keepalived, haproxy be a increase income, complete
Completely free tcp/http proxy agent software, keepalived is to ensure in cluster management that one of cluster High Availabitity is increased income
Service software, is avoided that haproxy Single Point of Faliure it is ensured that the high availability of haproxy itself；Based on tcp/http
Proxy, business service just enables the High Availabitity read operation for a59 memory node data.
As currently preferred technical scheme, a described M IS and storage services module ti-a59-
Builder is constituted together with the isomeric data storage server single server datastore of a described single-point deployment
One described a59 memory node；In order to ensure the high availability of reading and writing data, need in set of environmental to build the a59 of more than 2
Memory node；Described M IS and storage services module ti-a59-builder are in order to ensure the idempotent of write data requests
Property, daily record data memory module ctrl log datastore can be controlled by accessing described write request before writing data every time
Guarantee that identical Data Data will not be repeatedly written.
The data write request submitted to as currently preferred technical scheme, described client class libraries ti-a59-client
Copied to by described data queue framework ti-queue and also can replicate a data while each a59 example memory node to institute
State data write operation log service module ti-a59-log-srv, write operation can be serialized daily record and deposit by ti-a59-log-srv
Storage, daily record data be used for subsequent node fault or newly-increased when data recovery；Ti-a59-log-srv default policy is to retain 7 days
Interior data logging, the data before cleaning out daily 7 days.
As currently preferred technical scheme, main base when described data queue framework ti-queue and ti-a59 is integrated
Replicate in described Message Queuing Services module mq server completion message, described client class libraries ti-a59-client is submitted to
Data duplication to each a59 memory node and described data write operation log service module ti-a59-log-srv daily record section
Point.
As currently preferred technical scheme, described Message Queuing Services module mq server adopts rabbitmq to take
Business is as a59 message-oriented middleware；When M IS and storage services module ti-a59-builder or data write operation day
When will service module ti-a59-log-srv fault does not normally connect, store data in Message Queuing Services module it is ensured that
Message is not lost, and continues with the data in queue after ti-a59-builder or ti-a59-log-srv reconnects；By many
The cluster that Message Queuing Services module mq server described in platform is set up constitutes mq server cluster it is ensured that mq server
The high availability of itself.
Additionally, the present invention also provides a kind of method realizing asynchronous data write using said system, comprise the steps:
1) application program is when initiating asynchronous data write operation, by calling the writer of ti-a59-client by number
According to each a59 memory node of write；
2) writer calls the queue of ti-queue to add data for after data interpolation write time stamp and unique mark
Queue；
3) ti-queue data of joining the team is serialized and is assembled into message object；
4) message object is sent on corresponding mq broker in mq server ti-queue；ti-a59-client
Need to specify the group number of a59 memory node in initialization, a group number corresponds to a mq broker；
5) after mq broker receives message, rely on message to replicate distribution mechanisms, message is write all of all bindings
In queue；
6) ti-a59-builder, after receiving message, can carry out unserializing and decompression obtains initial data；
7) ti-a59-builder can be according to the write time of data and unique mark in ctrl log datastore
Checked；
8) if write operation was not carried out, builder is called to initiate data write operation；
9), after ti-a59-builder receives data, call the data conversion code that user specifies and by calling single
Standard api of server datastore completes data write；
10) after write success, ti-a59-builder one write operation daily record of record to ctrl log datastore.
For this reason, the present invention also provides a kind of method that said system realizes synchronized write data, comprise the steps:
1) application program, when initiating data syn-chronization write operation, needs first to call the scenario of ti-a59-client
Start a scene, the effect of scene is for being associated with an a59 data storage；
2), after scene creation finishes, write data into each a59 storage section by calling the writer of ti-a59-client
Point；
3), when writer receives synchronous write request, current bindings a59 memory node can be read from scenario, if referred to
Rule can wait the write operation returning result of node to be specified, if do not specified, waits the write operation result of first return；
4) writer calls the queue of ti-queue to add data for after data interpolation write time stamp and unique mark
Queue；
5) ti-queue data of joining the team is serialized and is assembled into message object；
6) message object is sent on corresponding mq broker ti-queue；Ti-a59-client needs in initialization
Specify the group number of a59 memory node, a group number corresponds to a mq broker；
7) after mq broker receives message, rely on message to replicate distribution mechanisms, message is write all of all bindings
In queue；
8) ti-a59-builder, after receiving message, can carry out unserializing and decompression obtains initial data；
9) ti-a59-builder can be according to the write time of data and unique mark in ctrl log datastore
Checked；
10) if write operation was not carried out, builder is called to initiate data write operation；
11), after ti-a59-builder receives data, call the data conversion code that user specifies and by calling
Standard api of single server datastore completes data write；
12) after write success, ti-a59-builder one write operation daily record of record to ctrl log datastore；
13) result of write operation can be sent to by mq server, ti-queue and be write by initiation by ti-a59-builder
In the specified queue that the ti-a59-client of request monitors, finally receiving write operation by the writer initiating write operation at that time should
Answer；
14) if step 1) create scene when not specified node, then the node serial number of return can be updated in scene,
So that subsequent write operation can be locked on same node；
15) call scenario whole synchronization call process.
For this reason, the present invention also provides a kind of method that said system realizes digital independent, the digital independent of ti-a59 is base
Store primary api in one point data, the load-balancing algorithm acted on behalf of by tcp/http proxy is sent read request to any
One available a59 memory node completes digital independent, specifically includes following steps:
1) the primary api of application call single server datastore initiates data read operation；Primary
Api points to the tcp/http proxy address of ti-a59；
2) tcp/http proxy, according to load-balancing algorithm, forwards read the request to an available one point data and deposits
Complete read operation on storage server, and former road returns to application program by reading result.
Compared with prior art, the beneficial effects of the present invention is:
1st, the present invention, in addition to the characteristic that the high availability that the single-point comprising available frame disposes storage is supported, also has
For higher cross-platform deployment ability, and the synchronous weight that cannot support with available frames such as fault data recoveries of isomeric data
Want characteristic.Provide a set of more sophisticated, data reliability is higher, the higher single-point deployment data storage of framework extensibility is high
Availableization implementation.
2. the present invention provides general high availability to support for one point data storage, supports that multinode is simultaneously written, loads
Digital independent under balance policy and and ensure the concordance of data.Only support linux/unix operation system compared to existing technology
System, and the framework of the present invention can be deployed in nearly all mainstream operation system.Prior art supports 2 kinds of data storage classes
Type, the framework of the present invention then supports more data storage class.
3. support the data syn-chronization between isomeric data, such as by the data syn-chronization of relevant database to chart database
In, or relation data is simultaneously written in nosql data base, prior art does not support the synchronization between isomeric data at present.
4. data backup data during supporting node fault recovers function, and prior art does not possess data after fault at present
Recover function.
Brief description
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1 is the structural representation of the High Availabitity framework that existing single-point disposes data storage；Wherein, dyno: existing list
Point storage High Availabitity framework；The client class libraries of dyno-client: existing framework；Dyno-proxy: the data of existing framework is deposited
Storage agency service；The isomorphism data storage service of single server datastore: single-point deployment；Rack a/b: single-point portion
Administration's storage server, is made up of dyno-proxy and single server datastore.
Fig. 2 is the structural representation of the High Availabitity frame system of single-point isomeric data storage of the present invention；Wherein, ti-a59:
The single-point isomery storage High Availabitity framework of present patent application；The client class libraries of ti-a59-client:ti-a59；ti-a59-
The M IS of builder:ti-a59 and storage service；The data write operation daily record of ti-a59-log-srv:ti-a59
Service；Data queue's framework of ti-queue:ti-59 built-in general-purpose；Mq server: Message Queuing Services；mq server
Cluster: the cluster set up by multiple stage mq server；Tcp/http proxy (ha): provide high availability, load balancing with
And the agent software based on tcp/http, and ensure the High Availabitity of agent software by cluster management software；single server
The isomeric data storage service of datastore: single-point deployment；Write log datastore (ha): ti-a59 write request day
Will service data stores；Ctrl log datastore (ha): ti-a59 write request controls daily record data storage；a59log
Server:a59 log server, is made up of ti-a59-log-srv and write log datastore；a59instance a/
B:a59 memory node, is made up of ti-a59-builder and single server datastore.
Fig. 3 is the flow chart of asynchronous write data in the present invention；Wherein, application: application program, calls ti-a59
Framework completes data write；The writer interface providing in ti-a59-client.writer:ti-a59-client；ti-
Queue.amqpqueue: class is realized based on the ti-a59 of amqp advanced message queue protocol；The disappearing of mq broker: message queue
Breath controller；The message queue of ti-a59-builder.queuewithreplylistener:ti-a59-builder is monitored
Device；Ctrl log datastore: preserve the data storage that write operation controls daily record；Ti-a59-builder.builder:ti-
The onboard data composer of a59-builder；The data storage of single server datastore: single-point deployment.
Fig. 4 is the flow chart of synchronized write data in the present invention；Wherein, application: application program, calls ti-a59
Framework completes data write；Call scene in ti-a59-client.scenairo:ti-a59-client, need during synchronization call
Scene to be opened and closed；The writer interface providing in ti-a59-client.writer:ti-a59-client；ti-
Queue.amqpqueue: class is realized based on the ti-a59 of amqp advanced message queue protocol；The disappearing of mq broker: message queue
Breath controller；The message queue of ti-a59-builder.queuewithreplylistener:ti-a59-builder is monitored
Device；Ctrl log datastore: preserve the data storage that write operation controls daily record；Ti-a59-builder.builder:ti-
The onboard data composer of a59-builder；The data storage of single server datastore: single-point deployment.
Fig. 5 is the flow chart of digital independent in the present invention；Wherein, application: application program, calls ti-a59 frame
The tcp/http proxy assembly of frame completes digital independent；The mark that the storage of datastore native api: one point data provides
Quasi- data access interface；Tcp/http proxy: provide high availability, load balancing and based on tcp/http application proxy soft
Part；The isomeric data storage of single server datastore: single-point deployment.
Specific embodiment
In conjunction with the accompanying drawings, the present invention is further detailed explanation.These accompanying drawings are the schematic diagram of simplification, only with
Illustration illustrates the basic structure of the present invention, and therefore it only shows the composition relevant with the present invention.
One as shown in Fig. 2 the intraware of the High Availabitity frame system of single-point isomeric data of the present invention storage is introduced such as
Under:
The entitled ti-a59 of High Availabitity frame system of the single-point isomeric data storage of the present invention.
Ti-a59 framework comprise four sub-projects ti-a59-client, ti-a59-builder, ti-a59-log-srv,
Framework ti-queue is realized in ti-a59-share and conventional data queue, and all items code adopts java language real
Existing.
Rely on the professional platform independence of java language, this frame system can be deployed in nearly all mainstream operation system, and
Prior art framework can only be deployed on linux/unix platform.
1. below (refer to Fig. 2) is introduced to each assembly being related in ti-a59 overall plan and module.
1.1 ti-a59-client
The client class libraries of ti-a59, provides the High Availabitity read-write operation of single-point isomeric data.
Data can be compressed and be distributed to each a59 memory node via ti-queue framework by data write operation, own
Node all can receive identical data and complete data write.
Data reading operation is to store primary api, the load balancing acted on behalf of by tcp/http proxy based on one point data
Read request is sent and completes digital independent to any one available a59 memory node by algorithm.
The complicated equalization algorithm that tcp/http proxy agency supports typically has several as follows:
1) roundrobin: simple poll, each server uses in turn according to weight, in the process time of server
In the case of mean allocation, this is algorithm the most smooth and fair.
2) leastconn: connect the minimum server priority of number and receive connection.
3) source: Hash is carried out to request source ip address, with the weight of available server sum divided by cryptographic Hash, according to
Result is allocated.As long as server is normal, always same client ip address accesses same server.If Hash
Result changes with available server quantity, then client can be directed to different servers.
1.2 ti-a59-builder
The M IS of ti-a59 and storage services module, are responsible for receiving the write submitted to by ti-a59-client
Data, is converted to the form of isomeric data storage support and is written in one point data storage.
Isomeric data storage (the single server datastore) that one ti-a59-builder is disposed with single-point
Rise and constitute an a59 memory node (instance).In order to ensure the high availability of reading and writing data, at least need in set of environmental
Build the a59 memory node of 2 or more.
Ti-a59-builder, in order to ensure the idempotence of write data requests, can be controlled by access before writing data every time
Daily record (ctrllog) processed guarantees that identical Data Data will not be repeatedly written.
1.3 ti-a59-log-srv
The data write operation log service module of ti-a59, for the data recovery after a59 memory node fault and newly-increased
Data initialization during node.
The data write request that ti-a59-client submits to copies to the same of each a59 example memory node by ti-queue
When also can replicate a data to ti-a59-log-srv, write operation can be serialized daily record storage by ti-a59-log-srv,
Daily record data can be used for subsequent node fault or newly-increased when data recovery.
Ti-a59-log-srv default policy is the data logging retaining in 7 days, the number before cleaning out daily 7 days
According to.
1.4 ti-queue
Ti-queue is general data queue's framework, provides high availability to support for ti-a59 data write operation.
It is based primarily upon mq server completion message when integrated with ti-a59 to replicate, the data duplication that ti-a59-client is submitted to arrives often
Individual a59 memory node and ti-a59-log-srv Journal node.
The middleware that 2.ti-a59 framework uses when building:
2.1 mq server
Mq server is Message Queuing Services module, provides the route of data write request, duplication and persistently for a59 framework
The functions such as change.Mq server has realization of much optionally increasing income, and author compares recommendation rabbitmq and services as in a59 message
Between part.
Rabbitmq is a Message Queuing system increasing income, and uses elrang language development, is amqp (advanced message queue
Agreement) standard implementation, support message persistence.
When ti-a59-builder or ti-a59-log-srv fault does not normally connect, data can be stored in and disappear
It is ensured that message is not lost in breath queue service module mq server, when ti-a59-builder or ti-a59-log-srv weight
Lian Houke continues with the data in queue.
Mq server cluster (adopting rabbitmq cluster in Fig. 2) is the collection set up by multiple stage mq server
Group is it is ensured that the mq server high availability of itself.
2.2 tcp/http proxy(ha)
Tcp/http proxy provides high availability, load balancing and the agent software based on tcp and http application, leads to
Cross cooperation ha software and realize agency service high availability.
The present embodiment preferably uses haproxy+keepalived as the realization of tcp/http proxy (ha).
Haproxy be a increase income, completely free tcp/http proxy agent software.
Keepalived is a service software of increasing income ensureing cluster High Availabitity in cluster management, can avoid haproxy
Single Point of Faliure occurs it is ensured that the high availability of haproxy itself.、
Based on tcp/http proxy, business service just can achieve that the High Availabitity for a59 memory node data reads behaviour
Make.
3.ti-a59 the data storage that framework is related to:
3.1 single server datastore
The isomeric data storage server of single-point deployment, the storage of responsible isomeric data, and data write is provided and reads
Interface.These data storages itself do not have the support of High Availabitity, do not support the synchronizing function of isomeric data yet.
The target of ti-a59 is just to provide for data access and the isomeric data synchronizing function of High Availabitity.
In an a59 memory node cluster, different types of isomeric data memory node can be disposed simultaneously, and existing
In one cluster of framework, all memory node storage classes must be identical.
3.2 write log datastore(ha)
The write request log services data storage of ti-a59, preserves the data write operation that ti-a59-log-srv receives, and uses
Data syn-chronization after fault or during newly-increased a59 memory node.
The present embodiment preferably employs mysql data base as data storage, and adopts master+slave+keepalived
Deployment way guarantees the high availability (ha) of daily record storage.
3.3 ctrl log datastore(ha)
The write request of ti-a59 controls daily record data storage, preserves the execution of the data write operation of current a59 memory node
Daily record is it is ensured that same data write operation will not be performed a plurality of times.
The present embodiment preferably employs mysql data base as data storage, and adopts master+slave+keepalived
Deployment way guarantees the high availability (ha) of daily record storage.
The High Availabitity reading and writing data of 4.ti-a59 is supported
The ability of High Availabitity read-write can be provided for the data storage of single-point deployment, be main characteristics and the innovation of ti-a59
One of point.
The data write operation that application call ti-a59-client initiates, this number is written simultaneously one according to meeting
In all enabled nodes of group a59 memory node, each node can write identical data.The data class of these enabled nodes
Type identical can rely on the synchronous function of the isomeric data of a59 it is also possible to differ, and a secondary data of ti-a59-client is write
Operation can be simultaneously written in different types of data storage.
Even if had one malfunctions cannot write at that time, if having at least one enabled node it is ensured that
This write operation can correctly execute.
During the data reading operation that application program is initiated, the tcp/http proxy assembly in ti-a59 framework can be
This reads and selects an available a59 memory node, if there is multiple enabled nodes, then can be selected according to load-balancing algorithm
One optimal accessed node.Even if so initiating certain one malfunctions of read operation, as long as there being at least one enabled node,
This data read operation would not be affected.
For write operation, ti-a59 supports the different data High Availabitity data writing mode of two kinds of asynchronous and synchronous.For
Read operation, ti-a59 provides the High Availabitity data reading mode of holding load balanced way.
5. isomeric data is synchronous
One of synchronous key property as ti-a59 of isomeric data, is that prior art is not supported at present.
One group of a59 memory node cluster supports the one point data storage service of deployment different types of data, supports isomery number
According to synchronous exactly one of characteristic of a59 framework, and in available frame storage cluster, only support the number of deployment same type
According to storage.
Ti-a59 (with reference to Fig. 3 and Fig. 4) during carrying out data write, can pass through ti-a59-
Builder.builder (the onboard data composer of ti-a59-builder) class completes M IS and assembling.
Different builder achieves the conversion data write operation from java categorical data to different types of data,
By configure different builder in ti-a59-builder and realize it is possible to complete such as neo4j, memsql,
Redis, the isomeric data between the types of databases such as memsql, mysql is synchronous.
6. the fault data of memory node recovers
Ti-a59 supports to recover function for memory node fault data, this be available frame do not support another is heavy
Want characteristic.
By the data write operation backup services building together with ti-a59-log-srv with write log datastore,
A data can be replicated when every secondary data writes and backed up from mq server queue.
After certain a59 memory node breaks down and restarts, the request of data not write persistence can be saved in assembly team
In row, automatically can obtain all data after fault time point after node is restarted from queue.
If the data in assembly queue is surprisingly lost, can also by ti-a59-log-srv will fault time point so far
All of batch data is directed in the ti-a59-builder of malfunctioning node, and ti-a59-builder can ensure data write
In the case of conforming, the data change operation that all nodes were not carried out all is reset one time, to reach data after fault
The purpose recovered.
Equally, fault data recovers the data initialization that function can also be used for newly-increased node, and newly-increased node adds a59 storage
The a certain secondary data full backup in nearly 7 days is comprised, by specifying data backup time, ti-a59-log-srv meeting during node group
Automatically all batch data so far after this time point are synchronized in newly-increased node.
2nd, the present invention also provides the implementation method of the High Availabitity frame system of above-mentioned single-point isomeric data storage, concrete ginseng
See Fig. 3-Fig. 5.
1st, the High Availabitity reading and writing data of ti-a59 is supported, for write operation, ti-a59 supports two kinds of differences of asynchronous and synchronous
Data High Availabitity data writing mode.For read operation, ti-a59 provides the High Availabitity data of holding load balanced way
Reading manner.
1.1 asynchronous data writes
The sequential chart that call of an asynchronous data write request be presented herein below and specification specified (referring to Fig. 3) realized in inside:
1) application program is when initiating asynchronous data write operation, by calling the writer of ti-a59-client by number
According to writing each a59 memory node, because being asynchronous call, application program, without waiting for write operation response, directly can execute
Follow-up code.
2) writer (the writer interface providing in ti-a59-client.writer, ti-a59-client) is data
Call ti-queue that data is added queue after adding write time stamp and unique mark.
3) ti-queue data of joining the team is serialized and is assembled into message object, the meeting when data exceedes specified size
It is compressed operating.
4) message object is sent to corresponding mq broker (the message control of message queue in mq server by ti-queue
Device processed) on.Ti-a59-client needs to specify the group number of a59 memory node in initialization, and a group number corresponds to a mq
broker.
5) after mq broker receives message, rely on message to replicate distribution mechanisms, message is write all of all bindings
In queue, each queue is monitored by the ti-a59-builder service of an a59 memory node.As long as there being message to enter
Queue, will be received by the queuewithreplylistener of ti-a59-builder and be processed.
6) ti-a59-builder, after receiving message, can carry out unserializing and decompression obtains initial data.
7) ti-a59-builder can be according to the write time of data and unique mark in ctrl log datastore
Checked, if write operation exists, ignored this write operation.
8) if write operation was not carried out, builder is called to initiate data write operation.
9), after ti-a59-builder receives data, call the data conversion code that user specifies and by calling single-point portion
Standard api of the isomeric data storage server single server datastore of administration completes data write.
10) after write success, ti-a59-builder one write operation daily record of record to ctrl log datastore.
So far, an asynchronous data write flow process all terminates.
1.2 synchronized write data flow processs
Synchronous writing operation and the main difference of asynchronous write operation are embodied in (sequential in the mode that ti-a59-client calls
Figure refers to Fig. 4).
1) application program, when initiating data syn-chronization write operation, needs first to call ti-a59-client.scenario to open
A dynamic scene, the effect of scene is for being associated with an a59 data storage, can specify one when creating scene
A59 memory node is it is also possible to automatically select a writing speed node the fastest by system.
2), after scene creation finishes, write data into each a59 storage section by calling the writer of ti-a59-client
Point, because being synchronization call, application program needs to wait the response of write operation.
3), when writer receives synchronous write request, current bindings a59 memory node can be read from scenario, if referred to
Rule can wait the write operation returning result of node to be specified, if do not specified, waits the write operation result of first return.
4) the 4th to the 12nd step is similar with asynchronous write data requests, and except for the difference that ti-a59-client can monitor ti-queue
Response queue, receive write operation returned data.
5) the 13rd to the 15th step, the result of write operation can be passed through mq server (in Fig. 4 by ti-a59-builder
Rabbitmq), ti-queue be sent to by initiate write request ti-a59-client monitor specified queue in, finally by work as
The writer of Shi Faqi write operation receives write operation response.Because each ti-a59-builder completes data and writing and returns
The write operation response of oneself, if specified a59 memory node when creating scene at that time, then writer only can process specified node
The data returning, otherwise processes first data returning to ti-a59-client.
6) the 16th step, if not specified node when creating scene, then the node serial number of return can be updated in scene,
So that subsequent write operation can be locked on same node.Although it should be noted that having locked write node, follow-up number
Still execute on all a59 nodes according to write operation, simply returning result receives from specified node.
7) the 17th step, calls scenario to complete whole synchronization call process.If program needs order is initiated repeatedly same
Step writes data, only need to recall scenario after all write operations finish.
So far, a synchrodata write flow process all terminates.
1.3 digital independent flow processs
The digital independent of ti-a59 is to store primary api based on one point data, by bearing that tcp/http proxy acts on behalf of
Read request is sent and completes digital independent to any one available a59 memory node by load equalization algorithm.Referring to Fig. 5, concrete bag
Include as follows:
1) the primary api of application call one point data storage initiates data read operation；The target that primary api points to
Address is not one point data storage address, and is directed to the tcp/http proxy address of ti-a59.
2) tcp/http proxy, according to load-balancing algorithm, forwards read the request to an available one point data and deposits
Complete read operation in storage server (single server datastore), and former road returns to application journey by reading result
Sequence.
So far, a digital independent flow process all terminates.
2. isomeric data is synchronous
By realizing the builder interface of ti-a59-builder, user can voluntarily extend all kinds of isomeric data types
Between data syn-chronization function.
Ti-a59-builder provides primary affairs for builder and supports, the statement formula thing based on spring
Business definition is it is ensured that the acid attribute (atomicity, concordance, isolation and persistency) of data write operation.Work as data write operation
When abnormal occur, can automatically trigger transaction rollback it is ensured that data write operation will not be persisted in storage.
User can be by a data write operation, and be converted into once even multiple different pieces of information according to the actual requirements writes behaviour
Make, the primary affairs support of ti-a59-builder can ensure atomicity and the concordance of multiple write operation.
Based on the message copy function of assembly, ti-a59 can support to copy to a secondary data write request different types of
Isomeric data memory node group, completes the write operation of different isomerization data simultaneously.
2. the fault data of memory node recovers
When memory node breaks down, the data being currently written into is because the guarantee of affair mechanism is will not persist
To in storage.The same transactional attribute because of mq server message queue, the message of this write request still can be retained in message
In queue, with the follow-up all write requests reaching together persistence in message queue.
Because the presence of mq server cluster, even if event in the mq server node being responsible for storage message at that time
Barrier, it is to be written in queue again that a59 memory node still can get all preservations by other available mq server nodes
Data.
If data to be written is surprisingly removed in mq server queue, user still can be taken by ti-a59-log-srv
Batch data is synchronized to a59 memory node to be restored by file mode by business, is responsible in order by ti-a59-builder
Again all untreatment datas are imported.This batch data reset mode avoids the number that user passes through manual reversion malfunctioning node
According to solving the problems, such as that the reset mode of different isomerization data type is different.
Ti-a59-builder equally can carry out data write and control inspection, the write having executed in data recovery
Operation will not be repeatedly executed the data consistency it is ensured that present node.
Fault data recovers the data initialization that function is equally applicable to newly-increased node, by specifying the initial of data syn-chronization
Time, ensured by the way of standby recoverys+increment synchronization entirely newly-increased node can quickly and other memory nodes data holding
Synchronous.
With the above-mentioned desirable embodiment according to the present invention for enlightenment, by above-mentioned description, relevant staff is complete
Entirely various change and modification can be carried out in the range of without departing from this invention technological thought.The technology of this invention
The content that property scope is not limited in description it is necessary to determine its technical scope according to right.
Claims (10)
1. a kind of High Availabitity frame system of single-point isomeric data storage is it is characterised in that include:
Client class libraries ti-a59-client, provides the High Availabitity read-write operation of single-point isomeric data；
M IS and storage services module ti-a59-builder, for receiving by client class libraries ti-a59-
The write data that client submits to, is converted to the form of isomeric data storage support and is written in one point data storage；
Data write operation log service module ti-a59-log-srv, for the data recovery after memory node fault and newly-increased section
Data initialization during point；
Data queue framework ti-queue, provides high availability to support for data write operation；
Message Queuing Services module mq server, is route, duplication and the persistence function of systems with data write request；
The isomeric data storage server single server datastore of single-point deployment, the storage of responsible isomeric data, and
Data write is provided and reads interface；
Write request log services data memory module write log datastore, preserves data write operation log service module
The data write operation that ti-a59-log-srv receives, for the data syn-chronization after fault or during newly-increased a59 memory node；
Write request controls daily record data memory module ctrl log datastore, and the data preserving current a59 memory node is write
The execution journal of operation is it is ensured that same data write operation will not be performed a plurality of times.
2. the system as claimed in claim 1 it is characterised in that: described client class libraries ti-a59-client provides data to write
Data can be compressed and be distributed to each a59 memory node via data queue framework ti-queue by operation, and all nodes are all
Identical data can be received and complete data write；It is base that described client class libraries ti-a59-client provides data reading operation
Store primary api in one point data, the load-balancing algorithm acted on behalf of by tcp/http proxy is sent read request to any
One available a59 memory node completes digital independent.
3. system as claimed in claim 2 it is characterised in that: a described M IS and storage services module ti-
A59-builder is together with the isomeric data storage server single server datastore of a described single-point deployment
Constitute a described a59 memory node；In order to ensure the high availability of reading and writing data, need in set of environmental to build more than 2
A59 memory node；Described M IS and storage services module ti-a59-builder are in order to ensure write data requests
Idempotence, daily record data memory module ctrl log can be controlled by accessing described write request before writing data every time
Datastore guarantees that identical Data Data will not be repeatedly written.
4. the system as claimed in claim 1 it is characterised in that: the data that described client class libraries ti-a59-client submits to
Write request is copied to by described data queue framework ti-queue and also can replicate a number while each a59 example memory node
According to described data write operation log service module ti-a59-log-srv, write operation can be serialized by ti-a59-log-srv
Daily record store, daily record data be used for subsequent node fault or newly-increased when data recovery；Ti-a59-log-srv default policy is
Retain the data logging in 7 days, the data before cleaning out daily 7 days.
5. the system as claimed in claim 1 it is characterised in that: when described data queue framework ti-queue and ti-a59 are integrated
It is based primarily upon described Message Queuing Services module mq server completion message to replicate, by described client class libraries ti-a59-
The data duplication that client submits to is to each a59 memory node and described data write operation log service module ti-a59-
Log-srv Journal node.
6. the system as claimed in claim 1 it is characterised in that: described Message Queuing Services module mq server adopts
Rabbitmq services as a59 message-oriented middleware；When M IS and storage services module ti-a59-builder or number
When normally not connecting according to write operation log service module ti-a59-log-srv fault, store data in Message Queuing Services mould
It is ensured that message is not lost in block, continue with queue after ti-a59-builder or ti-a59-log-srv reconnects
Data；Described in multiple stage Message Queuing Services module mq server set up cluster constitute mq server cluster it is ensured that
The mq server high availability of itself.
7. system as claimed in claim 2 it is characterised in that: described tcp/http proxy be provide high availability, load
Equilibrium and the agent software based on tcp and http application, realize the height for a59 memory node data by coordinating ha software
Available read operation；Described tcp/http proxy software is realized using haproxy+keepalived, and haproxy is a opening
Source, completely free tcp/http proxy agent software, keepalived is to ensure the one of cluster High Availabitity in cluster management
Individual service software of increasing income, is avoided that haproxy Single Point of Faliure it is ensured that the high availability of haproxy itself；Based on tcp/
Http proxy, business service just enables the High Availabitity read operation for a59 memory node data.
8. a kind of using the system as claimed in claim 1 realize asynchronous data write method it is characterised in that include as follows
Step:
1) data, when initiating asynchronous data write operation, is write by application program by calling the writer of ti-a59-client
Enter each a59 memory node；
2) writer calls the queue of ti-queue that data is added team for after data interpolation write time stamp and unique mark
Row；
3) ti-queue data of joining the team is serialized and is assembled into message object；
4) message object is sent on corresponding mq broker in mq server ti-queue；Ti-a59-client is first
Need to specify the group number of a59 memory node during beginningization, a group number corresponds to a mq broker；
5), after mq broker receives message, rely on message to replicate distribution mechanisms, message is write all queue of all bindings
In；
6) ti-a59-builder, after receiving message, can carry out unserializing and decompression obtains initial data；
7) ti-a59-builder can be carried out in ctrl log datastore according to the write time of data and unique mark
Check；
8) if write operation was not carried out, builder is called to initiate data write operation；
9), after ti-a59-builder receives data, call the data conversion code that user specifies and by calling single
Standard api of server datastore completes data write；
10) after write success, ti-a59-builder one write operation daily record of record to ctrl log datastore.
9. a kind of the method for synchronized write data is realized it is characterised in that including following walking using the system as claimed in claim 1
Rapid:
1) application program, when initiating data syn-chronization write operation, needs first to call the scenario of ti-a59-client to start
One scene, the effect of scene is for being associated with an a59 data storage；
2), after scene creation finishes, write data into each a59 memory node by calling the writer of ti-a59-client；
3), when writer receives synchronous write request, current bindings a59 memory node can be read from scenario, if specified,
The write operation returning result of the nodes to be specified such as meeting, if do not specified, waits the write operation result of first return；
4) writer calls the queue of ti-queue that data is added team for after data interpolation write time stamp and unique mark
Row；
5) ti-queue data of joining the team is serialized and is assembled into message object；
6) message object is sent on corresponding mq broker ti-queue；Ti-a59-client needs to refer in initialization
Determine the group number of a59 memory node, a group number corresponds to a mq broker；
7), after mq broker receives message, rely on message to replicate distribution mechanisms, message is write all queue of all bindings
In；
8) ti-a59-builder, after receiving message, can carry out unserializing and decompression obtains initial data；
9) ti-a59-builder can be carried out in ctrl log datastore according to the write time of data and unique mark
Check；
10) if write operation was not carried out, builder is called to initiate data write operation；
11), after ti-a59-builder receives data, call the data conversion code that user specifies and by calling single
Standard api of server datastore completes data write；
12) after write success, ti-a59-builder one write operation daily record of record to ctrl log datastore；
13) result of write operation can be sent to by initiation write request by ti-a59-builder by mq server, ti-queue
Ti-a59-client monitor specified queue in, finally write operation response is received by the writer initiating write operation at that time；
14) if step 1) create scene when not specified node, then the node serial number of return can be updated in scene so that
Subsequent write operation can be locked on same node；
15) scenario is called to complete whole synchronization call process.
10. a kind of the method for digital independent is realized it is characterised in that the number of ti-a59 using the system as claimed in claim 1
It is that primary api is stored based on one point data according to reading, the load-balancing algorithm acted on behalf of by tcp/http proxy is by read request
Send and complete digital independent to any one available a59 memory node, specifically include following steps:
1) the primary api of application call single server datastore initiates data read operation；Primary api refers to
Tcp/http proxy address to ti-a59；
2) tcp/http proxy, according to load-balancing algorithm, forwards read the request to available one point data storage clothes
Complete read operation on business device, and former road returns to application program by reading result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610872590.6A CN106357452B (en) | 2016-09-29 | 2016-09-29 | A kind of the High Availabitity frame system and its implementation of the storage of single-point isomeric data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610872590.6A CN106357452B (en) | 2016-09-29 | 2016-09-29 | A kind of the High Availabitity frame system and its implementation of the storage of single-point isomeric data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106357452A true CN106357452A (en) | 2017-01-25 |
| CN106357452B CN106357452B (en) | 2019-06-04 |
Family
ID=57866030
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610872590.6A Active CN106357452B (en) | 2016-09-29 | 2016-09-29 | A kind of the High Availabitity frame system and its implementation of the storage of single-point isomeric data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106357452B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107612950A (en) * | 2016-07-11 | 2018-01-19 | 阿里巴巴集团控股有限公司 | A kind of method, apparatus, system, electronic equipment that service is provided |
| CN107832138A (en) * | 2017-09-21 | 2018-03-23 | 南京邮电大学 | A kind of implementation method of the High Availabitity namenode models of flattening |
| CN108255929A (en) * | 2017-11-30 | 2018-07-06 | 杭州博世数据网络有限公司 | For the processing method of database neo4j |
| CN108549584A (en) * | 2018-01-25 | 2018-09-18 | 北京奇艺世纪科技有限公司 | A kind of server-side gray scale down method and device |
| CN108628885A (en) * | 2017-03-20 | 2018-10-09 | 腾讯科技（深圳）有限公司 | A kind of method of data synchronization, device and storage device |
| CN108647237A (en) * | 2018-03-30 | 2018-10-12 | 厦门海迈科技股份有限公司 | Method of data synchronization between isomeric relationship type database |
| CN108933824A (en) * | 2018-06-28 | 2018-12-04 | 郑州云海信息技术有限公司 | A kind of method, system and relevant apparatus keeping RabbitMQ service |
| CN109542330A (en) * | 2017-09-21 | 2019-03-29 | 杭州海康威视系统技术有限公司 | Date storage method, data query method and device |
| CN109933447A (en) * | 2019-03-29 | 2019-06-25 | 杭州弯流科技有限公司 | A kind of asynchronous message Queue Algorithm based on data line lock |
| CN111241049A (en) * | 2020-01-06 | 2020-06-05 | 深圳市同洲电子股份有限公司 | Distributed operation log implementation system based on micro-service architecture |
| CN111338834A (en) * | 2020-02-21 | 2020-06-26 | 北京百度网讯科技有限公司 | Data storage method and device |
| WO2022048495A1 (en) * | 2020-09-01 | 2022-03-10 | 阿里云计算有限公司 | Data backup method and apparatus, data recovery method and apparatus, and electronic device |
| CN114745438A (en) * | 2022-03-22 | 2022-07-12 | 北京乐我无限科技有限责任公司 | Cache data processing method, device and system for multiple data centers and electronic equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183377A (en) * | 2007-12-10 | 2008-05-21 | 华中科技大学 | High availability data-base cluster based on message middleware |
| CN102156720A (en) * | 2011-03-28 | 2011-08-17 | 中国人民解放军国防科学技术大学 | Method, device and system for restoring data |
| CN102314375A (en) * | 2011-03-18 | 2012-01-11 | 北京神州数码思特奇信息技术股份有限公司 | Heterogeneous database storage unified interface and database access method |
| CN102567334A (en) * | 2010-12-15 | 2012-07-11 | 上海杉达学院 | Office automation system based on heterogeneous data |
| CN103078927A (en) * | 2012-12-28 | 2013-05-01 | 合一网络技术(北京)有限公司 | Key-value data distributed caching system and method thereof |
| CN104391930A (en) * | 2014-11-21 | 2015-03-04 | 用友软件股份有限公司 | Distributed file storage device and method |
-
2016
- 2016-09-29 CN CN201610872590.6A patent/CN106357452B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101183377A (en) * | 2007-12-10 | 2008-05-21 | 华中科技大学 | High availability data-base cluster based on message middleware |
| CN102567334A (en) * | 2010-12-15 | 2012-07-11 | 上海杉达学院 | Office automation system based on heterogeneous data |
| CN102314375A (en) * | 2011-03-18 | 2012-01-11 | 北京神州数码思特奇信息技术股份有限公司 | Heterogeneous database storage unified interface and database access method |
| CN102156720A (en) * | 2011-03-28 | 2011-08-17 | 中国人民解放军国防科学技术大学 | Method, device and system for restoring data |
| CN103078927A (en) * | 2012-12-28 | 2013-05-01 | 合一网络技术(北京)有限公司 | Key-value data distributed caching system and method thereof |
| CN104391930A (en) * | 2014-11-21 | 2015-03-04 | 用友软件股份有限公司 | Distributed file storage device and method |
Non-Patent Citations (1)
| Title |
|---|
| 罗荣凌: "《基于异构数据库的容灾模型及数据一致性维护技术的研究与实现》", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107612950A (en) * | 2016-07-11 | 2018-01-19 | 阿里巴巴集团控股有限公司 | A kind of method, apparatus, system, electronic equipment that service is provided |
| CN108628885A (en) * | 2017-03-20 | 2018-10-09 | 腾讯科技（深圳）有限公司 | A kind of method of data synchronization, device and storage device |
| CN109542330A (en) * | 2017-09-21 | 2019-03-29 | 杭州海康威视系统技术有限公司 | Date storage method, data query method and device |
| CN107832138A (en) * | 2017-09-21 | 2018-03-23 | 南京邮电大学 | A kind of implementation method of the High Availabitity namenode models of flattening |
| CN109542330B (en) * | 2017-09-21 | 2020-11-10 | 杭州海康威视系统技术有限公司 | Data storage method, data query method and device |
| CN108255929A (en) * | 2017-11-30 | 2018-07-06 | 杭州博世数据网络有限公司 | For the processing method of database neo4j |
| CN108549584B (en) * | 2018-01-25 | 2020-11-27 | 北京奇艺世纪科技有限公司 | Method and device for degrading gray level of server side |
| CN108549584A (en) * | 2018-01-25 | 2018-09-18 | 北京奇艺世纪科技有限公司 | A kind of server-side gray scale down method and device |
| CN108647237A (en) * | 2018-03-30 | 2018-10-12 | 厦门海迈科技股份有限公司 | Method of data synchronization between isomeric relationship type database |
| CN108933824A (en) * | 2018-06-28 | 2018-12-04 | 郑州云海信息技术有限公司 | A kind of method, system and relevant apparatus keeping RabbitMQ service |
| CN109933447A (en) * | 2019-03-29 | 2019-06-25 | 杭州弯流科技有限公司 | A kind of asynchronous message Queue Algorithm based on data line lock |
| CN111241049A (en) * | 2020-01-06 | 2020-06-05 | 深圳市同洲电子股份有限公司 | Distributed operation log implementation system based on micro-service architecture |
| CN111338834A (en) * | 2020-02-21 | 2020-06-26 | 北京百度网讯科技有限公司 | Data storage method and device |
| CN111338834B (en) * | 2020-02-21 | 2023-06-30 | 北京百度网讯科技有限公司 | Data storage method and device |
| WO2022048495A1 (en) * | 2020-09-01 | 2022-03-10 | 阿里云计算有限公司 | Data backup method and apparatus, data recovery method and apparatus, and electronic device |
| CN114745438A (en) * | 2022-03-22 | 2022-07-12 | 北京乐我无限科技有限责任公司 | Cache data processing method, device and system for multiple data centers and electronic equipment |
| CN114745438B (en) * | 2022-03-22 | 2024-02-09 | 北京乐我无限科技有限责任公司 | Cache data processing method, device and system of multi-data center and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106357452B (en) | 2019-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106357452A (en) | High-availability framework system with single-point heterogeneous data storage function and realizing method thereof | |
| CN112118565B (en) | Multi-tenant service gray level publishing method, device, computer equipment and storage medium | |
| CN109947773B (en) | Deploying changes to key patterns in a multi-tenant database system | |
| US11689638B2 (en) | Embedded database as a microservice for distributed container cloud platform | |
| CN109491776B (en) | Task arranging method and system | |
| US5586312A (en) | Method and apparatus for using an independent transaction processing application as a service routine | |
| US11249983B2 (en) | Transaction change data forwarding | |
| CN113396407A (en) | System and method for augmenting database applications using blockchain techniques | |
| CN109522029A (en) | A kind of method and device for disposing cloud platform technology component | |
| US10467070B2 (en) | Processing cloud services and intelligence cloud services integration | |
| US11263236B2 (en) | Real-time cross-system database replication for hybrid-cloud elastic scaling and high-performance data virtualization | |
| US11681683B2 (en) | Transaction compensation for single phase resources | |
| US20130304705A1 (en) | Mirror file system | |
| WO2002091178A2 (en) | Method and apparatus for upgrading managed application state for a java based application | |
| US10698921B2 (en) | Persistence and initialization of synchronization state for serialized data log replay in database systems | |
| US20070174315A1 (en) | Compressing state in database replication | |
| US20190317840A1 (en) | Integrating transaction processing system interfaces with event-driven polyglot runtime modules | |
| US8527995B2 (en) | Synchronization system for entities maintained by multiple applications | |
| CN101727496A (en) | Method for realizing load balancing cluster of MICROSOFT SQL SERVER database | |
| CN108762982A (en) | A kind of database restoring method, apparatus and system | |
| US7752225B2 (en) | Replication and mapping mechanism for recreating memory durations | |
| CN111414356A (en) | Data storage method and device, non-relational database system and storage medium | |
| US20180253481A1 (en) | Testing database replication with multi-container system architecture | |
| CN114205333B (en) | IP configuration method, cluster construction method, computer device and storage medium | |
| Olmsted et al. | The cost of increased transactional correctness and durability in distributed databases |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wang Jialin Inventor after: Yang Zige Inventor after: Bao Haiwei Inventor before: Wang Jialin Inventor before: Yang Zige |