EP1370965A1 - Communication de service a service pour des services de reseau - Google Patents

Communication de service a service pour des services de reseau

Info

Publication number
EP1370965A1
EP1370965A1 EP02719261A EP02719261A EP1370965A1 EP 1370965 A1 EP1370965 A1 EP 1370965A1 EP 02719261 A EP02719261 A EP 02719261A EP 02719261 A EP02719261 A EP 02719261A EP 1370965 A1 EP1370965 A1 EP 1370965A1
Authority
EP
European Patent Office
Prior art keywords
service
publisher
subscriber
data
message
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.)
Withdrawn
Application number
EP02719261A
Other languages
German (de)
English (en)
Other versions
EP1370965A4 (fr
Inventor
Steven D. White
Lijiang Fang
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Microsoft Technology Licensing LLC
Original Assignee
Microsoft Corp
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority claimed from US10/017,680 external-priority patent/US7302634B2/en
Priority claimed from US10/033,177 external-priority patent/US20030061365A1/en
Application filed by Microsoft Corp filed Critical Microsoft Corp
Publication of EP1370965A1 publication Critical patent/EP1370965A1/fr
Publication of EP1370965A4 publication Critical patent/EP1370965A4/fr
Withdrawn legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L63/00—Network architectures or network communication protocols for network security
    • H04L63/10—Network architectures or network communication protocols for network security for controlling access to devices or network resources
    • H04L63/102—Entity profiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
    • G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management
    • G06Q10/109—Time management, e.g. calendars, reminders, meetings or time accounting
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00—Data switching networks
    • H04L12/02—Details
    • H04L12/16—Arrangements for providing special services to substations
    • H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
    • H04L12/1859—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast adapted to provide push services, e.g. data channels
    • 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
    • H04L67/51—Discovery or management thereof, e.g. service location protocol [SLP] or web services
    • 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
    • H04L67/56—Provisioning of proxy services
    • 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
    • H04L67/56—Provisioning of proxy services
    • H04L67/564—Enhancement of application control based on intercepted application data
    • 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
    • H04L69/30—Definitions, standards or architectural aspects of layered protocol stacks
    • H04L69/32—Architecture of open systems interconnection [OSI] 7-layer type protocol stacks, e.g. the interfaces between the data link level and the physical level
    • H04L69/322—Intralayer communication protocols among peer entities or protocol data unit [PDU] definitions
    • H04L69/329—Intralayer communication protocols among peer entities or protocol data unit [PDU] definitions in the application layer [OSI layer 7]
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00—Data switching networks
    • H04L12/02—Details
    • H04L12/16—Arrangements for providing special services to substations
    • H04L12/18—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast
    • H04L12/1863—Arrangements for providing special services to substations for broadcast or conference, e.g. multicast comprising mechanisms for improved reliability, e.g. status reports
    • H04L12/1877—Measures taken prior to transmission
    • 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
    • H04L67/02—Protocols based on web technology, e.g. hypertext transfer protocol [HTTP]

Definitions

  • the invention relates generally to computer network services for user data access, and more particularly to systems, methods and data structures for communication between the services.
  • corporate networks and the like can provide users with remote access to some of their data, but many users do not have access to such a network. For many of those that have access, connecting to a network with the many different types of devices, assuming such devices can even connect to a network, can be a complex or overwhelming problem. Moreover, even if a user has centrally stored data, the user needs the correct type of device running the appropriate application program to access that data. For example, a user with a PDA that runs a simple note taking application program ordinarily will not be able to use that program to open documents stored by a full-blown word processing program at work. In general, this is because the data is formatted and accessed according to the way the application program wants it to be formatted.
  • What is needed is a model wherein data is centrally stored for users, with a set of services that control access to the data with defined methods, regardless ofthe application program and/or device.
  • the data for each service should be structured in a defined way that complies with defined rules for that data, regardless ofthe application program or device that is accessing the data.
  • the data should be controllable by a user so as to automatically adjust for changes made thereto by other users.
  • This model should scale and interrelate the data of millions of users in virtually any combination, in a highly efficient robust manner.
  • the present invention provides a set of services for central (e.g., Internet) access to per-user data, based on each user's identity, with a service-to-service communications protocol that handles change information for millions of users.
  • the protocol enables the automatic publication and subscription by services of changes made to data.
  • the protocol is role-based in that a user controls the users that can subscribe for the user's data changes.
  • the protocol is efficient in that data is change data for users are combined and batched, and robust to handle failure scenarios.
  • the a "publisher” refers to the .NET MyServices service which is the source ofthe data
  • a "subscriber” refers to the .NET MyServices service that receives the data.
  • SSCP is a generic way for a .NET MyServices service to publish data changes to another .NET MyServices service.
  • the present invention establishes common message formats, and an accepted set of primitives that the parties involved understand, so that transactions among them follow predictable logical sequences.
  • SSCP also establishes handshaking procedures with ACK to handle lost messages.
  • the publisher maintains information about the identifier (ID) ofthe subscribing users.
  • ID identifier
  • the publisher maintains the ID ofthe user's data for which they have subscribed.
  • the publisher also maintains information regarding the role ofthe subscribing user. In order for the publisher to keep this information current, the subscriber notifies the publisher whenever one of its users wants to unsubscribe or add a new subscription.
  • SSCP provides for transmission of subscription updates from subscriber to publisher using the same robust mechanism as are used for transmitting data changes.
  • each request from a sender should have a response from the receiver. If the message fails to reach the receiver, e.g. due to transient network and/or service failure, it is resent during the next update interval. This resend process is repeated until a response is received, with a specified number of such retries performed, after which no further attempts are made for an appropriately longer time. More subtle types of failures also need to be handled. For example, consider a publisher sending a request to the subscriber, informing it ofthe change in a stored profile. The subscriber ordinarily receives and processes the request, and sends a response to the publisher.
  • the publisher will re-send its request it request during the next update interval.
  • SSCP the subscriber recognizes that this is a redundant request, and that it has already been processed, whereby the subscriber acknowledges the request again, but does not process it.
  • protocol handler For efficiency, because a typical service manages enormous amounts of data, partitioned over millions of users and the source data will be almost constantly changing, the protocol batches multiple requests and send them periodically. To this end, a protocol handler at the service periodically wakes up after a specified interval and sends the batched messages. Moreover, if a catastrophic failure (such as loss of power) occurs, this state data regarding the messages to send should not be lost, so data pertaining to protocol state should be stored in a durable manner, e.g., persisted to a hard disk. To implement SSCP, protocol handlers the publisher and subscriber track of several pieces of information, such as in respective tables.
  • a publications table is used by the publisher to track the users across the services that have subscriptions with it.
  • the publisher includes a publications queue table that is used by the publisher for batching requests until the protocol handler sends the requests at an update interval.
  • the publisher also retries requests for which a response has not been received, and thus tracks messages that need to be sent for the first time, or need to be resent, in the publications queue table.
  • the subscriber service includes subscriptions table to track of its subscriptions that are in effect.
  • the subscribing user specifies the user's identity ofthe user whose data he or she wants to subscribe to.
  • a subscriptions queue table is used by the subscriber to batch its requests for sending by the protocol handler at the update interval. Also, the subscriber is required to retry requests for which a response has not been received, and thus keeps track of messages that need to be sent for the first time, or need to be resent, which is also done in the subscriptions queue table.
  • the amount of information that is transmitted from one service to another is significantly reduced in SSCP because the change information for one user at a publisher service that is subscribed to by multiple users at a subscriber service who are assigned the same role at the publishing service, are aggregated into a single message.
  • the publisher operates in a fan-in model to put change information together based on their roles, rather than separate it per user recipient, and leaves it up to the subscriber to fan the information out to the appropriate users.
  • FIGURE 1 is a block diagram representing an exemplary computer system into which the present invention may be incorporated;
  • FIG. 2 is a block diagram representing a generic data access model
  • FIG. 3 is a representation of services for identity-based data access
  • FIG. 4 is a block diagram representing a schema-based service for accessing data arranged in a logical content document based on a defined schema for that service;
  • FIGS. 5-7 are block diagram generally representing publishers and subscribers interconnected via a service-to-service communication protocol in accordance with one aspect of the present invention
  • FIGS. 8-16B comprise flow diagrams generally representing operation ofthe service-to-service communication protocol in accordance with one aspect ofthe present invention.
  • FIGS. 17-18 are block diagram generally representing publishers and subscribers interconnected via a service-to-service communication protocol in accordance with an alternative aspect of the present invention.
  • FIGS. 19-20 are block diagram generally representing models in which the service-to-service communication protocol may be implemented, in accordance with an aspect ofthe present invention.
  • FIGURE 1 illustrates an example of a suitable computing system environment 100 on which the invention may be implemented.
  • the computing system environment 100 is only one example of a suitable computing environment and is not intended to suggest any limitation as to the scope of use or functionality ofthe invention. Neither should the computing environment 100 be interpreted as having any dependency or requirement relating to any one or combination of components illustrated in the exemplary operating environment 100.
  • the invention is operational with numerous other general purpose or special purpose computing system environments or configurations.
  • Examples of well known computing systems, environments, and/or configurations that may be suitable for use with the invention include, but are not limited to: personal computers, server computers, handheld or laptop devices, tablet devices, multiprocessor systems, microprocessor-based systems, set top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments that include any ofthe above systems or devices, and the like.
  • the invention may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer.
  • program modules include routines, programs, objects, components, data structures, and so forth, that perform particular tasks or implement particular abstract data types.
  • the invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network.
  • program modules may be located in local and/or remote computer storage media including memory storage devices.
  • an exemplary system for implementing the invention includes a general purpose computing device in the form of a computer 110.
  • Components ofthe computer 110 may include, but are not limited to, a processing unit 120, a system memory 130, and a system bus 121 that couples various system components including the system memory to the processing unit 120.
  • the system bus 121 may be any of several types of bus structures including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of bus architectures.
  • such architectures include Industry Standard Architecture (ISA) bus, Micro Channel Architecture (MCA) bus, Enhanced ISA (EISA) bus, Video Electronics Standards Association (VESA) local bus, and Peripheral Component Interconnect (PCI) bus also known as Mezzanine bus.
  • ISA Industry Standard Architecture
  • MCA Micro Channel Architecture
  • EISA Enhanced ISA
  • VESA Video Electronics Standards Association
  • PCI Peripheral Component Interconnect
  • the computer 110 typically includes a variety of computer-readable media.
  • Computer-readable media can be any available media that can be accessed by the computer 110 and includes both volatile and nonvolatile media, and removable and nonremovable media.
  • Computer-readable media may comprise computer storage media and communication media.
  • Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data.
  • Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by the computer 110.
  • Communication media typically embodies computer-readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media.
  • modulated data signal means a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal.
  • communication media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared and other wireless media. Combinations ofthe any ofthe above should also be included within the scope of computer-readable media.
  • the system memory 130 includes computer storage media in the form of volatile and/or nonvolatile memory such as read only memory (ROM) 131 and random access memory (RAM) 132.
  • ROM read only memory
  • RAM random access memory
  • a basic input/output system 133 (BIOS) containing the basic routines that help to transfer information between elements within computer 110, such as during start-up, is typically stored in ROM 131.
  • RAM 132 typically contains data and/or program modules that are immediately accessible to and/or presently being operated on by processing unit 120.
  • FIG. 1 illustrates operating system 134, application programs 135, other program modules 136 and program data 137.
  • the computer 110 may also include other removable/non-removable, volatile/nonvolatile computer storage media.
  • FIG. 1 illustrates a hard disk drive 141 that reads from or writes to non-removable, nonvolatile magnetic media, a magnetic disk drive 151 that reads from or writes to a removable, nonvolatile magnetic disk 152, and an optical disk drive 155 that reads from or writes to a removable, nonvolatile optical disk 156 such as a CD ROM or other optical media.
  • removable/non-removable, volatile/nonvolatile computer storage media that can be used in the exemplary operating environment include, but are not limited to, magnetic tape cassettes, flash memory cards, digital versatile disks, digital video tape, solid state RAM, solid state ROM, and the like.
  • the hard disk drive 141 is typically connected to the system bus 121 through a non-removable memory interface such as interface 140, and magnetic disk drive 151 and optical disk drive 155 are typically connected to the system bus 121 by a removable memory interface, such as interface 150.
  • hard disk drive 141 is illustrated as storing operating system 144, application programs 145, other program modules 146 and program data 147. Note that these components can either be the same as or different from operating system 134, application programs 135, other program modules 136, and program data 137. Operating system 144, application programs 145, other program modules 146, and program data 147 are given different numbers herein to illustrate that, at a minimum, they are different copies.
  • a user may enter commands and information into the computer 20 through input devices such as a tablet, or electronic digitizer, 164, a microphone 163, a keyboard 162 and pointing device 161, commonly referred to as mouse, trackball or touch pad.
  • Other input devices not shown in FIG. 1 may include a joystick, game pad, satellite dish, scanner, or the like.
  • a user input interface 160 that is coupled to the system bus, but may be connected by other interface and bus structures, such as a parallel port, game port or a universal serial bus (USB).
  • a monitor 191 or other type of display device is also connected to the system bus 121 via an interface, such as a video interface 190.
  • the monitor 191 may also be integrated with a touch-screen panel or the like. Note that the monitor and/or touch screen panel can be physically coupled to a housing in which the computing device 110 is incorporated, such as in a tablet-type personal computer. In addition, computers such as the computing device 110 may also include other peripheral output devices such as speakers 195 and printer 196, which may be connected through an output peripheral interface 194 or the like.
  • the computer 110 may operate in a networked environment using logical connections to one or more remote computers, such as a remote computer 180.
  • the remote computer 180 may be a personal computer, a server, a router, a network PC, a peer device or other common network node, and typically includes many or all of the elements described above relative to the computer 110, although only a memory storage device 181 has been illustrated in FIG. 1.
  • the logical connections depicted in FIG. 1 include a local area network (LAN) 171 and a wide area network (WAN) 173, but may also include other networks.
  • LAN local area network
  • WAN wide area network
  • Such networking environments are commonplace in offices, enterprise-wide computer networks, intranets and the Internet.
  • the computer system 110 may comprise source machine from which data is being migrated, and the remote computer 180 may comprise the destination machine.
  • source and destination machines need not be connected by a network or any other means, but instead, data may be migrated via any media capable of being written by the source platform and read by the destination platform or platforms.
  • the computer 110 When used in a LAN networking environment, the computer 110 is connected to the LAN 171 through a network interface or adapter 170.
  • the computer 110 When used in a WAN networking environment, the computer 110 typically includes a modem 172 or other means for establishing communications over the WAN 173, such as the Internet.
  • the modem 172 which may be internal or external, may be connected to the system bus 121 via the user input interface 160 or other appropriate mechanism.
  • FIG. 1 illustrates remote application programs 185 as residing on memory device 181. It will be appreciated that the network connections shown are exemplary and other means of establishing a communications link between the computers may be used.
  • the present invention generally operates in an architecture / platform that connects network-based (e.g., Internet-based) applications, devices and services, and transforms them into a user's personal network which works on the user's behalf, and with permissions granted by the user.
  • network-based e.g., Internet-based
  • the present invention is generally directed to schema-based services that maintain user, group, corporate or other entity data in a commonly accessible virtual location, such as the Internet.
  • the present invention is intended to scale to millions of users, and be stored reliably, and thus it is likely that a user's data will be distributed among and/or replicated to numerous storage devices, such as controlled via a server federation.
  • a data access model 200 of Figure 2 illustrates a generic navigation module 202 through which applications 204 and the like may access a wide variety of identity-based data, such as maintained in an addressable store 206.
  • a common set of command methods may be used to perform operations on various data structures that are constructed from the data in the addressable store 206, even though each of those data structures may represent different data and be organized quite differently.
  • Such command methods may describe generic operations that may be desired on a wide variety of data structures. Such operations may include, for example, insert, delete, replace, update, query or changequery operations.
  • the data is arranged according to various schemas, with the schemas corresponding to identity-based services through which users access their data.
  • a "schema” generally comprises a set of rules that define how a data structure may be organized, e.g., what elements are supported, in what order they appear, how many times they appear, and so on.
  • a schema may define, via color-coding or other identification mechanisms, what portions of an XML document (that corresponds to the data structure) may be operated on. Examples of such XML documents are described below.
  • the schema may also define how the structure ofthe XML document may be extended to include elements not expressly mentioned in the schema.
  • the schemas vary depending on the type of data they are intended to organize, e.g., an email-inbox-related schema organizes data differently from a schema that organizes a user's favorite websites.
  • the services that employ schemas may vary.
  • the generic navigation module 202 has associated therewith a navigation assistance module 208 that includes or is otherwise associated with one or more schemas 210.
  • a navigation assistance module 208 as represented in FIG. 2 corresponds to one or more services, and possesses the information that defines how to navigate through the various data structures, and may also indicate what command methods may be executed on what portions ofthe data structure.
  • FIG. 2 only one navigation assistance module 208 is shown coupled to the generic navigation module 202, there may be multiple navigation assistance modules that may each specialize as desired. For example, each navigation assistance module may correspond to one service.
  • the navigation assistance module 208 is illustrated as a separate module, some or all ofthe operations ofthe navigation assistance module 208 may be incorporated into the generic navigation module 202, and vice versa.
  • the various data structures constructed from the schema and addressable store data may comprise XML documents of various XML classes.
  • the navigation assistance module 208 may contain a schema associated with each of the classes of XML documents.
  • a number of schema-based services facilitate data access based on the identity of a user.
  • the user need not obtain a separate identity for each service, but rather obtains a single identity via a single set of credentials, such as with the Microsoft® Passport online service.
  • a user can access data via these services from virtually any network connectable device capable of running an application that can call the methods of a service.
  • .NET My Services comprises identity-centric services which may be generally implemented in XML (extensible Markup Language) Message Interfaces (XMIs). While the present invention will be described with respect to XML and XMI, it can readily be appreciated that the present invention is not limited to any particular language or set of interfaces.
  • the .NET My Services model essentially corresponds to one implementation ofthe generic data access model 200 of FIG. 2.
  • .NET My Services 300 is implemented as a set of Web services 301-316, each bound to a .NET Identity (PUID, such as a Passport ® unique identifier similar to a globally unique identifier when Passport ® is the authentication service).
  • the services 301-316 can communicate with one another via a service-to-service communications protocol (SSCP), described below with respect to FIG. 5-23.
  • SSCP service-to-service communications protocol
  • each service presents itself as a set of XML documents that can be manipulated from an application program 202 (FIG. 2) or the like using a set of standard methods and domain-specific methods.
  • a user device 320 running such application programs connects a user's applications to the services, and the data controlled by those services, such as over the Internet or an Intranet, such as over the Internet or an Intranet.
  • endpoints can be client devices, applications or services.
  • any device capable of executing software and connecting to a network in any means may thus give a user access to data that the user is allowed to access, such as the user's own data, or data that a friend or colleague has specified as being accessible to that particular user.
  • a .NET Identity is an identifier assigned to an individual, a group of individuals, or some form of organization or project.
  • each identity e.g., of a user, group or organization
  • each identity has tied to it a set of services that are partitioned along schema boundaries and across different identities.
  • the XML-document-centric architecture of .NET My Services provides a model for manipulating and communicating service state that is very different from prior data access models.
  • the XML-document-centric approach in conjunction with loose binding to the data exposed by the services, enables new classes of application programs.
  • the .NET My Services model 300 presents the various services 301-316 using a uniform and consistent service and method model, a uniform and consistent data access and manipulation model, and a uniform and consistent security authorization model.
  • the .NET My Services model 300 is based upon open Internet standards. Services are accessed by means of SOAP (Simple Object Access Protocol) messages containing an XML payload. Service input and output is expressed as XML document outlines, and each of these document outlines conform to an XML schema document. The content is available to a user from the interaction with the .NET My Services service endpoint 320.
  • SOAP Simple Object Access Protocol
  • a schema essentially describes a web service, such as in XML. More particularly, a service author begins to write a web service by defining an XML schema that defines what the data model looks like, e.g., the supported elements, their relative ordering, how many times they appear, and other similar definitions, as will become apparent below. This service definition also applies to an author determining what roles and methods are supported, e.g., which operations are supported, and the extent ofthe data that can be returned for each method. Another way of stating this concept is that the author starts by building a complete definition of a service, such as in XML, and specifies the verbs (methods) that an application will use to talk to it.
  • the service author has an XML definition that has been declared, and this declarative definition may be run through a compilation process, resulting in a fully operational service.
  • a general purpose interpreter-like mechanism already exists that, when fed one of these declarative XML definitions, adapts to the declarative XML definitions, thereby knowing what to do and how to act. From that point on a service exists that is capable of operating. In a simple service (e.g., with no domain-specific methods or complex logic), no new code needs to be written to provide such an operational service. As will be understood, such authoring of a service without coding is possible due to the data driven model ofthe present architecture.
  • an application 400 requests performance of a method that operates on data structures, in a manner that is generic with respect to the type of data structure being operated upon and without requiring dedicated executable code for manipulating data structures of any particular data type.
  • the application first contacts a special myServices service 314 to obtain the information needed to communicate with a particular service 404, through a set of methods 406 of that service 404.
  • the needed information received from the myServices service 314 includes a URI of that service 404.
  • the service 404 may correspond to essentially any ofthe services represented in FIG. 3.
  • the service 404 includes or is otherwise associated with a set of methods 406 including standard methods 408, such as to handle requests directed to insert, delete, replace, update, query or changequery operations on the data.
  • the set of methods of a particular service may also include service specific methods 410. In general, the only way in which an application can communicate with a service are via that service's methods.
  • Each service includes service logic 412 for handling requests and providing suitable responses.
  • the service logic performs various functions such as authorization, authentication, and signature validation, and further limits valid users to data which they are permitted to access.
  • the security aspect of a service is not discussed herein, except to note that in general, for otherwise valid users, the user's identity determines whether a user can access data in a requested manner.
  • a roleMap 414 comprising service-wide roleList document templates 415 and scopes (e.g., part of the overall service's schema 416) in conjunction with user-based data maintained in an addressable store 418 determines whether a particular requested method is allowed, e.g., by forming an identity-based roleList document 420.
  • the scope information in the roleMap 414 determines a shape of data to return, e.g., how much content is allowed to be accessed for this particular user for this particular request.
  • the content is obtained in accordance with a content document 422 in the service's schema 416 and the actual user data corresponding to that content document in the addressable store 418.
  • a per-identity shaped content document 424 may be essentially constructed for returning to the user, or for updating the addressable store, as appropriate for the method.
  • FIG. 4 includes a number of ID-based roleList documents and ID-based content documents, to emphasize that the service 406 is arranged to serve multiple users.
  • a system document 426 is present as part ofthe schema 416, as described below.
  • access to .NET My Services 300 is accomplished using SOAP messages formatted with .NET My Services-specific header and body content.
  • Each ofthe .NET My Services will accept these messages by means of an HTTP POST operation, and generate a response by "piggy-backing" on the HTTP Response, or by issuing an HTTP POST to a .NET My Services response-processing endpoint 320.
  • HTTP HyperText Transfer Protocol
  • .NET My Services will support raw SOAP over TCP, a transfer protocol known as Direct Internet Message Encapsulation (or DIME). Other protocols for transferring messages are feasible.
  • DIME Direct Internet Message Encapsulation
  • .NET My Services are accessed by protocol, no particular client-side binding code, object models, API layers, or equivalents are required, and are thus optional.
  • the .NET My Services will support Web Services Description Language (WSDL). It is not mandatory that applications wishing to interact with .NET My Services make use of any particular bindings, and such bindings are not described herein. Instead, the present invention will be generally described in terms of messages that flow between requestors of a particular service and the service endpoints.
  • a service In order to interact with .NET My Services, a service needs to format a .NET My Services message and deliver that message to a .NET My Services endpoint.
  • a client In order to format a message, a client needs to manipulate XML document outlines, and typically perform some simple, known (public-domain) cryptographic operations on portions ofthe message.
  • each .NET My Services service presents three logical XML documents, a content document 422, roleList document 415 (ofthe roleMap 414), and a system document 426. These documents are addressable using .NET My Services message headers, and are manipulated using standard .NET My Services methods.
  • each service may include additional domain-specific methods. For example, as described below, the ".NET Calendar” service 303 might choose to expose a "getFreeBusy” method rather than expose free/busy as writeable fragments in the content document.
  • Each .NET MyServices service thus logically includes a content document 422, which in general is the main, service-specific document.
  • the schema for this document 422 is a function of the class of service, as will become apparent from the description of each service's schema below.
  • the content document presents data in the shape dictated by the .NET My Services .NET Calendar schema
  • the ".NET FavoriteWebSites" service 308 the content document presents data in the shape dictated by the .NET My Services .NET FavoriteWebSites schema.
  • Each service also includes a roleList document 415 that contains roleList information, comprising information that governs access to the data and methods exported by the service 404.
  • the roleList document is manipulated using the .NET My Services standard data manipulation mechanisms. The shape of this document is governed by the .NET My Services core schema's roleListType XML data type.
  • Each service also includes a system document 426, which contains service-specific system data such as the roleMap, schemaMap, messageMap, version information, and service specific global data.
  • the document is manipulated using the standard .NET My Services data manipulation mechanism, although modifications are limited in a way that allows only the service itself to modify the document.
  • the shape of this system document 426 may be governed by the system document schema for the particular service, in that each service may extend a base system document type with service specific information.
  • the base system document is described once, rather than for each service, with only those services having extended service specific information separately described. Notwithstanding, it should be understood that each service includes at least the base system portion in its system document.
  • the present invention is generally directed to schemas, which in general comprise a set of rules or standards that define how a particular type of data can be structured.
  • schemas which in general comprise a set of rules or standards that define how a particular type of data can be structured.
  • the meaning of data may be communicated between computer systems.
  • a computer device may recognize that a data structure that follows a particular address schema represents an address, enabling the computer to "understand" the component part of an address.
  • the computer device may then perform intelligent actions based on the understanding that the data structure represents an address. Such actions may include, for example, the presentation of an action menu to the user that represents things to do with addresses.
  • Schemas may be stored locally on a device and/or globally in the federation's "mega- store.”
  • a device can keep a locally stored schema updated by subscribing to an event notification service (in this case, a schema update service) that automatically passes messages to the device when the schema is updated. Access to globally stored schemas is controlled by the security infrastructure.
  • the various .NET MyServices services described above are loosely coupled services, and have the ability to share data with each other. It is thus possible for the data to be stored and managed by one service, regardless of how many services or applications make use ofthe data.
  • this data sharing can take place (assuming that appropriate security constraints are satisfied), a first of which is that one service that wants data queries another service that has the data, i.e., a pull model.
  • a service that wants data can informs the service that has the data to send it the current copy of the data and places an outstanding request to send it any changes to that data. The said changes are sent asynchronously. This is a push model.
  • the .NET services defines verbs such as query, update, etc., which can be used as a basis for the pull data pipe between services. But for reasons of bandwidth optimization and robustness, the push model turns out to be a better choice for service to service communication.
  • SSCP service-to-service communication protocol
  • a "publisher” refers to the .NET MyServices service which is the source ofthe data
  • a “subscriber” refers to the .NET MyServices service that receives the data.
  • SSCP is a generic way for a .NET MyServices service to publish data changes to another .NET MyServices service.
  • SSCP does not make any assumptions on what data is being published, and the data may be from any source, e.g., .NET Contacts, .NET Profile, .NET Presence, .NET Inbox and so forth.
  • SSCP also does not make any assumptions on which services can be publishers and which services can be subscribers.
  • the same service can be a publisher and subscriber, publishers can publish to multiple subscribers and subscribers can subscribe to multiple publishers.
  • a given service can publish/subscribe to a static list of other services, e.g., .NET Contacts (alternatively referred to as myContacts) may be configured with the list of services (e.g., .NET Profile / myProfile, .NET Inbox / mylnbox and so on) that it wants to publish to and/or subscribe from, and this list will ordinarily not change.
  • myContacts e.g.NET Contacts
  • myContacts may be configured with the list of services (e.g.NET Profile / myProfile, .NET Inbox / mylnbox and so on) that it wants to publish to and/or subscribe from, and this list will ordinarily not change.
  • the services are static, the instances of services are not. For example, once a service A is configured with the ability to publish or subscribe from service B, service A can do so with any instance of B For security reasons and the like, only .NET services can participate in data communication over SSCP.
  • an email whitelist which is a list of addresses that are allowed to send email to a particular recipient. Email from people belonging to the whitelist is put in the inbox; all other email is sent to a Junk Mail folder or to the deleted folder. Sometimes, the whitelist of a user is the same as her contact list - this would be the case with the .NET Inbox service. Even if this is not the case, it is fairly straightforward to store a whitelist in .NET Contacts by the use of a categorization mechanism present in .NET My Services.
  • a white list can be implemented in a brute force fashion by arranging the .NET Inbox service (e.g., directly or in conjunction with an application program) to look at the sender address whenever a message is received.
  • the .NET Inbox service may query the user's .NET Contacts service to see if the sender is in the contact list, whereby depending on the result of the query, .NET Inbox service either puts the message in the Inbox or puts it in the Junk Mail folder.
  • this approach has obvious performance and scaling problems, as it is impractical or impossible for any service that handles hundreds of millions of email messages every day to use such a model; the sheer volume of traffic between .NET Inbox and .NET Contacts would bring down both the services.
  • a superior approach is for the .NET Inbox service to maintain a local copy ofthe whitelist, and subscribe to the .NET Contacts data of every user that has enabled a Junk Mail filter. Whenever changes occur to the whitelist, the .NET Contacts service uses SSCP to send those changes to the .NET Inbox service.
  • the .NET Inbox service has a local copy ofthe white list, the performance/scaling issues are avoided, and any traffic between the .NET Inbox service and the .NET Contacts service occurs only when the whitelist changes in the .NET Contacts service, the .NET Inbox service subscribes to the .NET Contacts service document of a new user (or a user who has newly activated her junk mail filters) or the .NET Inbox service asks the .NET Contacts service to delete the subscription of a currently subscribed user.
  • Whitelists represent a simple publish-subscribe scenario in that user id's of the publisher and subscriber are the same. There is no requirement to take into account the role ofthe subscriber in the publisher's document, the assumption being that the same id plays the "owner" role on both sides of this communication channel.
  • a more complex example is that of Live Contacts.
  • .NET Contacts service it is likely that many are users of .NET My Services. As a result, these contacts will have a .NET Profile service which manages data in their profile.
  • the data stored in a contact record of .NET Contacts is a subset of what is stored in that contact's profile, the boundaries of the said subset being determined by the role of the subscriber in the originating profile's role list.
  • .NET Contacts service to subscribe to the .NET Profile service to get the data for many ofthe contacts that it manages. From the other perspective, the .NET Profile service publishes its data to the .NET Contacts service.
  • .NET Profile of this user publishes any changes to the .NET Contacts service of each appropriately authorized user (e.g., in a trusted circle of friends), whenever a user updates his profile, such as to change his or her email address, that change becomes immediately visible to the users in his or her trusted circle when they look up his email via their .NET Contacts service.
  • SSCP works across realms as well as between services in the same realm, e.g., a subscriber contacts service in a realm corresponding to MSN.com will be able to receive published changes from a publisher profile service in a realm corresponding to a provider such as XYZ.com, as well as from an MSN.com profile service.
  • the present invention favors the push model over the pull model. While the pull model is usually simpler, its conceivable use is limited to data pipes with low traffic and/or few subscribers. However, the push model, while a little more involved, provides a bandwidth optimized, robust data pipe and is ideal for high-traffic and/or large number of subscribers. To ensure robustness in such an environment of transient network and/or service failures, the present invention establishes common message formats, and an accepted set of primitives that the parties involved understand, so that transactions among them follow predictable logical sequences. SSCP also establishes handshaking procedures with ACK to handle lost messages.
  • FIG. 5 provides a representation of an example publisher-subscriber relationship.
  • FIG. 5 there are two .NET Profile services 501 and 502, each managing the profiles of three users, 504-506 and 508-510, respectively.
  • each of these services 501, 502 and 520 will typically manage the data for hundreds of millions of users. Note that for each user, access to the various contact information sets is on a per-identifier basis, e.g., a contact that is specified as a friend by a user may be assigned different access rights to the user's contacts than a contact that is specified as an associate by the same user.
  • the .NET Contacts service 520 has subscriptions in two different .NET Profile services, namely 501 and 502. Similarly, it is likely that a given publisher will publish to multiple subscribers. Note that a single service may act both as a subscriber and a publisher, e.g., in the whitelist example above, the .NET Contacts service is a publisher, while in the Live Contacts example, .NET Contacts service is a subscriber.
  • .NET Profile When User ⁇ 's role in User ⁇ 's .NET Profile is that of an associate, while the role of User2 is that of a friend.
  • .NET Profile publishes the data, it sends data visible to an associate to Userl, and data visible to a friend to User2.
  • SSCP sends changes only to subscribed users within a subscribing service, and determines the role of each subscribing user and filters the data based on the role.
  • User3's role was also that of an associate, then only one copy ofthe associate data would be sent to the subscribing service, thus optimizing usage of network resources.
  • the publisher maintains information about the identifier (ID) ofthe subscribing users, (e.g., Userl, User2). Also, for each subscribing user, the publisher maintains the ID ofthe user's data for which they have subscribed, e.g., for Userl of .NET Contacts , this is User2 and User3 in .NET Profile service 1. The publisher also maintains information regarding the role ofthe subscribing user, e.g., in the context of User ⁇ in .NET Profile service2, this is associate for Userl, friend for User2).
  • the subscriber In order for the publisher to keep this information current, the subscriber notifies the publisher whenever one of its users wants to unsubscribe or add a new subscription. For example, consider that Userl wants to add User4 into his live contact list, and remove User ⁇ .
  • SSCP provides for transmission of subscription updates from subscriber to publisher using the same robust mechanism as are used for transmitting data changes.
  • the SSCP data pipe is robust and as such, is tolerant of transient network and/or service failures.
  • the publisher or subscriber needs to know that their transmitted messages have reached the destination, which means that each request from a sender should have a response from the receiver. If the message fails to reach the receiver, e.g.
  • a subscriber can process a repeat request any number of times, however the result of any subsequent processing should not change the first processing result. This property is referred to as idempotency.
  • a protocol handler at the service periodically wakes up after a specified interval and sends the batched messages.
  • this state data regarding the messages to send should not be lost, so data pertaining to protocol state should be stored in a durable manner, e.g., persisted to a hard disk.
  • SSCP is implemented at a publisher (service) 600 and subscriber (service) 610 by respective protocol handlers 602, 612, such as daemon processes or the like running with respect to a service.
  • the publisher 600 and subscriber 610 exchange messages, and use this as a mechanism to communicate changes.
  • SSCP handlers 602, 612 maintain several pieces of data, the sum total of which represents the state of a publisher or subscriber. As conceptually represented in FIG. 6, this data can be viewed as being segmented over several data structures 604-618. Note however that the arrangements, formats and other description presented herein are only logically represent the schema; the actual storage format is not prescribed, and an implementation may store in any fashion it deems fit as long as it logically conforms to this schema.
  • a publisher 600 communicates with a subscriber 610 using request and response messages. For example, when data changes at the publisher 600, the publisher 600, sends a request message to the subscriber 610 informing the subscriber that data has changed, normally along with the new data.
  • the subscriber 610 receives the message, makes the required updates, and sends back an ACK message acknowledging that the message was received and that the changes were made.
  • a subscriber 610 can also send a request message, such as when the subscriber 610 wants to subscribe or un-subscribe to a piece of datum.
  • the publisher 600 updates its list of subscriptions (in a publications table 608) and sends back a response acknowledging the request.
  • SSCP is agnostic to whether a response message for a given request is synchronous or asynchronous.
  • SSCP there are two primary parts to SSCP, a first from the publisher to the subscriber, which deals with sending changes made to the publisher's data, and a second from subscriber to the publisher, which deals with keeping the list of subscriptions synchronized. Furthermore, every service is required to provide notification to all other services that have subscriptions with it, or services with which it has subscriptions, when it is going offline or online.
  • Protocol parameters are supported by both the publisher and the subscriber and control the behavior ofthe protocol.
  • SSCP supports the ability to batch request messages. Whenever there is a need to send a request message, such as when there are changes in publisher data or subscriptions, the service puts the corresponding request message into a publisher message queue 606. Periodically, the protocol handler 602 in the publishing service 600 wakes up and processes the messages in the queue 606. This period is called as the Updatelnterval, and is a configurable parameter. To satisfy the robustness requirement, the publisher's protocol handler 602 needs to periodically resend requests until the publisher service 600 receives an acknowledge message (ACK). If the ACK for a message is successfully received, this message is purged from the queue 606.
  • ACK acknowledge message
  • the message remains in the queue, flagged as having been sent at least once, so it will be retried at the next update interval.
  • the number of times the publisher the publisher service 600 retries sending a message to the subscriber service 610 is configurable by the parameter RetryCount, i.e., after retrying this many times, the publisher service 600 assumes that the subscriber service 610 is dead. Then, once the maximum number of retries is over, the publisher service 600 waits for a relatively longer time. Once this longer time is elapsed, the publisher service 600 sets the RetryCount parameter to zero and begins resending the queued up requests over again. This longer time (before beginning the retry cycle), is configurable by the parameter Resetlnterval.
  • the protocol handlers 602, 610 at the publisher and subscriber, respectively track of several pieces of information, such as in their respective tables 604-618.
  • SSCP relies on the entities (services and users) being uniquely identifiable by the use of identifiers, e.g., every user in .NET has a unique identifier assigned by the Microsoft ® Passport service.
  • Each service be it acting as a publisher or subscriber, also has a unique identifier, and in practice, a service ID will be a certificate issued by a certification authority.
  • connections table e.g., the connections table 604 for the publishing service 600 and the connections table 614 for the subscribing service 610.
  • SID comprises the service ID of a Subscriber or Publisher
  • TO comprises the URL at which the service is expecting requests comprises
  • CLUSTER comprises the cluster number of this service
  • RETRY comprises the current retry number ofthe service.
  • RetryCount ⁇ RETRY ⁇ Resetlnterval the target service is assumed to be dead. Note that when an unknown service is recognized (i.e., one that is not present in the connections table), an attempt is made to contact immediately, without waiting until the next interval.
  • a publications table 608 is used by the publisher 600 to track the users across the services that have subscriptions with it.
  • the publications table 608 includes records with the following fields:
  • PUID comprises the identifier ofthe publishing user
  • SUID comprises the identifier ofthe subscribing user
  • SSID comprises the identifier ofthe subscribing service
  • ROLE comprises the role assign to this SUED
  • CN comprises the last known change number ofthe publisher's data which was delivered to the subscriber (for updating deltas).
  • the CN field is required to ensure recovery from certain catastrophic failures, as described below.
  • the publications table 608 may be made visible at the schema level, but ordinarily should be read-only.
  • the set SM is refe ⁇ ed to as the subscription map of P with respect to S.
  • the subscription map is obtained by the following query:
  • the publisher 600 includes a publications queue table 606 that is used by the publisher for batching requests until the protocol handler 602 sends the requests when the Updatelnterval time is achieved.
  • the publisher also retries requests for which a response has not been received, and thus tracks messages that need to be sent for the first time, or need to be resent, in the publications queue table 606.
  • An entry in the table 606 looks like this:
  • SUID comprises the identifier ofthe subscribing user
  • PUID comprises the identifier ofthe publishing user
  • SSID comprises the identifier ofthe subscribing service.
  • the publication queue 606 does not store messages, because a publisher services millions of users, whereby at any given instant, the publications queue 606 is likely have thousands of entries, and thus the amount of change data may be enormous.
  • the publisher 600 uses the entries in the queue table 606 to look up the ROLE ofthe SUID (from the publications table 608), and dynamically generates the request message during an update interval.
  • a subscriptions table 618 is used by the subscriber 610 to track of its subscriptions that are in effect.
  • An entry in the table 618 looks like this:
  • SUED comprises the identifier of the subscribing user
  • PUID comprises the identifier ofthe publishing user
  • PSID comprises the identifier ofthe publishing service
  • CN comprises the last known change number received from the publisher. Note that the existence of a row in this table implies that the associated publishing service 600 has one or more associated entries in its publications table 608. The CN field is required to ensure that publisher retries are idempotent.
  • the subscribing user specifies the PUID ofthe user whose data he or she wants to subscribe to. For example, if a userl changes a telephone number in userl 's profile, user2 can subscribe to see the change in user2's contacts, whereby (if user2 is properly authorized) the profile service becomes a publisher of userl 's changes and the contacts service becomes of subscriber of userl 's changes. The subscriber queries .NET Services (myServices) to find out the ID of the publisher (PSID) and stores the SUED/ PUID/PSED in subscriptions table 618.
  • myServices myServices
  • a subscriptions queue table 616 is used by the subscriber 610 to batch its requests for sending by the protocol handler 610 whenever the Updatelnterval timer goes off. Also, the subscriber is required to retry requests for which a response has not been received, and thus keeps track of messages that need to be sent for the first time, or need to be resent, which is also done in the subscriptions queue table 616. An entry in the table looks like this:
  • SUID comprises the identifier of the subscribing user
  • PUID comprises the identifier ofthe publishing user
  • PSID comprises the identifier of the publishing service
  • OPERATION comprises the Boolean (TRUE is an addition of a subscription and FALSE is a deletion of a subscription) and GENERATION indicates whether this message is fresh or has been sent one or more times already.
  • the subscription queue 616 does not store the messages, but rather during an update interval, the protocol handler simply looks at the OPERATION field (which indicates whether this request is to add a subscription or delete a subscription) and dynamically generates the appropriate request message.
  • the publisher simply deletes an added subscription. Note that if the publisher did not receive the original add or delete requests, it is equivalent to asking it to add an existing row or delete a non-existent row, which is handled by the idempotency rules.
  • UpdateSubscriptionData message is used when a user's document gets modified, to send change information to the subscribers.
  • the publishing service 600 checks the contents ofthe publications table 608 for interested subscribers by issuing the following logical query:
  • the publisher 600 uses the resultant information to create an entry in the queue; the said entry records the information necessary to construct an UpdateSubscriptionData message to each affected subscribing service.
  • an associated set of filtered data is created in a service- dependent manner.
  • the data is then factored by SSED, and an UpdateSubscriptionData message is created for each affected subscriber and sent, arrives.
  • the message format for UpdateSubscriptionData follows the following schema using the XMI conventions:
  • the data contained in the subscriptionData entity is defined by the participants in the service-to-service communication. Services which engage in multiple service-to- service communications should use the @topic attribute to disambiguate the meaning of the content.
  • the @topic attribute is a URI and is specific to the instance of service-to- service communication. For instance the .NET Profile to .NET Contacts communication could use a URI such as "urn:microsoft.com:profile-contacts: 1.0.” No service should attempt to accept an UpdateSubscriptionMap request for any conversation that they have not been explicitly configured to accept.
  • ⁇ updatedData> The function of ⁇ updatedData> is to inform the publisher, while the ⁇ deleteFromSubscriptionMap> is used by the subscriber to tell the publisher that this SUID has been deleted, as described below. Note that if a response is received for data that is not subscribed, an immediate delete may handle such a response.
  • the UpdateSubscriptionMap message is used when a set of one or more users changes their subscription status(es). When this occurs, the set of changes are sent to the affected publishers within an UpdateSubscriptionMap message. When the publisher receives this message it updates the records in the publications table 608. It is not an e ⁇ or to add an entry more than once, nor to delete a non-existent entry. In both these cases the response is formatted so that success is indicated. This is required to ensure that retries are idempotent.
  • the request message format for UpdateSubscriptionMap follows the following schema using the XMI conventions:
  • the addToSubscriptionMap section is used to make additions to the subscriptionMap, while the deleteFromSubscriptionMap removes entries.
  • the response message for updateSubscriptionMapResponse is formatted according to the following schema using the XMI conventions:
  • the ⁇ addedToSubscriptionMap> and ⁇ deletedFromSubscriptionMap> provide status information, while the entity ⁇ unknownPID> is used in situations where a publishing user is deleted.
  • Services also need to send out messages when they come on-line, e.g., to wake up other services which have stopped sending them messages.
  • the service should send out the following message to its partner services stored in its connections table (604 if a publisher, 614 if a subscriber, although it is understood that a service may be both a publisher and a subscriber and thus access both tables at such a time time).
  • the format of this message using the XMI conventions is: ⁇ serviceStatus> 1..1 ⁇ online/>0..1 ⁇ offline />0..1
  • a protocol handler wakes up when the interval timer goes off, whereby the handler sends the queued up requests, or when a request is received from another service, whereby the handler performs the requested action and sends a response.
  • a "Live Contacts" example as generally discussed above, will be used herein.
  • three .NET Profile services, having IDs of PSID), PSID 2 , and PSID 3 will be described.
  • PSIDi contains the profile documents of three users, namely PUIDn, PUID ⁇ 2 , and PUID ⁇ ;
  • PSED 2 contains profile documents of two users: PUID 2 i and PUID 22 ; and
  • PSED contains profile documents of two users: PUID ⁇ and PUID 32 .
  • There are two .NET Contacts services whose IDs are SSID1 and SSID2, wherein SSID1 manages contact documents of three users, SUID ⁇ , SUED ⁇ 2 , and SUID ⁇ 3 , and SSED2 manages contact documents of two users SUID 2 ⁇ and SUID 22 .
  • PUEDn friend(SUIDn), associate(SUID ⁇ 2 ) PUEDi.: other(SUID 2 ⁇ ) PUIDn: with respect to PS_D 2 : PUED 2 ⁇ : friend(SUID ⁇ )
  • PUID 22 friend(SUID 2 ⁇ ,SUID 22 ), associate(SUID, 2 ) and with respect to PSID 3 :
  • PUID 3 ⁇ associate(SUID ⁇ ), other(SUED u ) PUED 32 : friend(SUID 2 ⁇ ), associate(SUID 22 ) and also indicating with respect to SSEDi : SUIDii: PUIDn, PUID 2 ⁇ , PUED31 SUED 12 : PUIDn, PUID 22 SUED ⁇ 3 : PUID31 and with respect to SS_D 2 : SUID 2 ⁇ : PUID ⁇ 2 , PUID 22 , PUED 32
  • SUID 22 PUID 22 , PUID 32
  • the two contacts services each include a connections table.
  • this table (with included information such as cluster and URL omitted for simplicity) looks like:
  • the three profile services each contain a publications table.
  • this table (with included information such as change number omitted for simplicity) looks like:
  • the user SUIDn adds links to PU_D, 2 and PUID 32 and deletes the link from PUIDn, while SUID, 2 deletes the link to PUIDn the contents ofthe subscriptions queue for SSIDi is:
  • this table When processed, this table will generate two different UpdateSubscriptionMap requests that are sent to the two affected .NET Profile services. PSID, is sent:
  • each .NET Profile service updates the contents of their publications table as follows (with the CN change number column omitted). For PSID, the resulting table looks like:
  • PUID changes the contents on its profile, whereby PSED, constructs the following UpdateSubscriptionData message to SSID,:
  • the amount of information that is transmitted from one service to another is significantly reduced in SSCP because the change information for one user at a publisher service that is subscribed to by multiple users at a subscriber service who are assigned the same role at the publishing service, are aggregated into a single message.
  • the publisher operates in a fan-in model to put change information together based on their roles, rather than separate it per user recipient, and leaves it up to the subscriber to fan the information out to the appropriate users.
  • a user may change his profile to reflect a new telephone number, address, occupation and so forth;, , based on what they are authorized to see, e.g., as friends (who can see all such changes) or associates (who can only see telephone number and occupation changes), SSCP constructs a message with one copy of the friends data and one copy of the associates data, and sends this message to the subscriber.
  • the implicit assumption in this description is that all the subscribers reside on the same service. Should any of the subscribers reside on a different service, a separate message will be sent to that service, following the same aggregation principles outlined above.
  • SSCP is a robust protocol which is able to handle many different kinds of failure scenarios, including when the publisher fails, the subscriber fails, the link between publisher and subscriber goes down before the subscriber can respond (after it has received a request), the link between publisher and subscriber goes down before the publisher can respond (after it has received a request), the publisher loses the subscription map, and the subscriber loses published data.
  • failure scenarios are handled by message retries and idempotency, as generally described below. Message retries will be described with respect to an example that assumes the publisher sends the request message. However the message-retry mechanism applies equally well when the subscriber sends the retry message.
  • the publisher When the publisher sends a request message, the publisher sends the message from the publications queue and waits for a response to this message. If the publisher gets a response, it deletes the message from the queue, otherwise it keeps the message in the queue and resends it the next time Update Interval timer goes off. As described above the number of retries occurs a specified maximum number of times, after which the subscriber is considered dead. After some longer interval time, the subscriber is automatically tested for aliveness, and the process begins all over. This aliveness testing can also be limited to some number of times. This method ensures that an alive subscriber does not miss an UpdateSubscriptionData message.
  • retry attempts should idempotent - that is, multiple retries of a request should behave as if the request had been sent only once. Idempotency is achieved by keeping track of the change number, or CN, which is a column in the publications and subscriptions tables as described above.
  • CN change number
  • change numbers are represented as an as an integer sequence, although it is understood that change numbers need not be sequential, but may be whatever the service has, as long as it increases (or decreases) monotonically.
  • the smallest unit of change is a .NET blue node, the smallest query-able, cacheable, unit of data in .NET.
  • the publisher 600 when a fresh subscription is created, the publisher 600 adds a row into the publications table 608 (FIG. 6), with CN being set to the lower (upper) bound for the change number. . Note that since every .NET blue node already has a change number associated with it, this value is guaranteed to be available.
  • the subscriber 610 also keeps track of the value of this CN in its subscriptions table 618. Whenever the publisher 600 sends an UpdateSubscriptionData request to the subscriber, it includes the value of CN that it cu ⁇ ently has for this [.NET blue]node. It records this CN in the publications table 608.
  • the subscriber 610 On receiving the UpdateSubscriptionData message, the subscriber 610 updates its copy ofthe CN (present in the CN field of subscriptions table 618) to the new value. If, due to a transient network failure, the publisher 600 fails to receive the response message from the subscriber, the publisher resends the request message again at the next update interval. On receiving this request, the subscriber inspects the CN, and determines that it has already processed this message because the CN in the message is the same as the CN that it has. The subscriber treats this as a no-op with respect to making any update, and sends back a response whereby the publisher will normally receive it and delete this message from the message queue. The net result is that any message received multiple times by the subscriber is processed exactly once, i.e., retries are idempotent.
  • the subscriber achieves idempotency because when a publisher receives a request to add a preexisting entry to its subscription map, it should treat this as a no-op, and not return an e ⁇ or. When the publisher receives a request to delete a non-existent entry from its subscription map, it should treat this as a no-op and not return an error. As can be readily appreciated, multiple add or delete from subscription map requests behave as if there was only one such request.
  • the publisher fails, the publisher will not be able to respond to subscriber requests to update the subscription map. This is handled by resending the message until a response is received. As with other retries, long-term or catastrophic failures are handled by having a limit on the number of retries and waiting for a longer time before starting all over, and then if still no response after some number of "longer" time cycles, requiring the attempted recipient to initiate contact.
  • the publisher will also not receive any responses that the subscriber may have sent to its UpdateSubscriptionData requests. From the point of view ofthe subscriber, this is logically indistinguishable from the case where the link between subscriber and publisher fails, and is handled as described below.
  • Subscriber failures are very similar to what happens when the publisher fails.
  • the subscriber continues to resend the UpdateSubscriptionMap requests until it receives a response from the publisher, or the retry limit is reached, whereupon the retry attempts will be held off for a longer delay time.
  • the non-reception of responses by the subscriber is the same as a link failure, the handling of which is explained below.
  • the subscriber has sent an UpdateSubscriptionMap message
  • the publisher has processed this message and sent a response, but the subscriber does not receive the response. As described above, this causes the subscriber to resend the message.
  • the publisher receives a duplicate UpdateSubscriptionMap message from the subscriber, detected via the change number. Since retries are idempotent, the publisher simply sends back a response to the subscriber. A subscriber to publisher link failure is handled similarly.
  • a PUID may be deleted from the publisher and for some reason the subscriber does not get notified of this event.
  • the publisher comes back with the ⁇ unknownPID> entity in the response. This tells the subscriber to update its image ofthe subscription map.
  • a SUID may be occasionally deleted at the subscriber and in general, the publisher has no way of knowing it. On data change, the publisher sends an update request to the deleted SUID, and when this happens, the subscriber sends a
  • ⁇ deleteFromSubscriptionMap> entity in its response to notify the publisher ofthe SUED deletion. This tells the publisher to update its subscription map.
  • One catastrophic form of failure is when a publisher loses its subscription map or the subscriber loses its subscription data. This can cause various levels of data loss. For example, if the publisher has experienced a catastrophic failure, such as disk crash, the publisher needs to revert to data from a back up medium such as tape. As a result, its subscription map is out of date. For the subscriber, a similar situation makes its subscribed data out of date.
  • the service that experienced the loss sends a message requesting an update.
  • the publisher's subscription map can be brought up to date by the information stored in subscriptions table in the subscriber, while a subscriber's data can be made up to date by the subscription map and the change number stored in the publications table.
  • // PUID is the user id whose data was changed. Query the publications // table for all SUIDs that are affected, and insert this data into // the PUBLIC ATIONS_QUEUE, if it does not exist already
  • VALUES (%suid%, %puid%, %psid%, TRUE, 0)
  • each user can have multiple instances of a .NET (or my*) service.
  • a user can have two instances ofthe myContacts service, one for company contacts and one for personal contacts, (although the same segmentation can also be achieved using categories).
  • INSTANCE an identifier stored in the myServices service.
  • OID owner-id
  • INSTANCE an identifier stored in the myServices service.
  • FIG. 17 shows an example of a publisher-subscriber relationship.
  • myProfile services 1701 and 1702 each managing the profiles of three users.
  • Useri has three instances (1704 ⁇ -1704 3 ) of a myProfile service
  • user 6 has four instances, one of which resides in the first myProfile service 1701, three of which reside in the second myProfile service 1702.
  • myContacts service 1720 which manages the contact information of two users; user 2 has two instances (1722 ⁇ and 1722 2 ) ofthe service.
  • each of these services will manage the data for millions or even hundreds of millions of users.
  • myContacts service has subscriptions in the two different myProfile services 1701 and 1702; it is similarly likely that a given publisher will publish to multiple .NET services.
  • a single service e.g., in the whitelist example, myContacts is a publisher; in the Live Contacts example, it is a subscriber).
  • myProfileDoc ⁇ i when the profile information for myProfileDoc ⁇ i changes, this information should be published by myProfile service 2 1702, to myContacts service 1720, as both myContactsDoc ! 1721 and myContactsDoc 2 ⁇ 1722 ⁇ have subscribed for the service.
  • SSCP enables the publishing of data as changes occur, via the push model. Furthermore, in keeping with the present invention, the publisher should make all attempts to batch the changes to maximally utilize bandwidth.
  • myProfileDoc 5 subscribes to the profile changes of myProfileDoc 5 . Thus, when User 5 's profile is changed, myProfile should publish the changes only to myContactsDoc 2 ⁇ , and myContactsDoci should not see these changes.
  • MyProfileDoc ⁇ i is that of an associate; the role of User 2 is that of a friend.
  • a myProfile service When a myProfile service publishes the data, it should send data visible to an associate to myContactsDoc, and data visible to a friend to myContactsDoc 2 ,.
  • SSCP sends changes only to subscribed documents (user/instance) within a subscribing service, and determines the role of each subscribing user, and filter the data based on the role. To this end, the publisher maintains information about documents wanting subscriptions, which is determined by the OED/INSTANCE pair (myContactsDoc, and myContactsDoc 2 >).
  • the publisher For each subscribing document, the publisher also maintains information about the document it is subscribing to (for myContactsDoc,, this is myProfileDoc 2 and myProfileDoc 3 in myProfile Service,), and about the role played by the owner ofthe subscribing document (for myProfileDoc ⁇ , in myProfile Service 2 , this is associate for myContactsDoc,, friend for myContactsDoc 2 ,).
  • the subscriber should notify the publisher whenever one of its users wants to unsubscribe or add a new subscription.
  • subscribes that is, a user specifies an instance of the service which wants to act as a subscriber, but for purposes of description the user can be thought of as a subscribing.
  • User wants to add User 4 into his live contact list and remove User 6 .
  • SSCP should allow for transmission of this information from subscriber to publisher.
  • SSCP allows the subscriber to send subscription updates to the publisher.
  • the alternative embodiment described in this section provides robustness, to guarantee that the publisher and subscriber see the messages that they are supposed to see.
  • the publisher or subscriber need to know that their messages have reached the destination, whereby a message from the sender has a co ⁇ esponding acknowledgement (ACK) returned from the receiver.
  • ACK need not be synchronous with respect to the message, and can instead be sent / received asynchronously.
  • the robust protocol ofthe present invention also handles the failures of publishers or subscribers, which is generally accomplished by resending a request until a response is received. However, to prevent a flood of retry messages in case of a catastrophic failure at the destination, a limited number of retries are specified, after which no further attempts are made for a longer time. This is accomplished via a reset interval (which is relatively much longer than the retry interval) after which the entire retry process begins.
  • a more subtle type of failure occurs when, for example, a publisher sends a request to the subscriber, informing it ofthe change in a stored profile, the subscriber processes the request, and sends a response to the publisher, but the network connection between the subscriber and the publisher has a transient failure and the response does not reach the publisher.
  • the publisher resends its request.
  • the subscriber recognizes that this is a redundant request that has already been processed. In other words, a request should be processed only once even if it is sent multiple times; alternatively, the request could be processed any number of times, but the next result should be as if it was processed only once.
  • retries are idempotent.
  • a typical service manages gigabytes of data, partitioned over millions of users. This means that in its role as a publisher, the source data will be frequently, if not almost constantly, changing. For efficiency, every change is not published immediately, but instead change requests are batched, and send occasionally (e.g., periodically). To this end, the protocol handler at the service periodically wakes up after a specified interval and sends the batched messages, as described above with respect to FIG. 6. As generally represented in FIG. 6, SSCP is implemented at a publisher (service)
  • protocol handlers 602, 612 such as daemon processes or the like running with respect to a service.
  • the publisher 600 and subscriber 610 exchange messages, and use this as a mechanism to communicate changes.
  • the requirements ofthe protocol dictate that SSCP handlers 602, 612 maintain several pieces of data, the sum total of which represents the state of a publisher or subscriber. As conceptually represented in FIG. 6, this data can be viewed as being segmented over several data structures 604-618. Note however that the a ⁇ angements, formats and other description presented herein are only logically represent the schema; the actual storage format is not prescribed, and an implementation may store in any fashion it deems fit as long as it logically conforms to this schema.
  • a publisher 600 communicates with a subscriber 610 using request and response messages. For example, when data changes at the publisher 600, the publisher 600, sends a request message to the subscriber 610 informing the subscriber that data has changed, normally along with the new data. The subscriber 610 receives the message, makes the required updates, and sends back an ACK message acknowledging that the message was received and that the changes were made. A subscriber 610 can also send a request message, such as when the subscriber 610 wants to subscribe or un-subscribe to a piece of datum. When the publisher 600 receives this message, the publisher 600 updates its list of subscriptions (in a publications table 608) and sends back a response acknowledging the request. Note that SSCP is agnostic to whether a response message for a given request is synchronous or asynchronous.
  • SSCP there are two primary parts to SSCP, a first from the publisher to the subscriber, which deals with sending changes made to the publisher's data, and a second from subscriber to the publisher, which deals with keeping the list of subscriptions synchronized. Furthermore, every service is required to provide notification to all other services that have subscriptions with it, or services with which it has subscriptions, when it is going offline or online.
  • Protocol parameters are supported by both the publisher and the subscriber and control the behavior ofthe protocol.
  • SSCP supports the ability to batch request messages. Whenever there is a need to send a request message, such as when there are changes in publisher data or subscriptions, the service puts the co ⁇ esponding request message into a publisher message queue 606. Periodically, the protocol handler 602 in the publishing service 600 wakes up and processes the messages in the queue 606. This period is called as the Updatelnterval, and is a configurable parameter.
  • the publisher's protocol handler 602 needs to periodically resend requests until the publisher service 600 receives an acknowledge message (ACK). If the ACK for a message is successfully received, this message is purged from the queue 606. Until then, the message remains in the queue, flagged as having been sent at least once, so it will be retried at the next update interval.
  • the number of times the publisher the publisher service 600 retries sending a message to the subscriber service 610 is configurable by the parameter RetryCount, i.e., after retrying this many times, the publisher service 600 assumes that the subscriber service 610 is dead. Then, once the maximum number of retries is over, the publisher service 600 waits for a relatively longer time. Once this longer time is elapsed, the publisher service 600 sets the RetryCount parameter to zero and begins resending the queued up requests over again. This longer time (before beginning the retry cycle), is configurable by the parameter Resetlnterval.
  • the protocol handlers 602, 610 at the publisher and subscriber, respectively track of several pieces of information, such as in their respective tables 604-618.
  • SSCP relies on the entities (services and users) being uniquely identifiable by the use of identifiers, e.g., every user in .NET has a unique identifier assigned by the Microsoft ® Passport service.
  • Each service be it acting as a publisher or subscriber, also has a unique identifier, and in practice, a service ID will be a certificate issued by a certification authority.
  • the service To send a request or a response, the service needs to know where the target is located. For purposes ofthe protocol a service is identified either by just the URL or by a series of URL/CLUSTER entries. To ensure proper handling ofthe number of retries for a particular service, the handler needs to keep track of how many retries have been done. All this information is kept in the CONNECTIONS table, which is used by both publishers and subscribers:
  • the publisher tracks the users'across the services with which it has subscriptions. This is done in the PUBLICATIONS table.
  • the PUBLICATIONS table used by the publisher, looks like:
  • the document (SO;, SIj) subscribes to the document (POi, PI;) with topic
  • the PUBLICATIONS_QUEUE table is used by the publisher to batches the requests for the protocol handler to send when the interval is achieved, e.g., the Updatelnterval timer goes off. Also, the publisher is required to retry requests for which a response has not been received. The publisher thus tracks the messages that need to be sent for the first time, or those that need to be resent. This is done in the PUBLICATIONS_QUEUE table, which looks like this:
  • the PCN field is required to ensure correct updates in situations when multiple updates happen to the underlying data before a response is received from the subscriber.
  • PCN (5).
  • change number six (6) occurs. This causes the PCN in the PUBLICATION_QUEUE to be updated from five (5) to six (6).
  • the response comes back from the subscriber for the original message containing the change number that it had received, which is equal to five (5).
  • the publisher compares this change number with the change number that it has stored in the PUBLICATION_QUEUE table, and finds that the one in the table has a value of 6. So, it knows that more changes need to be sent to the subscriber (those co ⁇ esponding to change number six (6)), and hence it retains the row in the queue.
  • the Publication Queue Store does not store messages, but the information needed to create the messages.
  • the storage required by these messages is likely to be huge, so rather than storing the actual messages in the table, during an update interval, the publisher uses entries in this table to look up the ROLE of the owner ofthe subscribing document (from the PUBLICATIONS table), and generates the request message at the time of sending it.
  • Another reason for not storing messages deals with multiple updates occurring within a single updateinterval. In this case multiple copies ofthe messages would needlessly get generated and then overwritten.
  • Another reason to not store messages in the queue is that messages are collated so that similar data payloads get combined into a single outbound request. Generating messages for every queue entry would mean a redundant effort, discarded at message send time.
  • TOPIC If the subscribing document is having multiple subscriptions with a publishing document, then a TOPIC is used to distinguish them.
  • the subscription queue does not store messages. Instead, the OPERATION field in the Queue indicates whether this request is to add a subscription or delete a subscription. During an update interval, the protocol handler simply looks at the OPERATION field and dynamically generates the appropriate request message. Thus, even though the subscription queue does not store the message, it has the information needed to formulate the message. Further, note that the subscription queue has multiple columns, while the publication-queue has only a key, because the publication queue only needs to identify which one ofthe pre-existing subscriptions needs a data update. Thus, it only needs to store the row-id in the PUBLICATIONS table. However, the subscription queue sometimes needs to add a subscription, and the information needed for this purpose should be in the subscription queue.
  • the SCN field is required to ensure co ⁇ ectness in cases where the user adds/deletes the same subscription multiple times - for example, the user adds a subscription, and then deletes it or deletes a subscription and then adds it - before the original request was sent to, and a response received from, the publisher.
  • each change of mind on the part ofthe user is treated as a change, and is assigned a change number. This number is passed back and forth between subscriber and publisher in the request and response messages and ensure that the multiple adds and deletes are processed properly.
  • This UpdateSubscriptionData message is provided when a user's document gets modified.
  • TOPIC %TOPIC% GROUP BY SSID, ROLE
  • the publisher uses this information to construct an UpdateSubscriptionData message to each affected subscribing service. For the set of distinct ROLES used within the result set an associated set of filtered data is created in a service dependent manner. Then, the data is factored by SSID and each affected subscriber is sent an UpdateSubscriptionData message (actually the messages are queued up and sent the next time the Update Interval timer goes off).
  • UpdateSubscriptionData follows the following schema using the XMI conventions:
  • the data contained in the subscriptionData entity is defined by the participants in the service-to-service communication. Documents which engage in multiple publish/subscribe relationships should use the @topic attribute to disambiguate the meaning of the content.
  • the @topic attribute is a URI and is specific to the instance of service-to-service communication. For instance the myProfile to myContacts communication topic could use a URI like: urn:microsoft.com:profile-contacts: 1.0. No service should attempt to accept an UpdateSubscriptionMap request for any conversation that they have not been explicitly configured to accept.
  • ⁇ updatedData> The function of ⁇ updatedData> is to inform the publisher, while ⁇ deleteFromSubscriptionMap> is used by the subscriber to tell the publisher that this SOID/SINST has been deleted.
  • UpdateSubscriptionMap When a set of users change their subscription statuses, the set of changes are sent to the affected Publishers within an UpdateSubscriptionMap message.
  • the Publisher receives this message it updates the records in the PUBLICATION TABLE. It is important to the correctaess ofthe protocol that all updates are handled robustly. In particular it is not an e ⁇ or to add an entry more than once. Likewise it is not an e ⁇ or to delete a non-existent entry. In both these cases it is important to format the response so that success is indicated for these cases.
  • the message format for UpdateSubscriptionMap follows the following schema using the XMI conventions:
  • the addToSubscriptionMap section is used to make additions to the subscriptionMap, while the deleteFromSubscriptionMap removes entries.
  • the ⁇ addedToSubscriptionMap> and ⁇ deletedFromSubscriptionMap> provide status information, while the entity ⁇ unknownPID> is used in situations where a publishing user is deleted. Services also need to send out messages when they come on-line, e.g., to wake up other services which have stopped sending them messages. To this end, whenever a service is going offline or coming online, the service should send out the following message to its partner services stored in its connections table (604 if a publisher, 614 if a subscriber, although it is understood that a service may be both a publisher and a subscriber and thus access both tables at such a time time).
  • the format of this message using the XMI conventions is:
  • SSCP is designed so that the protocol does not impose any indigenous restrictions on what can or cannot be subscribed to.
  • a service can request a subscription to all of publisher's data (at least, all that is visible to it). However, it may also subscribe to only a subset of it.
  • the "topic" attribute of UpdateSubscriptionMap message is used to specify this. From the perspective of SSCP, a topic is simply an identifier (mutually agreed upon by the subscriber and publisher) which specifies what the subscriber wants to subscribe to. For instance, if mylnbox service only wants to subscribe to an email address in myContacts service (which is the case for whitelists) then one way of using "topic" attribute would be:
  • a subscribing document S can have multiple subscriptions with a publishing document P where each subscription differs by only the topic attribute.
  • the protocol handler wakes up when the interval timer goes off, and the handler sends the queued requests, or a request is received from another service, and the handler performs the requested action and sends a response.
  • FIG. 18 In FIG. 18, in which there are three myProfile services whose IDs are PSIDi, PSID ⁇ and PSID 3 . In FIG. 18:
  • PSLDi contains the profile documents of three users: POIDn, POID ⁇ 2 , POID ⁇ 3
  • POEDn has three instance documents: 1, 2, and 3.
  • POID12 and POIDn have one instance document each.
  • PSID2 contains profile documents of two users: POID 2 ⁇ and POID 22 , each having one instance document.
  • PSID 3 contains profile documents of two users: POID.i and POID 32 .
  • POID 3 ⁇ has one instance document.
  • PO_D 32 has two instance documents: 1 and 2.
  • SSIDi manages contact documents of three users: SOID ⁇ , SOIDn, and SOIDn, each with one instance document.
  • SS_D 2 manages contact documents of two users: SOID21 and SOID 22 .
  • SOED21 has two instance documents: 1 and 2.
  • SOID 22 has one instance document.
  • the two contacts services each include a CONNECTIONS table (for simplicity, information such as cluster, URL, and so on, are not shown below).
  • the connections table includes:
  • the three profile services each contain a PUBLICATIONS table (for simplicity, information such as PKEY or SCN columns are not shown below). For PSED, this looks like:
  • document SOED,, /instance 1 adds links to the documents POID, 2 /instancel and PO_D3 2 /instance2 and deletes the link from POIDn/instancel, while SOID, 2 /instancel deletes the link from POID,, /instance 1 the contents ofthe SUBSCRIPTIONS_QUEUE for SSID, is:
  • each myProfile service updates the contents of their PUBLICATIONS table as follows (with the TOPIC and SCN columns not shown). For PSID, the resulting table looks like:
  • SSCP is a robust protocol which is able to handle many different kinds of failure scenarios, including:
  • SSCP When the publisher sends a request message, SSCP follows the following algorithm: 1 ) Publisher sends a message from the PUBLIC ATIONS_QUEUE.
  • retry attempts should idempotent, i.e., multiple retries of a request should behave as if the request had been sent only once. Idempotency is achieved by keeping track of the change number, or PCN (which is a column in the PUBLICATIONS and SUBSCRIPTIONS tables).
  • PCN which is a column in the PUBLICATIONS and SUBSCRIPTIONS tables.
  • the underlying service implementation has change number data, and keeps track of it, independent of SSCP.
  • change numbers are logically reflected as an integer sequence, however in general, the PCNs need not be sequential, but instead may be whatever the service has, as long as it increases or decreases monotonically.
  • the smallest unit of change is a .NET blue node, wherein cu ⁇ ently a blue node is the smallest query-able, cacheable, unit of data in .NET.
  • the publisher When a fresh subscription is created, the publisher adds a row into the PUBLICATIONS table, with PCN being set to 0 to indicate that no data has yet been exchanged. The subscriber also keeps track of the value of this PCN in its SUBSCRIPTIONS table. Whenever the publisher sends an UpdateSubscriptionData request to the subscriber, it includes the value of PCN that it cu ⁇ ently has for this (e.g., blue) node. It records this PCN in the PUBLICATIONS table. On receiving the UpdateSubscriptionData message, the subscriber updates its copy of the PCN (present in the PCN field of SUBSCRIPTIONS table) to the new value.
  • UpdateSubscriptionData On receiving the UpdateSubscriptionData message, the subscriber updates its copy of the PCN (present in the PCN field of SUBSCRIPTIONS table) to the new value.
  • the publisher fails to receive the response message from the subscriber, it resends the request message again at the next update interval.
  • the subscriber inspects the PCN; it knows that it has already processed this message because the publisher's change number in the message is the same as the PCN that it has, and thus treats this as a no-op and sends back a response.
  • the publisher deletes this message from the message queue, and the net result is, any message received multiple times by the subscriber is processed exactly once - i.e., retries are idempotent.
  • the subscriber achieves idempotency is by the following rules: when a publisher receives a request to add a preexisting entry to its subscription map, it should treat this as a no-op and not return an e ⁇ or. When the publisher receives a request to delete a nonexistent entry from its subscription map, it should treat this as a no-op and not return an e ⁇ or. As can be appreciated, multiple add or delete from subscription map requests behave as if there of only one such request.
  • the SCN field is required to ensure co ⁇ ectaess in cases where the user adds/deletes the same subscription multiple times - for example, the user adds a subscription, and then deletes it or deletes a subscription and then adds it - before the original request was sent to, and a response received from, the publisher.
  • each change of mind on the part of such a user is treated as a change, and is assigned a change number. Change numbers are monotonically increasing.
  • the SCN is updated to a new increased value.
  • the subscription queue contains only the last request made by the user; but the change number has increased every time the user changes his mind.
  • the UpdateSubscriptionMap request includes the cu ⁇ ent value ofthe change number from the queue for each add or delete entity present in the request.
  • the delete request arrives at the publisher before the add request.
  • the subscriber deletes the co ⁇ esponding entity from the queue.
  • the subscriber ignores this response since there is no entity in the subscription queue co ⁇ esponding to this response.
  • the delete request arrives at the publisher before the add request.
  • the subscriber deletes the co ⁇ esponding entity from the queue.
  • the subscriber ignores this response since there is no entity in the subscription queue co ⁇ esponding to this response.
  • this unusual case simply means that there will exist one or more rogue subscriptions at the publisher until such time that the data subscribed by these rogue subscriptions change. At this point, the protocol logic takes over and deletes the rogue subscription. Note that the vast majority ofthe time, the simple case (1) is what takes place, and the other cases occur only very rarely.
  • the publisher When the publisher fails, the publisher will not be able to respond to subscriber requests to update the subscription map, which is handled by resending the message until a response is received. Long-term or catastrophic failures are handled by having a limit on the number of retries and waiting for a "long time" before starting all over. The publisher will also not receive any responses that the subscriber may have sent to its UpdateSubscriptionData requests. From the point of view ofthe subscriber, this is
  • a failure case can occur when the subscriber has sent an UpdateSubscriptionMap message, and the publisher has processed this message and sent a response, but the link between the publisher and subscriber fails. As a result, the subscriber does the not receive the response. As described in the section "Message retries", this causes the subscriber to resend the message. Thus the publisher receives a duplicate UpdateSubscriptionMap message from the subscriber. Since retries are idempotent, the publisher simply sends back a response to the subscriber. When the subscriber to publisher link fails, it is handled similarly. Occasionally, POID/INSTANCE is deleted from the publisher, and the subscriber usually does not get notified of this event.
  • the publisher's subscription map can be brought up to date by the information stored in subscriptions table in the subscriber, while a subscriber's data can be made up to date by the subscription map and the change number stored in the publications table. In general, the service that experienced the loss has enough knowledge to send a message requests an update.
  • the publisher's subscription map can be brought up to date by the information stored in SUBSCRIPTIONS table in the subscriber.
  • a subscriber's data can be made up to date by the subscription map and the publisher's change number stored in the PUBLICATIONS fable.
  • // PUID/PINST is the user id whose data was changed. Query the publications // table for all SUEDs that are affected, and insert this data into // the PUBLICATIONS_QUEUE, if it does not exist already.
  • ______ ROLE FindRole(POID, PINST, SOID);
  • UpdateSingleSubscriptionMap SOID, SINST, POID, PINST, PSID, TOPIC
  • FIG. 19 represents one such cluster architecture.
  • the load balancer redirects the request to a front end server (FE), based on load balancing and fault-tolerance considerations.
  • the FE does some preliminary processing on the request, locates the back-end server (BE) servicing this user, and forwards the request to the back end server.
  • BE returns with a response, which the FE puts into an appropriate message format (e.g., .NET data language) and sends it off to its destination.
  • an appropriate message format e.g., .NET data language
  • the load-balancer on an incoming request, can distribute load by choosing an FE which is not busy.
  • the same property allows the load balancer to be fault-tolerant when an FE fails.
  • the BE is stateful; when required by the semantics of a service, the BE remembers history.
  • each BE services a subset ofthe users ofthe entire service, and while the choice of an FE is arbitrary, a given request always co ⁇ esponds to one specific BE - the one which stored the user's data.
  • the a ⁇ ows labeled with circled numerals one (1) through eight (8) represent the data flow on a typical request, with (1), a request comes to the service's load balancer 1900. Then, the load balancer determines that FE 3 is the right front-end to handle this request (based on load and failover considerations), and (2) provides the request to FE 3 which processes the request. FE 3 determines the user identity, and locates the BE that services this user, which in the present example, is BE,. FE 3 determines what data is needed from the backend, and FE3 sends database requests to BE, (arrow labeled three (3)).
  • BE retrieves the required data from the database (a ⁇ ows labeled four (4) and five (5)), and BE, sends data back to FE 3 , in the form of database response (arrow six (6)). Then, FE 3 returns the data back into an appropriate response and sends the message off to its destination (a ⁇ ows labeled seven (7) and eight (8)).
  • the model represented in FIG. 19 works fine for handling incoming SSCP requests. For example, when an UpdateSubscriptionMap request comes into a publisher, it is processed in the general manner described above. However, for outgoing requests, such as when the publisher needs to send the UpdateSubscriptionData message, an enhanced model is provided, generally because in the SSCP protocol, a publisher or a subscriber processes its queue every time the interval timer goes off, and for the protocol to function co ⁇ ectly, a single reader should drain the queue, and also because in the model described in the previous section, the BE has no reason to initiate a request message; its job is to process a request and generate an appropriate response. However, SSCP requires that the participating services generate requests when the interval timer goes off: a) A publisher sends UpdateSubscriptionData messages b) A subscriber sends UpdateSubscriptionMap messages
  • the word “service” refers to either the publisher or the subscriber
  • the word “queue” refers to either the publication queue or the subscription queue.
  • the FEs run code for inbound SSCP messages, just as they do for other inbound .NET data language messages. This means that the FEs run code for updating subscription data (on the subscribing side), code for updating subscription maps (on the publishing side), and processing SSCP responses (both subscriber and publisher).
  • the BEs run code for outbound SSCP messages. This code runs every time the interval timer goes off. This code handles the publication queue and generating UpdateSubscriptionData messages (publisher), handling subscription queue and generating UpdateSubscriptionMap messages (subscriber).
  • the process generally works as follows: 1) Each BE stores a slice ofthe persistent SSCP data. Taking the example of a publishing service, if BE1 manages user, and user, 2 , and BE 2 manages user 2 ⁇ and user 22 then
  • BE stores PUBLICATIONS and PUBLIC ATIONS_QUEUE and CONNECTION tables which handle the subscription / publication requirements for data from user, , and user, 2 .
  • BE 2 stores PUBLICATIONS and PUBLICATIONS_QUEUE and CONNECTION tables which handle the sub/pub requirements for data from user2i and user22-
  • the BE constructs the appropriate message(s) - such as UpdateSubscriptionData, or UpdateSubscriptionMap. For each message: a) The BE picks an FE (e.g., at random) and sends the message to it. b) The FE simply forwards the message along to its destination - i.e., it acts as a proxy. c) A response is handled in the usual way (since incoming SSCP messages don't require any changes)
  • FIG. 20 generally represents this model when the interval timer goes off and the following things happen at BE, (similar things also happen at other BEs).
  • BE has to send two requests, requestl and request2, as a result of processing its queue during this interval timer event.
  • a ⁇ ows labeled (A) BE, sends requestl through FE 3 , which is randomly picked.
  • the a ⁇ ows labeled (B) represent a response arriving from a destination service through FE 2 , which is picked by the load balancer according to its algorithms.
  • the a ⁇ ows labeled (C) represent BE sending request2 through randomly picked FE, .
  • the a ⁇ ows labeled (D) represent a response arriving from a destination service through FE, (which is picked by the load balancer according to its algorithms).

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Business, Economics & Management (AREA)
  • Computer Security & Cryptography (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computer Hardware Design (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Economics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Bioethics (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Storage Device Security (AREA)

Abstract

La présente invention concerne un protocole de communication de service à service robuste et efficace, qui gère des informations de modification dans une architecture d'accès à des données centrées sur l'identité. Ce protocole permet une publication et une inscription automatiques, au moyen de services de modifications effectuées sur des données de millions d'utilisateurs. Le protocole est basé sur les rôles en ce qu'un utilisateur commande les utilisateurs qui peuvent s'inscrire aux modifications de données d'utilisateur et est efficace en ce que des données qui sont des données de modification pour des utilisateurs sont combinées et regroupées. Ce protocole est assez robuste pour gérer des scénarios de panne. Dans un mode de réalisation de la présente invention, l'éditeur (600) se rapporte au service « .NET MyServices » qui est la source des données, alors qu'un abonné (610) se rapporte au « .NET MyServices » qui reçoit les données. L'éditeur et l'abonné conservent des informations mises à jour concernant les utilisateurs de chacun, afin de réaliser une communication et une filtration de données sélective. Des requêtes sont admises afin de fournir une certaine robustesse. Ces requêtes sont relancées régulièrement pendant un moment, avec des délais entre les relances régulières, jusqu'à être admises.
EP02719261A 2001-03-14 2002-03-14 Communication de service a service pour des services de reseau Withdrawn EP1370965A4 (fr)

Applications Claiming Priority (7)

Application Number Priority Date Filing Date Title
US27580901P 2001-03-14 2001-03-14
US275809P 2001-03-14
US10/017,680 US7302634B2 (en) 2001-03-14 2001-10-22 Schema-based services for identity-based data access
US33177 2001-10-22
US10/033,177 US20030061365A1 (en) 2001-03-14 2001-10-22 Service-to-service communication for network services
US17680 2001-10-22
PCT/US2002/008063 WO2002073442A1 (fr) 2001-03-14 2002-03-14 Communication de service a service pour des services de reseau

Publications (2)

Publication Number Publication Date
EP1370965A1 true EP1370965A1 (fr) 2003-12-17
EP1370965A4 EP1370965A4 (fr) 2007-03-07

Family

ID=27360857

Family Applications (1)

Application Number Title Priority Date Filing Date
EP02719261A Withdrawn EP1370965A4 (fr) 2001-03-14 2002-03-14 Communication de service a service pour des services de reseau

Country Status (2)

Country Link
EP (1) EP1370965A4 (fr)
WO (1) WO2002073442A1 (fr)

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7024662B2 (en) 2001-03-14 2006-04-04 Microsoft Corporation Executing dynamically assigned functions while providing services
US7302634B2 (en) 2001-03-14 2007-11-27 Microsoft Corporation Schema-based services for identity-based data access
US6975833B2 (en) 2002-02-07 2005-12-13 Sap Aktiengesellschaft Structural elements for a collaborative e-learning system
US7237189B2 (en) 2002-02-11 2007-06-26 Sap Aktiengesellschaft Offline e-learning system
US7014467B2 (en) 2002-02-11 2006-03-21 Sap Ag E-learning course structure
US7029280B2 (en) 2002-02-11 2006-04-18 Sap Ag E-learning course editor
US9886309B2 (en) 2002-06-28 2018-02-06 Microsoft Technology Licensing, Llc Identity-based distributed computing for device resources
US7376585B2 (en) 2002-09-04 2008-05-20 Sap Aktiengesellschaft Real-time sales information architecture
US7417959B2 (en) 2003-09-29 2008-08-26 Sap Aktiengesellschaft Audio/video-conferencing using content based messaging
US7287066B2 (en) * 2003-10-31 2007-10-23 Sap Aktiengesellschaft Publish-subscribe system having a reliability mechanism
US7840175B2 (en) 2005-10-24 2010-11-23 S&P Aktiengesellschaft Method and system for changing learning strategies
US8571462B2 (en) 2005-10-24 2013-10-29 Sap Aktiengesellschaft Method and system for constraining learning strategies
CN112600942B (zh) * 2021-02-18 2022-12-02 杭州网银互联科技股份有限公司 一种应用于提升sd-wan中的路由计算效率的方法及系统

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5974417A (en) * 1996-01-18 1999-10-26 Sun Microsystems, Inc. Database network connectivity product
US6088717A (en) * 1996-02-29 2000-07-11 Onename Corporation Computer-based communication system and method using metadata defining a control-structure

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6044372A (en) * 1997-07-18 2000-03-28 Dazel Corporation Method and apparatus for publishing information to a communications network and enabling subscriptions to such information
US6269369B1 (en) * 1997-11-02 2001-07-31 Amazon.Com Holdings, Inc. Networked personal contact manager
US6148301A (en) * 1998-07-02 2000-11-14 First Data Corporation Information distribution system
US6167408A (en) * 1998-08-31 2000-12-26 International Business Machines Corporation Comparative updates tracking to synchronize local operating parameters with centrally maintained reference parameters in a multiprocessing system
US6266690B1 (en) * 1999-01-27 2001-07-24 Adc Telecommunications, Inc. Enhanced service platform with secure system and method for subscriber profile customization

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5974417A (en) * 1996-01-18 1999-10-26 Sun Microsystems, Inc. Database network connectivity product
US6088717A (en) * 1996-02-29 2000-07-11 Onename Corporation Computer-based communication system and method using metadata defining a control-structure

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of WO02073442A1 *

Also Published As

Publication number Publication date
WO2002073442A1 (fr) 2002-09-19
EP1370965A4 (fr) 2007-03-07

Similar Documents

Publication Publication Date Title
US20030061365A1 (en) Service-to-service communication for network services
Cabrera et al. Herald: Achieving a global event notification service
AU2003225818B2 (en) Data replication system and method
AU2004290093B2 (en) A directory system
US20180007144A1 (en) Event queuing and distribution system
US6748534B1 (en) System and method for partitioned distributed scanning of a large dataset for viruses and other malware
RU2379755C2 (ru) Система и способ для совместного использования объектов между компьютерами по сети
US20020198943A1 (en) Web-enabled two-way remote messaging facility
US20070005711A1 (en) System and method for building instant messaging applications
US20020133535A1 (en) Identity-centric data access
US20040088357A1 (en) Method and apparatus for applying personalized rules to e-mail messages at an e-mail server
EP1370991A1 (fr) Services a base de schema pour acces a des donnees a base d'identite
KR102208935B1 (ko) 데이터 교환을 위한 컨텍스트를 설정하는 http 프로토콜을 통한 메시징 api
EP1370965A1 (fr) Communication de service a service pour des services de reseau
US8849946B2 (en) System and method for hypertext transfer protocol publish and subscribe server
US20090254977A1 (en) Method and Apparatus for Communicating Information Between Devices
Thomson et al. Generic event delivery using http push
AU2007329843B2 (en) Offsite centralized data center providing client functionality
CA2452893C (fr) Methode et appareil pour stocker et gerer des contacts dans un systeme de collaboration reparti
Barnes et al. CCMP: a novel standard protocol for conference management in the XCON framework
Teymourian et al. Implementation of a novel semantic web middleware approach based on triplespaces
Subramanian et al. Peer-to-peer corporate resource sharing and distribution with mesh
IL159337A (en) Method and apparatus for storing and managing contacts in a distributed collaboration system
Thomson Internet-Draft Mozilla Intended status: Standards Track E. Damaggio Expires: January 23, 2017 B. Raymor, Ed. Microsoft
Sharp Application Protocols

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

17P Request for examination filed

Effective date: 20030825

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE TR

AX Request for extension of the european patent

Extension state: AL LT LV MK RO SI

A4 Supplementary search report drawn up and despatched

Effective date: 20070205

17Q First examination report despatched

Effective date: 20080502

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: MICROSOFT TECHNOLOGY LICENSING, LLC

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20161105