WO2020005596A1 - Actor model programming - Google Patents

Actor model programming Download PDF

Info

Publication number
WO2020005596A1
WO2020005596A1 PCT/US2019/037405 US2019037405W WO2020005596A1 WO 2020005596 A1 WO2020005596 A1 WO 2020005596A1 US 2019037405 W US2019037405 W US 2019037405W WO 2020005596 A1 WO2020005596 A1 WO 2020005596A1
Authority
WO
WIPO (PCT)
Prior art keywords
actor
actors
recipient
message
list
Prior art date
Application number
PCT/US2019/037405
Other languages
French (fr)
Inventor
Sylvan Wesley CLEBSCH
Matthew John Parkinson
Original Assignee
Microsoft Technology Licensing, Llc
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 GBGB1810645.0A external-priority patent/GB201810645D0/en
Application filed by Microsoft Technology Licensing, Llc filed Critical Microsoft Technology Licensing, Llc
Publication of WO2020005596A1 publication Critical patent/WO2020005596A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/54Interprogram communication
    • G06F9/546Message passing systems or structures, e.g. queues
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/52Program synchronisation; Mutual exclusion, e.g. by means of semaphores
    • G06F9/522Barrier synchronisation

Definitions

  • the actor model is a mathematical model of concurrent computation that treats "actors" as the universal primitives of concurrent computation.
  • An actor is a type of object (in the sense of object-orientated programming): it is a data structure comprising data and one or more functions for operating on that data. Actors may modify their own private state, but can only affect each other through messages between them. In response to a message that it receives, an actor can: make local decisions, create more actors, send more messages, and determine how to respond to the next received message. These tasks could be carried out in parallel. Recipients of messages are identified by an address, sometimes called a "mailing address". Thus an actor can only communicate with actors whose addresses it has. It can obtain those from a message it receives, or if the address is for an actor it has itself created. For example, electronic mail (e-mail) can be modelled as an actor system, with email accounts being modelled as actors and email addresses as actor addresses.
  • e-mail electronic mail
  • Threads are portions of code which can, at least for a time, be run independently of one another, and which can be run concurrently or in parallel with one another.
  • Concurrency refers to the case where threads are interleaved with one another through the same execution unit of a processor, thus providing an effectively simultaneous form of execution.
  • Parallelism refers to the case where threads are executed truly simultaneously through different parallel execution units. In a given program or set of programs, two or more threads of the program may be run concurrently and/or in parallel at any one time, depending on the resources of the system.
  • Threads can take the form of hardware threads or software threads.
  • the processor itself comprises hardware support for each thread, at least in the form a set of context registers for each of the threads that can be run concurrently or in parallel at any one time.
  • Each set of context registers stores a program state of a respective thread, such as the program counter and operands.
  • the concurrency or parallelism is achieved not (or not only) though hardware support for each thread, but rather the thread states are maintained at a higher level in software. Either way, in many cases the threads may be scheduled for execution by the operating system running on the computer system.
  • the threads in question may comprise threads of the operating system itself or threads of one or more applications run on the operating system.
  • Actor model programming has two key concepts: actors and messages.
  • An actor is allowed to process at most one message at a time, and a message can be handled using only the state of a single actor.
  • a given actor can only ever be manipulating its own internal state or processing as message from a single external actor at any one time. Only one actor can be affecting the state of a given actor at any one time. As an actor is only doing one thing at a time its internal state can be safely manipulated without requiring locks (instead, messages are queued until they can be processed).
  • the present invention provides a message that can request multiple actors to perform operations at the same time.
  • the restriction that an actor can be associated with at most one executing message at any given time is maintained, but now a message is handled using a single execution thread running with access to the state of one or more actors. This enables the system to be composed of much smaller actors and thus increase the concurrency or parallelism in the system.
  • a method of operating a computer according to an actor model comprises defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data.
  • a wrapped message is generated to be transmitted from a transmitting one of the actors to multiple recipient ones of the actors.
  • the wrapped message comprises at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list.
  • the index is initially set to indicate the first recipient actor in the list.
  • the wrapped message is transmitted from the transmitting actor to the first recipient actor in the list.
  • a method of programming a computer according to an actor model comprises defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data.
  • a transmitting one of the actors is programmed to generate a wrapped message destined for multiple recipient ones of the actors.
  • the wrapped message comprises at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list.
  • the index is initially set to indicate the first recipient actor in the list.
  • the transmitting actor is programmed to transmit the wrapped message from the transmitting actor to the first recipient actor in the list.
  • Each of the recipient actors, except the last in the list is programmed so as upon receiving the wrapped message, to advance the index and then forward the wrapped message to the next actor in the list as indicated by the advanced index.
  • the methods may comprise operations in accordance with any of the embodiments disclosed herein.
  • software embodied on computer-readable storage for enabling the programming methods disclosed herein, the software comprising a compiler, interpreter or library comprising one or more dedicated functions or commands enabling the programming of the generation and transmission of the wrapped message by the transmitting actor, and the forwarding of the wrapped message by the recipient actors.
  • a computer system comprising one or more processors and memory comprising one or more memory units arranged to store code arranged to run on the one or more processors, the code being configured so as when run to perform operations in accordance with any of the methods disclosed herein.
  • Figure 1 is a schematic block diagram of a computer system
  • Figure 2 schematically illustrates the flow of a wrapped message from a transmitting actor to recipients actors
  • Figure 3 also schematically shows the flow of a wrapped messages to recipient actors.
  • FIG. 1 shows a computer system 100 in accordance with embodiments disclosed herein.
  • the computer system 100 comprises: processing apparatus 102, memory 104, and one or more I/O devices 106.
  • the memory 104 stores an operating system 108 and one or more applications 110.
  • the operating system 108 is arranged to run on the processing apparatus 102, and the one or more applications 110 are arranged to run on the operating system 102 on the processing apparatus 102.
  • the operating system 108 is configured to schedule the application(s) 110 for execution and to mediate access to the memory 104 and I/O resources 106 by the application(s) 110.
  • the memory 104 is also operable to store data to be accessed or operated on by the operating system 108 and/or application(s) 110, or to store data resulting from operations performed by the operating system 108 and/or applications 110.
  • the memory 104 on which the operating system 108, application(s) 110 and data are stored may comprise one or more memory devices employing one or more memory media, e.g. electronic memory such as flash memory or a solid state drive (SSD); or a magnetic memory such as a hard-disk dive (HDD).
  • SSD solid state drive
  • HDD hard-disk dive
  • Each of the operating system 108, applications 110 and data may be stored on a single memory device or distributed across multiple memory devices.
  • the program code of the operating system 108 and applications 110 and data may be stored in different regions of the same one or more memory devices as the data, or on a different one or more memory devices than the data.
  • the operating system 108 may be stored in different regions of the same one or more memory devices as the applications 110, or a different one or more memory devices; and in the case of multiple applications 1 10 these may be stored in different regions of the same or more memory device, or some or all of them may be stored in different one or more memory devices than each other.
  • the memory device(s) may be internal to the same housing as the processing apparatus or be external to it, or may comprise a combination of internal and external devices.
  • the processing apparatus 102 is arranged to run multiple concurrent or parallel threads 112. These may be threads of the operating system 108 itself and/or threads of one or more of the applications 110.
  • the processing apparatus 102 on which the operating system 108 and application(s) 110 are run, including the threads 112, may comprises one or more processors comprising one or more cores. In the case of multiple processors these may be implemented on the same computer terminal or distributed across multiple computer units at multiple computer units at different locations, e.g. different chassis in the same data centre, different terminals on the same local area network, or different geographic locations networked together via a wide-area network such as the Internet.
  • the threads 112 may comprise: different threads on a same processor core, threads on different processor cores, threads on different cores on the same die or IC package, threads on different IC packages in the same board, threads on different boards connected together in a same data centre, threads on different devices connected together over a local-area and/or wide-area network (e.g. the Internet or a mobile cellular network such as a 3 GPP network), or any combination of these.
  • the operating system 108 may take the form of a distributed operating system.
  • the memory 104 may comprise multiple memory devices distributed across multiple locations. Where desired, distributed computing techniques in themselves are known in the art.
  • An actor is an object (often described as an“active object”) that is a unit of execution, and as such is stored in memory 104. An actor is run on a single thread at any given time and handles messages in sequence.
  • Figure 2 shows the flow of a wrapped message 202 sent from a transmitting actor A0 via recipient actors A2 and A3.
  • Each actor within the defined actor model takes the form of a data structure comprising data 204 and one or more functions 206 for operating on that data.
  • a wrapped message 202 is generated (e.g. by one of the actors) which will be transmitted from a transmitting actor A0. That is, the transmitting actor A0 is programmed to generate the wrapped message 202 destined for multiple recipient actors.
  • the wrapped message 202 includes one or more constituent messages 208, a sorted list 210 of recipient actors, and an index 212 indicating an entry in the sorted list 210.
  • the sorted list 210 may include one or more recipient actors.
  • the actors may be sorted based on a memory address of the actors.
  • the index 212 is initially set to indicate the first recipient actor in the sorted list 210.
  • the transmitting actor may be actor A0.
  • the transmitting actor may be a managing function.
  • the sorted list may comprise recipient actors A2, A3, and A4, listed in that order.
  • the index is initially set to indicate actor A2.
  • the index may in general indicate the recipient actor in any manner.
  • the sorted list does not have to comprise contiguously numbered actors, nor does the list have to comprise actors numbered in ascending order.
  • the sorted list may contain actors A2, A5, A3, A6, and A4.
  • the wrapped message is transmitted from the transmitting actor (e.g. actor A0) to the first recipient actor in the sorted list (e.g. actor A2). That is, the transmitting actor is programmed to transmit the wrapped message to the first recipient actor in the list.
  • the first recipient actor increments the index to the next entry in the list. For example, if the list of recipient actors is A2, A3, and A4, the first recipient actor A2 would increment the index to the next entry which indicates actor A3.
  • the first recipient actor forwards the wrapped message (with the incremented index) to the next recipient actor, i.e. the second recipient actor.
  • the second recipient actor e.g. actor A3 repeats the process of incrementing the index and forwarding the wrapped message to the next recipient actor. The process is repeated until the wrapped message is forwarded to and received by the last recipient actor in the sorted list.
  • the wrapped message, nf may take the form of (m, A*, i), where m is the constituent message(s), A* is the sorted list of recipient actors and i is the index.
  • the constituent message m may be a function f(A*) that operates over the recipient actors.
  • the transmitting actor may itself be a recipient actor.
  • the wrapped message may be transmitted from the transmitting actor to one or more recipient actors in turn, and one of those recipient actors may forward the wrapped message back to the transmitting actor.
  • the sorted list of recipient actors may comprise actors A2, A4, A6, A0, and A3.
  • the transmitting actor A0 receives the wrapped message from actor A6 and forwards it, along with an incremented index, to actor A3.
  • the transmitting actor may also be the last recipient actor.
  • Each actor, other than the last recipient actor in the list, is programmed such that as the wrapped message passes through said actor, that actor is prevented from processing any further inter-actor messages until all the actors in the list have processed the constituent message.
  • An inter-actor message may be any message transmitted from one actor to another actor. That is, if an actor receives a wrapped message, it cannot receive a second message (wrapped message or otherwise) until all of the recipient actors have performed any actions defined in the constituent message. I.e. all but the last of the recipient actors may each be locked from receiving any further inter-actor messages until all the actors in the list have processed the constituent message.
  • embodiments of the present invention introduces a new descheduled state, which effectively locks the actor from processing any further messages.
  • the actor cannot leave the descheduled state until the wrapped message is received by all of the intended recipients.
  • the descheduling is asynchronous in that each actor is descheduled at a different point in time, i.e. when it forwards the wrapped message to the next recipient actor.
  • the constituent message may be a single constituent message for each of the recipient actors. That is, each recipient actor is sent the same constituent message to process. In other examples, the same constituent message may be sent to some but not all of the recipient actors. For example, recipient actors A2 and A3 may be sent the same constituent message, whereas recipient actor A4 receives a different constituent message. That is, the constituent message comprises multiple constituent messages. Each of the multiple constituent messages may be intended (and therefore sent) to one or more of the recipient actors. For instance, each recipient actor may receive its own specific constituent message. E.g. recipient actors A2, A3 and A4 may receive constituent messages M2, M3 and M4 respectively.
  • An actor is an object (often described as an“active object”) that is a unit of execution, and as such is stored in memory 104.
  • An actor is run on a single thread at any given time and handles messages in sequence. Actors are initially placed in a queue to be claimed for processing by one of a plurality of threads. Each actor may only be claimed by (and therefore run on) a single thread at a time. Each thread may have its own respective single-producer multi-consumer queue (SPMCQ) of actors. The thread is itself the producer for its own queue, but any other thread may claim an actor from that thread’s queue by means of a message. I.e. each thread may claim an actor with pending messages from any other thread’ s SPMCQ.
  • SPMCQ single-producer multi-consumer queue
  • an actor is claimed by a thread other than the thread that produced the queue in which the actor is initially placed, the actor may be run on multiple threads (e.g. multiple operating system threads). Here, the actor still processes messages in sequence such that it follows a logical“thread of execution”. Different threads may claim different ones of the recipient actors. For example, thread Tl and T2 may claim recipient actors A2 and A4 respectively. A given thread may claim a plurality of recipient actors at the same time. In some examples, the transmitting actor may be claimed by a thread other than the thread(s) claiming the recipient actors.
  • the queue of actors for each thread is held in a part of the memory 104.
  • the queue may be maintained by the individual respective thread, or centrally by another function (e.g. a supervising thread of the operating system 108 or the application 110).
  • Previous actor model programming has made the manipulation of databases using actors difficult.
  • one application of actor model programming could be to create a database program.
  • a first table for instance may map identifiers to employees’ names and a second table may map the identifiers to the employees’ jobs.
  • Identifier 1 may map“Susan” to her job“Greengrocer”.
  • Actor A2 may be responsible for modifying the first table whilst actor A3 may be responsible for modifying the second table.
  • each actor operates on its respective tables in separate atomic transactions. This is adequate if one type of data entry is being changed, e.g. the names.
  • a wrapped message includes a sorted list of the recipient actors (A2 and A3) and an index initially set to the first recipient actor in the list (A2).
  • Actor A2 receives the wrapped message from a transmitter actor (e.g. actor A0) and is locked, i.e. prevented from receiving another message.
  • the wrapped message may comprise a first constituent message for actor A2.
  • the first constituent message may cause actor A2 to change Susan’s name to “Sue”.
  • Actor A2 advances the index to the next entry in the list, i.e. A3.
  • the wrapped message is forwarded to the next recipient actor (i.e. from A2 to A3).
  • the wrapped message may comprise a second constituent message for actor A3.
  • the second constituent message may cause actor A3 to change the job mapped to Susan’s ID to“Astronaut”.
  • actor A3 When the last recipient actor A3 receives the wrapped message it knows it is the last recipient as the index has been advanced to an entry corresponding to that actor (i.e. to A3). Now, when each actor has performed the operations of its respective data, each recipient actor is rescheduled (or released) so that it can receive and process other messages.
  • the sorted list ensures that actors can never deadlock the system. That is, no actor can receive more than one message.
  • Figure 3 illustrates the concept of when to reschedule an actor.
  • Figure 3 shows schematically the flow of a wrapped message 302 sent via multiple actors 304.
  • Each actor may be placed in a queue 306 to be claimed for processing by one of a plurality of threads 308.
  • threads 308 In normal actor runtimes, once a message is processed on an actor it can immediately deal with a new message.
  • a wrapped message i.e. a multi-message
  • actors are not rescheduled after they have been acquired, but instead they are left in a“limbo state” waiting for the wrapped message to propagate through all the required recipient actors.
  • the invention ensures that actors are always acquired in a defined order.
  • actor A2 processes its constituent message and forwards the wrapped message to the next recipient actor in the sorted list.
  • the next recipient actor is actor A6.
  • actor A2 advances the index to correspond with the next recipient actor in the sorted list.
  • actor A2 rather than actor A2 rescheduling itself, it instead gets the current thread to schedule a new actor (i.e. actor A6), and does not add itself to the work queue. This prevents actor A2 from being claimed by a thread.
  • the second thread processes the constituent message intended for actor A6, and can execute safely with the internal state of both actors A2 and A6.
  • recipient actor A6 Since the index corresponds to the last actor in the sorted list of recipient actors, recipient actor A6 knows that it is the last actor and does not need to forward the wrapped message to another actor.
  • recipient actor A6 At (E), once actors A2 and A6 have finished processing their constituent message(s), both actors are rescheduled in the work queue to be claimed by threads, and the system can start processing messages for a new actor (e.g. actor Al).

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer And Data Communications (AREA)

Abstract

A method of operating a computer according to an actor model, the method comprising: defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data; generating a wrapped message to be transmitted from a transmitting actor to multiple recipient actors, the wrapped message comprising at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list, the index initially being set to indicate the first recipient actor in the list; transmitting the wrapped message from the transmitting actor to the first recipient actor in the list; each of the recipient actors, except the last in the list, upon receiving the wrapped message, advancing the index and forwarding the wrapped message to the next actor in the list as indicated by the advanced index.

Description

ACTOR MODEL PROGRAMMING
Background
[0001] The actor model is a mathematical model of concurrent computation that treats "actors" as the universal primitives of concurrent computation. An actor is a type of object (in the sense of object-orientated programming): it is a data structure comprising data and one or more functions for operating on that data. Actors may modify their own private state, but can only affect each other through messages between them. In response to a message that it receives, an actor can: make local decisions, create more actors, send more messages, and determine how to respond to the next received message. These tasks could be carried out in parallel. Recipients of messages are identified by an address, sometimes called a "mailing address". Thus an actor can only communicate with actors whose addresses it has. It can obtain those from a message it receives, or if the address is for an actor it has itself created. For example, electronic mail (e-mail) can be modelled as an actor system, with email accounts being modelled as actors and email addresses as actor addresses.
[0002] What characterises an actor over simply any object of an object-orientated programming is that an actor is run on only one thread at any given time.
[0003] Threads are portions of code which can, at least for a time, be run independently of one another, and which can be run concurrently or in parallel with one another. Concurrency refers to the case where threads are interleaved with one another through the same execution unit of a processor, thus providing an effectively simultaneous form of execution. Parallelism refers to the case where threads are executed truly simultaneously through different parallel execution units. In a given program or set of programs, two or more threads of the program may be run concurrently and/or in parallel at any one time, depending on the resources of the system.
[0004] Threads can take the form of hardware threads or software threads. In the case of hardware threads, the processor itself comprises hardware support for each thread, at least in the form a set of context registers for each of the threads that can be run concurrently or in parallel at any one time. Each set of context registers stores a program state of a respective thread, such as the program counter and operands. In the case of software threads, the concurrency or parallelism is achieved not (or not only) though hardware support for each thread, but rather the thread states are maintained at a higher level in software. Either way, in many cases the threads may be scheduled for execution by the operating system running on the computer system. The threads in question may comprise threads of the operating system itself or threads of one or more applications run on the operating system.
Summary
[0005] Actor model programming has two key concepts: actors and messages. An actor is allowed to process at most one message at a time, and a message can be handled using only the state of a single actor. A given actor can only ever be manipulating its own internal state or processing as message from a single external actor at any one time. Only one actor can be affecting the state of a given actor at any one time. As an actor is only doing one thing at a time its internal state can be safely manipulated without requiring locks (instead, messages are queued until they can be processed).
[0006] However, this creates a restriction in terms of the types of programs that can be executed in an actor model. For example, it is very difficult (if not impossible) to build a transactional database using an actor model as the programmer is either required to implement locking on top of the model, or alternatively all of the data ends up in a single actor, which then harms the scaling of the application, as all data access effectively becomes single threaded. Implementing locking on top of the Actor model negates the benefits of the Actor model as it introduces the possibility of deadlocks.
[0007] To address such problems or similar, the present invention provides a message that can request multiple actors to perform operations at the same time. The restriction that an actor can be associated with at most one executing message at any given time is maintained, but now a message is handled using a single execution thread running with access to the state of one or more actors. This enables the system to be composed of much smaller actors and thus increase the concurrency or parallelism in the system.
[0008] According to one aspect disclosed herein, there is provided a method of operating a computer according to an actor model. The method comprises defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data. A wrapped message is generated to be transmitted from a transmitting one of the actors to multiple recipient ones of the actors. The wrapped message comprises at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list. The index is initially set to indicate the first recipient actor in the list. The wrapped message is transmitted from the transmitting actor to the first recipient actor in the list. Each of the recipient actors, except the last in the list, upon receiving the wrapped message, advances the index and then forwards the wrapped message to the next actor in the list as indicated by the advanced index. [0009] According to another aspect disclosed herein, there is provided a method of programming a computer according to an actor model. The method comprises defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data. A transmitting one of the actors is programmed to generate a wrapped message destined for multiple recipient ones of the actors. The wrapped message comprises at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list. The index is initially set to indicate the first recipient actor in the list. The transmitting actor is programmed to transmit the wrapped message from the transmitting actor to the first recipient actor in the list. Each of the recipient actors, except the last in the list, is programmed so as upon receiving the wrapped message, to advance the index and then forward the wrapped message to the next actor in the list as indicated by the advanced index.
[0010] In embodiments the methods may comprise operations in accordance with any of the embodiments disclosed herein.
[0011] According to another aspect disclosed herein, there is provided software embodied on computer-readable storage for enabling the programming methods disclosed herein, the software comprising a compiler, interpreter or library comprising one or more dedicated functions or commands enabling the programming of the generation and transmission of the wrapped message by the transmitting actor, and the forwarding of the wrapped message by the recipient actors.
[0012] According to another aspect disclosed herein there is provided software (e.g. an operating system) embodied on a computer-readable medium and configured so as when run on one or more processors to perform operations in accordance with any of the methods disclosed herein.
[0013] According to another aspect disclosed herein, there is provided a computer system comprising one or more processors and memory comprising one or more memory units arranged to store code arranged to run on the one or more processors, the code being configured so as when run to perform operations in accordance with any of the methods disclosed herein.
[0014] This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter. Nor is the claimed subject matter limited to implementations that solve any or all of the disadvantages noted herein.
Brief Description of the Drawings
[0015] To assist understanding of embodiments disclosed herein and to illustrate how such embodiments may be put into effect, reference is made, by way of example only, to the accompanying drawings in which:
[0016] Figure 1 is a schematic block diagram of a computer system,
[0017] Figure 2 schematically illustrates the flow of a wrapped message from a transmitting actor to recipients actors, and
[0018] Figure 3 also schematically shows the flow of a wrapped messages to recipient actors.
Detailed Description of Embodiments
[0019] Figure 1 shows a computer system 100 in accordance with embodiments disclosed herein. The computer system 100 comprises: processing apparatus 102, memory 104, and one or more I/O devices 106. The memory 104 stores an operating system 108 and one or more applications 110. The operating system 108 is arranged to run on the processing apparatus 102, and the one or more applications 110 are arranged to run on the operating system 102 on the processing apparatus 102. The operating system 108 is configured to schedule the application(s) 110 for execution and to mediate access to the memory 104 and I/O resources 106 by the application(s) 110.
[0020] The memory 104 is also operable to store data to be accessed or operated on by the operating system 108 and/or application(s) 110, or to store data resulting from operations performed by the operating system 108 and/or applications 110. The memory 104 on which the operating system 108, application(s) 110 and data are stored may comprise one or more memory devices employing one or more memory media, e.g. electronic memory such as flash memory or a solid state drive (SSD); or a magnetic memory such as a hard-disk dive (HDD). Each of the operating system 108, applications 110 and data may be stored on a single memory device or distributed across multiple memory devices. The program code of the operating system 108 and applications 110 and data may be stored in different regions of the same one or more memory devices as the data, or on a different one or more memory devices than the data. Similarly, the operating system 108 may be stored in different regions of the same one or more memory devices as the applications 110, or a different one or more memory devices; and in the case of multiple applications 1 10 these may be stored in different regions of the same or more memory device, or some or all of them may be stored in different one or more memory devices than each other. The memory device(s) may be internal to the same housing as the processing apparatus or be external to it, or may comprise a combination of internal and external devices.
[0021] The processing apparatus 102 is arranged to run multiple concurrent or parallel threads 112. These may be threads of the operating system 108 itself and/or threads of one or more of the applications 110. The processing apparatus 102 on which the operating system 108 and application(s) 110 are run, including the threads 112, may comprises one or more processors comprising one or more cores. In the case of multiple processors these may be implemented on the same computer terminal or distributed across multiple computer units at multiple computer units at different locations, e.g. different chassis in the same data centre, different terminals on the same local area network, or different geographic locations networked together via a wide-area network such as the Internet.
[0022] For instance, the threads 112 may comprise: different threads on a same processor core, threads on different processor cores, threads on different cores on the same die or IC package, threads on different IC packages in the same board, threads on different boards connected together in a same data centre, threads on different devices connected together over a local-area and/or wide-area network (e.g. the Internet or a mobile cellular network such as a 3 GPP network), or any combination of these. Note therefore that in some embodiments the operating system 108 may take the form of a distributed operating system. Also, in embodiments the memory 104 may comprise multiple memory devices distributed across multiple locations. Where desired, distributed computing techniques in themselves are known in the art.
[0023] An actor is an object (often described as an“active object”) that is a unit of execution, and as such is stored in memory 104. An actor is run on a single thread at any given time and handles messages in sequence.
[0024] Figure 2 shows the flow of a wrapped message 202 sent from a transmitting actor A0 via recipient actors A2 and A3.
[0025] Each actor within the defined actor model takes the form of a data structure comprising data 204 and one or more functions 206 for operating on that data. A wrapped message 202 is generated (e.g. by one of the actors) which will be transmitted from a transmitting actor A0. That is, the transmitting actor A0 is programmed to generate the wrapped message 202 destined for multiple recipient actors. The wrapped message 202 includes one or more constituent messages 208, a sorted list 210 of recipient actors, and an index 212 indicating an entry in the sorted list 210. The sorted list 210 may include one or more recipient actors. The actors may be sorted based on a memory address of the actors. The index 212 is initially set to indicate the first recipient actor in the sorted list 210.
[0026] For example, the transmitting actor may be actor A0. The transmitting actor may be a managing function. The sorted list may comprise recipient actors A2, A3, and A4, listed in that order. In this example, the index is initially set to indicate actor A2. Note that the index may in general indicate the recipient actor in any manner. Also note that the sorted list does not have to comprise contiguously numbered actors, nor does the list have to comprise actors numbered in ascending order. For example, the sorted list may contain actors A2, A5, A3, A6, and A4.
[0027] The wrapped message is transmitted from the transmitting actor (e.g. actor A0) to the first recipient actor in the sorted list (e.g. actor A2). That is, the transmitting actor is programmed to transmit the wrapped message to the first recipient actor in the list. When the first recipient actor receives the wrapped message, the first recipient actor increments the index to the next entry in the list. For example, if the list of recipient actors is A2, A3, and A4, the first recipient actor A2 would increment the index to the next entry which indicates actor A3. The first recipient actor forwards the wrapped message (with the incremented index) to the next recipient actor, i.e. the second recipient actor. The second recipient actor (e.g. actor A3) repeats the process of incrementing the index and forwarding the wrapped message to the next recipient actor. The process is repeated until the wrapped message is forwarded to and received by the last recipient actor in the sorted list.
[0028] The wrapped message, nf, may take the form of (m, A*, i), where m is the constituent message(s), A* is the sorted list of recipient actors and i is the index. The constituent message m may be a function f(A*) that operates over the recipient actors.
[0029] The transmitting actor may itself be a recipient actor. For example, the wrapped message may be transmitted from the transmitting actor to one or more recipient actors in turn, and one of those recipient actors may forward the wrapped message back to the transmitting actor. For example, if the transmitting actor is actor A0, the sorted list of recipient actors may comprise actors A2, A4, A6, A0, and A3. Here, the transmitting actor A0 receives the wrapped message from actor A6 and forwards it, along with an incremented index, to actor A3. The transmitting actor may also be the last recipient actor.
[0030] Each actor, other than the last recipient actor in the list, is programmed such that as the wrapped message passes through said actor, that actor is prevented from processing any further inter-actor messages until all the actors in the list have processed the constituent message. An inter-actor message may be any message transmitted from one actor to another actor. That is, if an actor receives a wrapped message, it cannot receive a second message (wrapped message or otherwise) until all of the recipient actors have performed any actions defined in the constituent message. I.e. all but the last of the recipient actors may each be locked from receiving any further inter-actor messages until all the actors in the list have processed the constituent message.
[0031] Unlike prior implementations of the actor model in which an actor processes a message upon receiving said message, embodiments of the present invention introduces a new descheduled state, which effectively locks the actor from processing any further messages. The actor cannot leave the descheduled state until the wrapped message is received by all of the intended recipients. The descheduling is asynchronous in that each actor is descheduled at a different point in time, i.e. when it forwards the wrapped message to the next recipient actor.
[0032] In some examples, the constituent message may be a single constituent message for each of the recipient actors. That is, each recipient actor is sent the same constituent message to process. In other examples, the same constituent message may be sent to some but not all of the recipient actors. For example, recipient actors A2 and A3 may be sent the same constituent message, whereas recipient actor A4 receives a different constituent message. That is, the constituent message comprises multiple constituent messages. Each of the multiple constituent messages may be intended (and therefore sent) to one or more of the recipient actors. For instance, each recipient actor may receive its own specific constituent message. E.g. recipient actors A2, A3 and A4 may receive constituent messages M2, M3 and M4 respectively.
[0033] An actor is an object (often described as an“active object”) that is a unit of execution, and as such is stored in memory 104. An actor is run on a single thread at any given time and handles messages in sequence. Actors are initially placed in a queue to be claimed for processing by one of a plurality of threads. Each actor may only be claimed by (and therefore run on) a single thread at a time. Each thread may have its own respective single-producer multi-consumer queue (SPMCQ) of actors. The thread is itself the producer for its own queue, but any other thread may claim an actor from that thread’s queue by means of a message. I.e. each thread may claim an actor with pending messages from any other thread’ s SPMCQ. If an actor is claimed by a thread other than the thread that produced the queue in which the actor is initially placed, the actor may be run on multiple threads (e.g. multiple operating system threads). Here, the actor still processes messages in sequence such that it follows a logical“thread of execution”. Different threads may claim different ones of the recipient actors. For example, thread Tl and T2 may claim recipient actors A2 and A4 respectively. A given thread may claim a plurality of recipient actors at the same time. In some examples, the transmitting actor may be claimed by a thread other than the thread(s) claiming the recipient actors.
[0034] The queue of actors for each thread is held in a part of the memory 104. The queue may be maintained by the individual respective thread, or centrally by another function (e.g. a supervising thread of the operating system 108 or the application 110).
[0035] Previous actor model programming has made the manipulation of databases using actors difficult. For example, one application of actor model programming could be to create a database program. In such a case, a first table for instance may map identifiers to employees’ names and a second table may map the identifiers to the employees’ jobs. Identifier 1 may map“Susan” to her job“Greengrocer”. Actor A2 may be responsible for modifying the first table whilst actor A3 may be responsible for modifying the second table. In previous models, each actor operates on its respective tables in separate atomic transactions. This is adequate if one type of data entry is being changed, e.g. the names. However, it would be desirable to perform two different simultaneous changes of state in an actor model in a single atomic transaction, e.g. changing name and job title. To do this, a single message is generated that computes over the states of multiple actors (actors A2 and A3).
[0036] A wrapped message includes a sorted list of the recipient actors (A2 and A3) and an index initially set to the first recipient actor in the list (A2). Actor A2 receives the wrapped message from a transmitter actor (e.g. actor A0) and is locked, i.e. prevented from receiving another message. The wrapped message may comprise a first constituent message for actor A2. The first constituent message may cause actor A2 to change Susan’s name to “Sue”. Actor A2 advances the index to the next entry in the list, i.e. A3. The wrapped message is forwarded to the next recipient actor (i.e. from A2 to A3). The wrapped message may comprise a second constituent message for actor A3. The second constituent message may cause actor A3 to change the job mapped to Susan’s ID to“Astronaut”. When the last recipient actor A3 receives the wrapped message it knows it is the last recipient as the index has been advanced to an entry corresponding to that actor (i.e. to A3). Now, when each actor has performed the operations of its respective data, each recipient actor is rescheduled (or released) so that it can receive and process other messages. The sorted list ensures that actors can never deadlock the system. That is, no actor can receive more than one message.
[0037] Figure 3 illustrates the concept of when to reschedule an actor. Figure 3 shows schematically the flow of a wrapped message 302 sent via multiple actors 304. Each actor may be placed in a queue 306 to be claimed for processing by one of a plurality of threads 308. In normal actor runtimes, once a message is processed on an actor it can immediately deal with a new message. Here, when processing a wrapped message (i.e. a multi-message), actors are not rescheduled after they have been acquired, but instead they are left in a“limbo state” waiting for the wrapped message to propagate through all the required recipient actors. To avoid deadlock, the invention ensures that actors are always acquired in a defined order.
[0038] In the example of Figure 3, at (A) transmitting actor A3 is attempting to send a wrapped message M’, that will acquire recipient actors A2 and A6. Let us assume that A2 is the first in the order of acquisition. The wrapped message M’ includes a sorted list comprising recipient actors A2 and A6. The wrapped message also includes an index set to indicate the first recipient actor A2. At (B) the wrapped message has been transmitted to recipient actor A2 and therefore actor A3 may reschedule itself on the work queue. A3 can reschedule as it has not received an inter-actor message. Recipient actor A2 is taken out of the work queue (i.e. claimed by thread 1) to execute a constituent message of the wrapped message. At (C), actor A2 processes its constituent message and forwards the wrapped message to the next recipient actor in the sorted list. In this example, the next recipient actor is actor A6. Before forwarding the wrapped message to actor A6, actor A2 advances the index to correspond with the next recipient actor in the sorted list. At (D), rather than actor A2 rescheduling itself, it instead gets the current thread to schedule a new actor (i.e. actor A6), and does not add itself to the work queue. This prevents actor A2 from being claimed by a thread. Now, the second thread processes the constituent message intended for actor A6, and can execute safely with the internal state of both actors A2 and A6. Since the index corresponds to the last actor in the sorted list of recipient actors, recipient actor A6 knows that it is the last actor and does not need to forward the wrapped message to another actor. At (E), once actors A2 and A6 have finished processing their constituent message(s), both actors are rescheduled in the work queue to be claimed by threads, and the system can start processing messages for a new actor (e.g. actor Al).
[0039] It will be appreciated that the above embodiments have been described by way of example only. Other applications or variants of the disclosed techniques may become apparent to a person skilled in the art given the disclosure herein. The scope of the present disclosure is not limited by the above-described embodiments but only by the accompanying claims.

Claims

Claims
1. A method of operating a computer according to an actor model, the method comprising:
defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data; generating a wrapped message to be transmitted from a transmitting one of the actors to multiple recipient ones of the actors, the wrapped message comprising at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list, the index initially being set to indicate the first recipient actor in the list;
transmitting the wrapped message from the transmitting actor to the first recipient actor in the list;
each of the recipient actors, except the last in the list, upon receiving the wrapped message, advancing the index and then forwarding the wrapped message to the next actor in the list as indicated by the advanced index.
2. The method of claim 1 , wherein all but the last of the recipient actors are each locked from receiving any further inter-actor messages until all the actors in the list have processed the constituent message.
3. The method of claim 1 or claim 2, wherein the actors are initially placed in a queue to be claimed for processing by one of a plurality of threads, each actor being claimed by only one of the threads, at least at any one time.
4. The method of claim 3, including enabling different threads to claim different ones of the recipient actors at a given time.
5. The method of claim 4, including enabling a different thread to claim the transmitting actor than the thread or threads claiming the recipient actors at a given time.
6. The method of any of claims 3 to 5, wherein the queue is a single producer, multiple consumer queue.
7. The method of any of claims 3 to 6, wherein the threads are operating system threads.
8. The method of any of claims 3 to 6, wherein the threads comprise multiple threads of a database program.
9. The method of any preceding claim, wherein the at least one constituent message comprises multiple constituent messages, one for each of some or all of the multiple recipient actors.
10. The method of any preceding claim, wherein the at least one constituent message comprises a same constituent message for each of some or all of the multiple recipient actors.
11. The method of any preceding claim, wherein the transmitting actor is also a recipient actor of the wrapped message.
12. A computer system programmed to perform the method of any preceding claim.
13. A method of programming a computer according to an actor model, the method comprising:
defining a plurality of actors, each taking form of a data structure comprising respective data and one or more respective functions for operating on the respective data; programming a transmitting one of the actors to generate a wrapped message destined for multiple recipient ones of the actors, the wrapped message comprising at least one constituent message, a sorted list of the recipient actors, and an index indicating an entry in the list, the index initially being set to indicate the first recipient actor in the list;
programming the transmitting actor to transmit the wrapped message from the transmitting actor to the first recipient actor in the list;
programming each of the recipient actors, except the last in the list, so as upon receiving the wrapped message, to advance the index and then forward the wrapped message to the next actor in the list as indicated by the advanced index.
14. The method of claim 13, further comprising programming the computer to perform the method of any of claims 2 to 11.
15. Software embodied on computer-readable storage for enabling the programming method of claim 13 or 14, the software comprising a compiler, interpreter or library comprising one or more dedicated functions or commands enabling the programming of the generation and transmission of the wrapped message by the transmitting actor, and the forwarding of the wrapped message by the recipient actors.
PCT/US2019/037405 2018-06-28 2019-06-17 Actor model programming WO2020005596A1 (en)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
GBGB1810645.0A GB201810645D0 (en) 2018-06-28 2018-06-28 Generalized actor model programming
GB1810645.0 2018-06-28
US16/043,142 US10768902B2 (en) 2018-06-28 2018-07-23 Actor model programming
US16/043,142 2018-07-23

Publications (1)

Publication Number Publication Date
WO2020005596A1 true WO2020005596A1 (en) 2020-01-02

Family

ID=67220858

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2019/037405 WO2020005596A1 (en) 2018-06-28 2019-06-17 Actor model programming

Country Status (1)

Country Link
WO (1) WO2020005596A1 (en)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0178235A1 (en) * 1984-09-20 1986-04-16 Institut National De Recherche En Informatique Et En Automatique Electronic method and device for the distributed execution of an activity between a plurality of different sites

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0178235A1 (en) * 1984-09-20 1986-04-16 Institut National De Recherche En Informatique Et En Automatique Electronic method and device for the distributed execution of an activity between a plurality of different sites

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
SHAMS M IMAM ET AL: "Savina - An Actor Benchmark Suite", PROGRAMMING BASED ON ACTORS AGENTS & DECENTRALIZED CONTROL, ACM, 2 PENN PLAZA, SUITE 701 NEW YORK NY 10121-0701 USA, 20 October 2014 (2014-10-20), pages 67 - 80, XP058063013, ISBN: 978-1-4503-2189-1, DOI: 10.1145/2687357.2687368 *
ZACHARI DICHEV: "Improving Akka Dispatchers", 20 July 2017 (2017-07-20), XP055616483, Retrieved from the Internet <URL:https://scalac.io/improving-akka-dispatchers/> [retrieved on 20190829] *

Similar Documents

Publication Publication Date Title
JP5650952B2 (en) Multi-core / thread workgroup calculation scheduler
US8595735B2 (en) Holistic task scheduling for distributed computing
US8997101B2 (en) Scalable thread locking with customizable spinning
CN107113341B (en) System for high throughput processing of transactions in a distributed relational database management system for data partitioning
US20150040140A1 (en) Consuming Ordered Streams of Messages in a Message Oriented Middleware
CN110597606B (en) Cache-friendly user-level thread scheduling method
US20150205633A1 (en) Task management in single-threaded environments
US9052972B2 (en) Determining the processing order of a plurality of events
WO2019047441A1 (en) Communication optimization method and system
Chrysanthakopoulos et al. An asynchronous messaging library for c
CN107077390A (en) A kind of task processing method and network interface card
KR20080005523A (en) Multithreading with concurrency domains
US11507386B2 (en) Booting tiles of processing units
CN109154897A (en) Reconfigurable distributed treatment
JP2012160013A (en) Data analysis and machine learning processing unit, method, and program
US10768902B2 (en) Actor model programming
CN111767995A (en) Operation method, device and related product
US8356300B2 (en) Multi-threaded processes for opening and saving documents
WO2020005596A1 (en) Actor model programming
Crawford et al. OpenCL command-buffer extension: Design and implementation
Duy et al. Enhanced virtual release advancing algorithm for real-time task scheduling
US11947517B1 (en) Stream-based transaction processing
JP2008159019A (en) Execution processing device of program
JP3931883B2 (en) Job scheduling method, computer system, and job scheduling program
Briot et al. Object-oriented concurrent programming

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19737617

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19737617

Country of ref document: EP

Kind code of ref document: A1