Overview of the OMG Data Distribution Service Douglas C. Schmidt & Jeff Parsons {schmidt, parsons}@dre. vanderbilt. edu http: //www. dre. vanderbilt. edu/~schmidt/ Professor of EECS Vanderbilt University Nashville, Tennessee
Tutorial on DDS Douglas C. Schmidt Past R&D Successes: Platform-centric Systems From this design paradigm… Nav Air Frame WTS AP FLIR GPS IFF Cyclic Exec Legacy Do. D systems are designed to be: • Stovepiped • Proprietary • Brittle & non-adaptive • Expensive to develop & evolve • Vulnerable 2 Problem: Small changes can break nearly anything & everything
Tutorial on DDS Douglas C. Schmidt Past R&D Successes: Platform-centric Systems …and this operation paradigm… Utility “Curve” Utility Real-time Qo. S requirements for platform-centric Do. D systems: • Ensure end-to-end Qo. S, e. g. , • Minimize latency, jitter, & footprint • Bound priority inversions • Allocate & manage resources statically “Broken” “Works” Resources “Harder” Requirements 3 Problem: Lack of any resource can break nearly anything & everything
Tutorial on DDS Douglas C. Schmidt Past R&D Successes: Network-centric Systems …to this design paradigm… Air Frame AP Event Channel GPS Nav WTS Replication Service IFF FLIR Object Request Broker Today’s leading-edge Do. D systems are designed to be: • Layered & componentized • More standard & COTS • Robust to expected failures & adaptive for non-critical tasks • Expensive to evolve & retarget • Vulnerable Problem: Unanticipated changes can break many things 4
Tutorial on DDS Douglas C. Schmidt Past R&D Successes: Network-centric Systems …and this operational paradigm… Applications Interceptor Sys Cond Middleware Workload & Replicas Local Resource Managers Connections & priority bands } Sys Cond Mechanism & Property Managers Qo. S Contracts Interceptor Middleware { Workload & Replicas Connections & priority bands CPU & memory Network latency & bandwidth Local Resource Managers Network latency & bandwidth Endsystem Adaptive Real-time Middleware 5
Tutorial on DDS Douglas C. Schmidt Past R&D Successes: Network-centric Systems …and this operational paradigm… Utility Desired Utility Curve Problem: Network-centricity is an afterthought in today’s systems “Working Range” Resources “Softer” Requirements 6
Tutorial on DDS Douglas C. Schmidt Case Study: Qo. S-enabled Publish/Subscribe Technologies for Tactical Information Management Coordination Of Multiple UAVs Feedback & Control Dynamic Mission Replanning Image Processing & Tracking DARPA PCES Capstone demo, 4/14/05, White Sands Missile Range 8
Tutorial on DDS Douglas C. Schmidt Case Study: Qo. S-enabled Publish/Subscribe Technologies for Tactical Information Management Coordination Of Multiple UAVs Feedback & Control Dynamic Mission Replanning Image Processing & Tracking Asynchronous Event Handling Physical Memory Access Memory Management Asynchronous Transfer of Control Synchronization Scheduling Modeling Tools Real-time JVMs Model Checking Real-time ORBs 9 Aspect Languages
Tutorial on DDS Douglas C. Schmidt Limitations with Demo’d PCES Information Management Technologies • Shooter information management was Real-time Info to Cockpit based on platform-centric Real-time CORBA & Real-time CORBA Event Real-time Event Service • Well-suited for point-to-point & static Object Request pub/sub command processing over Broker wireline networks Tactical • e. g. , statically provisioned Qo. S Network & RTOS policies • Poorly suited for dynamic pub/sub info dissemination to multiple nodes in MANETs • e. g. , too many layers, excessive Problem: Net-centricity is afterthought in platform-centric technologies time/space overhead, inflexible Qo. S 10
Tutorial on DDS Douglas C. Schmidt Limitations with Demo’d PCES Information Management Technologies Track Processing • C 2 & C 4 ISR information management based on J 2 EE Java Messaging & Java Messaging Service (JMS) J 2 EE • Best suited for operational Middleware enterprise environments Enterprise • e. g. , large data centers Network & OS connect via wireline networks • Poorly suited for tactical environments • e. g. , lack of Qo. S policies & RTOS integration, extremely Problem: Enterprise technologies are ill suited for tactical systems 11 high time/space
Tutorial on DDS Douglas C. Schmidt Our R&D Goal: Evaluate Qo. S-enabled Info Brokering for Tactical Systems • Solutions must function properly where • Communication bandwidth is limited/variable • Connectivity is intermittent • Connections are noisy • Processing & storage capacity are limited • Power & weight limits affect usage patterns • Unanticipated workflows are common • Dynamic network topology & membership changes are frequent • Ideally, solutions should be COTS, standard, & integrate with heterogeneous legacy systems Goal is “better than best-effort, ” subject to resource constraints… 12
Tutorial on DDS Douglas C. Schmidt Overview of the Data Distribution Service (DDS) • DDS is an highly efficient OMG pub/sub standard • e. g. , fewer layers, less overhead Topic R Data Writer R Data Reader R Publisher Subscriber RT Info to Cockpit & Track Processing DDS Pub/Sub Infrastructure Tactical Network & RTOS 13
Tutorial on DDS Douglas C. Schmidt Overview of the Data Distribution Service (DDS) • DDS is an highly efficient OMG pub/sub standard • e. g. , fewer layers, less overhead • DDS provides meta-events for detecting dynamic changes Topic NEW TOPIC R Data Writer R Data Reader R NEW SUBSCRIBER Publisher NEW PUBLISHER 14 Subscriber
Tutorial on DDS Douglas C. Schmidt Overview of the Data Distribution Service (DDS) • DDS is an highly efficient OMG pub/sub standard • e. g. , fewer layers, less overhead • DDS provides meta-events for detecting dynamic changes • DDS provides policies for specifying many Qo. S requirements of tactical information management systems, e. g. , • Establish contracts that precisely specify a wide variety of Qo. S policies at multiple system layers Topic HISTORY RESOURCE LIMITS R Data Writer R S 1 Data Reader R S 2 S 3 S 4 S 5 Publisher Subscriber S 6 S 7 LATENCY S 7 X S 7 RELIABILITY 15 S 6 S 5 S 4 S 3 S 2 S 1 COHERENCY
Tutorial on DDS Douglas C. Schmidt Overview of the Data Distribution Service (DDS) • DDS is an highly efficient OMG pub/sub standard • e. g. , fewer layers, less overhead • DDS provides meta-events for detecting dynamic changes • DDS provides policies for specifying many Qo. S requirements of tactical information management systems, e. g. , • Establish contracts that precisely specify a wide variety of Qo. S policies at multiple system layers • Move processing closer to data DESTINATION Topic FILTER R Data Writer R S 1 S 2 SOURCE FILTER S 3 Data Reader R S 4 S 5 Publisher Subscriber S 6 S 7 TIME-BASED FILTER 16
Tutorial on DDS Douglas C. Schmidt DDS Architectural Elements Data-Centric Publish-Subscribe Data Local Reconstruction Layer (DLRL) (DCPS) – The upper layer APIs that define how to – The lower layer APIs apps can use to build a local object cache so apps can exchange topic data with other DDSaccess topic data as if it were local enabled apps according to designated Qo. S policies • DDS spec only defines policies & interfaces between application & service • Doesn’t address protocols & techniques for different actors implementing the service • Doesn’t address management of internal DDS resources 18
Tutorial on DDS Douglas C. Schmidt DDS Application Architecture { The Application DLRL DCPS Communication 19 Application DLRL
Tutorial on DDS Douglas C. Schmidt DDS Domains & Domain Participants • The domain is the basic construct used to bind individual applications together for communication Domain 3 Domain 2 • Like a VPN 2 1 Node 3 1 Node Domain. Participant Domain 1 20 Node
Tutorial on DDS Douglas C. Schmidt DCPS Entities include • Data can be accessed in two ways – Topics – Wait-based (synchronous calls) • Typed data – Listener-based (asynchronous callbacks) – Publishers • Sophisticated support for filtering • Contain Data. Writers – e. g. , Topic, Content-Filtered. Topic, or Multi. Topic – Subscribers • Configurable via (many) Qo. S policies • Contain Data. Readers – Domain. Participants Topic • Entry points Data Reader Domain Participant Subscriber Data Writer Topic Data Reader Publisher Data Domain 21 Data Reader Subscriber Data Writer Publisher
Tutorial on DDS Douglas C. Schmidt Qo. S Policies Supported by DDS • DCPS entities (i. e. , topics, data readers/writers) configurable via Qo. S policies • Qo. S tailored to data distribution in tactical information systems – DEADLINE – RELIABILITY (BEST_EFFORT, RELIABLE) • Establishes contract regarding rate at which periodic data is refreshed • Enables use of real-time transports for data – LATENCY_BUDGET – HISTORY (KEEP_LAST, KEEP_ALL) • Establishes guidelines for acceptable end-to-end delays • Controls which (of multiple) data values are delivered – TIME_BASED_FILTER • Mediates exchanges between slow consumers & fast producers – DURABILITY (VOLATILE, TRANSIENT, PERSISTENT) • Determines if data outlives time when they are written – RESOURCE_LIMITS • Controls resources utilized by service – … and many more … • Request/offered compatibility checked by DDS 22
Tutorial on DDS Douglas C. Schmidt Best Effort Reliability Qo. S in DDS Qo. S Reliability = BEST_EFFORT Subscriber Publisher Subscriber Notification of new data objects deadline timeout time-based filter no notification • Very predictable • Data is sent without reply • Publishers and subscribers match and obey Qo. S Deadline policy settings • Time-based Filter Qo. S policy gives bandwidth control 23 time
Tutorial on DDS Douglas C. Schmidt Reliable Qo. S in DDS Qo. S Reliability = RELIABLE Topic history Data Writer R R S 1 Data Reader R S 2 S 3 S 4 S 5 Publisher Subscriber S 6 S 7 S 6 S 5 S 4 S 3 S 2 Ordered instance delivery is guaranteed 24 S 1
Tutorial on DDS Douglas C. Schmidt Tradeoff Between Reliability & Determinism Qo. S Reliability = BEST_EFFORT • Can’t have reliability and determinism. • Best effort mode for streaming data. X • No retries of dropped packets. • Minimizes latency. • Reliable mode for commands & events. • Retry dropped packets until timeout. • Every packet received in order. Qo. S Reliability = RELIABLE X X • Speculative cache mechanism. • DDS reliability is tunable. • Can be adjusted per subscription. 25
Tutorial on DDS Douglas C. Schmidt All Qo. S Policies in DDS • Deadline • Partition • Destination Order • Presentation • Durability • Reader Data Lifecycle • Entity Factory • Reliability • Group Data • Resource Limits • History • Time-Based Filter • Latency Budget • Topic Data • Lifespan • Transport Priority • Liveliness • User Data • Ownership • Writer Data Lifecycle • Ownership Strength 26
Tutorial on DDS Douglas C. Schmidt Single vs. Multiple Domain Systems 27
Tutorial on DDS Douglas C. Schmidt Data Writers & Publishers • Data Writers are the primary access point for an application to publish data into a DDS data domain • The Publisher entity is a container to group together individual Data Writers • User applications – Associate Data Writers with Topics – Provide data to Data Writers – Data is typically defined using OMG IDL • It can therefore be as strongly or weakly typed as you desire 28
Tutorial on DDS Douglas C. Schmidt Data Readers & Subscribers • A Data Reader is the primary access point for an application to access data that has been received by a Subscriber • Subscriber is used to group together Data Readers • Subscribers & Data Readers can have their own Qo. S policies • User applications – Associate Data Readers with Topics – Receive data from Data Readers using Listeners (async) or Wait-Sets (sync) 29
Tutorial on DDS Douglas C. Schmidt Types & Keys • A DDS Type is represented by a collection of data items. – e. g. “IDL struct” in the CORBA PSM struct Analog. Sensor { string sensor_id; // key float value; // other sensor data }; • A subset of the collection is designated as the Key. – The Key can be indicated by IDL annotation in CORBA PSM, e. g. , #pragma DDS_KEY Analog. Sensor: : sensor_id • It’s manipulated by means of automatically-generated typed interfaces. – IDL compiler may be used in CORBA PSM implementation • A Type is associated with generated code: – Analog. Sensor. Data. Writer // write values – Analog. Sensor. Data. Reader // read values – Analog. Sensor. Type // can register itself with Domain 30
Tutorial on DDS Douglas C. Schmidt Topics • A DDS Topic is the connection between publishers & subscribers. • A Topic is comprised of a Name and a Type. – Name must be unique in the Domain. – Many Topics can have the same Type. • Provision is made for contentbased subscriptions. – Multi. Topics correspond to SQL join – Content-Filtered Topics correspond to SQL select. Domain ID 35 Topic name “My. Sensor” Type name “Analog. Sensor” string float 31 sensor_id value [ key ]
Tutorial on DDS Douglas C. Schmidt Topic-Based Publish/Subscribe Pressure Temperature • Data. Writer is bound (at creation time) to a single Topic • Data. Reader is bound (at creation time) with one or more topics (Topic, Content. Filtered. Topic, or Multi. Topic) – Content. Filtered. Topic & Multi. Topic provide means for content-based subscriptions & “joins”, respectively 32
Tutorial on DDS Douglas C. Schmidt Subscription = Topic + Data. Reader application Data Reader Qo. S Topic 1 Topic 2 Topic n Qo. S subscriber 33
Tutorial on DDS Douglas C. Schmidt Content-based Subscriptions • Content. Filtered. Topic (like a DB-selection) – Enables subscriber to only receive data-updates whose value verifies a condition. – e. g. subscribe to “Pressure” of type Analog. Data – where “value > 200” • Multi. Topic (like a DB-join operation) – Enables subscription to multiple topics & re-arrangement of the dataformat – e. g. combine subscription to “Pressure” & “Temperature” & rearrange the data into a new type: struct { float pres; float temp; }; 34
Tutorial on DDS Douglas C. Schmidt DDS Content-Filtered Topics Topic Instances in Domain Instance 1 Value = 249 Instance 2 Value = 230 Instance 3 Value = 275 Instance 4 Value = 262 Instance 5 Value = 258 Filter Expression: Value > 260 Instance 6 Value = 261 Expression Params Instance 7 Value = 259 Content-Filtered Topic Filter Expression and Expression Params determine which Topic instances the Subscriber receives. 35
Tutorial on DDS Douglas C. Schmidt DDS Multi. Topic Subscriptions Topic Multi. Topic Domain Participant Data Reader Subscriber Data Reader Subscriber Multi. Topics can combine, filter, and rearrange data from multiple Topics. 36
Tutorial on DDS Douglas C. Schmidt Example: Create Domain Participant • Domain. Participant object acts as factory for Publisher, Subscriber, Topic and Multi. Topic entity objects // used to identify the participant Domain. Id_t domain_id =. . . ; // get the singleton factory instance Domain. Participant. Factory_var dpf = Domain. Participant. Factory: : get_instance (); // create domain participant from factory Domain. Participant_var dp = dpf->create_participant (domain_id, PARTICIPANT_QOS_DEFAULT, NULL); 37
Tutorial on DDS Douglas C. Schmidt Example: Create Topic …… // register the data type associated with the topic Foo. Data. Type foo_dt; foo_dt. register_type (dp, “Foo”); // create a topic Topic_var foo_topic = dp->create_topic (“Foo_topic”, //topic name “Foo”, // type name TOPIC_QOS_DEFAULT, // Qos policy NULL); // topic listener 38
Tutorial on DDS Douglas C. Schmidt Example: Create Subscriber and Data. Reader …… // create a subscriber from the domain participant Subscriber. Qos s. Qos; dp->get_default_subscriber_qos (s. Qos); Subscriber_var s = dp->create_subscriber (s. Qos, NULL); // create a data reader from the subscriber // and associate it with the created topic Data. Reader_var reader = s->create_datareader (foo_topic. in (), DATAREADER_QOS_DEFAULT, NULL); Foo. Data. Reader_var foo_reader = Foo. Data. Reader: : _narrow (reader. in ()); 39
Tutorial on DDS Douglas C. Schmidt Example: Create Publisher and Data. Writer …… // create a publisher from the domain participant Publisher. Qos p. Qos; dp->get_default_publisher_qos (p. Qos); Publisher_var p = dp->create_publisher (p. Qos, NULL); // create a data writer from the publisher // and associate it with the created topic Data. Writer_var writer = p->create_datawriter (foo_topic. in (), DATAWRITER_QOS_DEFAULT, NULL); // narrow down to specific data writer Foo. Data. Writer_var foo_writer = Foo. Data. Writer: : _narrow (writer); // publish user-defined data Foo foo_data; foo_writer->write (foo_data); 40
Tutorial on DDS Douglas C. Schmidt How to Get Data (Async Listener-based) Listener_var subscriber_listener = new My. Listener(); foo_reader->set_listener(subscriber_listener); My. Listener: : on_data_available(Data. Reader reader) { Foo. Seq_var received_data; Sample. Info. Seq_var sample_info; reader->take(received_data. out (), sample_info. out (), ANY_SAMPLE_STATE, ANY_LIFECYCLE_STATE); // Use received_data …… } 41
Tutorial on DDS Douglas C. Schmidt How to Get Data (Sync Wait-based) Condition_var foo_condition = reader->create_readcondition(ANY_SAMPLE_STATE, ANY_LIFECYCLE_STATE); Wait. Set waitset; waitset->attach_condition(foo_condition); Condition. Seq_var active_conditions; Duration_t timeout = {3, 0}; waitset->wait(active_conditions. out (), timeout); . . . Foo. Seq_var received_data; Sample. Info. Seq_var sample_info; reader->take_w_condition(received_data. out (), sample_info. out (), foo_condition); // Use received_data 42
Tutorial on DDS Douglas C. Schmidt Benchmark Scenario • Two processes perform IPC in which a client initiates a request to transmit a number of bytes to the server along with a seq_num (pubmessage), & the server simply replies with the same seq_num (ackmessage). – The invocation is essentially a two-way call, i. e. , the client/server waits for the request to be completed. • The client & server are collocated. • DDS & JMS provides topic-based pub/sub model. • Notification Service uses push model. • SOAP uses p 2 p schema-based model. 43
Tutorial on DDS Douglas C. Schmidt Testbed Configuration • Hostname blade 14. isislab. vanderbilt. edu • OS version (uname -a) Linux version 2. 6. 14 -1. 1637_FC 4 smp (bhcompile@hs 20 -bc 1 -4. build. redhat. com) • GCC Version g++ (GCC) 3. 2. 3 20030502 (Red Hat Linux 3. 2. 3 -47. fc 4) • CPU info Intel(R) Xeon(TM) CPU 2. 80 GHz w/ 1 GB ram 44
Tutorial on DDS Douglas C. Schmidt Test Parameters // Complex Sequence Type struct Inner { string info; long index; }; typedef sequence<Inner> Inner. Seq; struct Outer { long length; Inner. Seq nested_member; }; • Average round-trip latency & dispersion • Message types: – sequence of bytes – sequence of complex type • Lengths in powers of 2 • Ack message of 4 bytes • 100 primer iterations • 10, 000 stats iterations typedef sequence<Outer> Complex. Seq; 45
Tutorial on DDS Douglas C. Schmidt Roundtrip Latency Results (1/2) 46
Tutorial on DDS Douglas C. Schmidt Roundtrip Latency Results (2/2) 47
Tutorial on DDS Douglas C. Schmidt Roundtrip Latency Results Analysis • From the results we can see that DDS has significantly better performance than other SOA & pub/sub services. • Although there is a wide variation in the performance of the DDS implementations, they are all at least twice as fast as other pub/sub services. 48
Tutorial on DDS Douglas C. Schmidt Encoding/Decoding Benchmarks • Measured overhead and dispersion of – encoding C++ data types for transmission – decoding C++ data types from transmission • DDS 3 and GSOAP implementations compared • Same data types, platform, compiler and test parameters as for roundtrip latency benchmarks 49
Tutorial on DDS Douglas C. Schmidt Encoding/Decoding Results (1/2) 50
Tutorial on DDS Douglas C. Schmidt Encoding/Decoding Results (2/2) 51
Tutorial on DDS Douglas C. Schmidt Encoding/Decoding Results Analysis • Slowest DDS implementation is compared with GSOAP. • DDS is faster. – Almost always by a factor of 10 or more. – GSOAP is encoding XML strings. • Difference is larger for byte sequences. – DDS implementation has optimization for byte seq. • Encodes sequence as a single block – no iteration. – GSOAP always iterates to encode sequences. • Jitter discontinuities occur at consistent payload sizes. 52
Tutorial on DDS Douglas C. Schmidt DDS Information consumer subscribe to information of interest Information producer publish information DDS matches & routes relevant info to interested subscribers • Efficient Publish/Subscribe interfaces • Qo. S suitable for real-time systems – deadlines, levels of reliability, latency, resource usage, time-based filter • Listener & wait-based data access suits different application styles • Support for content-based subscriptions • Support for data-ownership • Support for history & persistence of data-modifications 53 publishers subscribers Summary of DCPS features
Tutorial on DDS Douglas C. Schmidt Data Local Reconstruction Layer (DLRL) Track 3 D_Track DLRL Cache Track_Topic 3 D -Track 54 DCPS
Tutorial on DDS Douglas C. Schmidt Goals of DLRL • Data Local Reconstruction Layer (DLRL) model is local to an application • “Object-oriented” access to data • Application developers can – describe classes with their methods, data fields, & relations – attach some of those data fields to DCPS entities – manipulate these objects (i. e. , create, read, write, delete) using native language constructs – activate attached DCPS entities to update objects – have these objects managed in a cache • Like a mapping or binding (intuition only) 55
Tutorial on DDS Douglas C. Schmidt DLRL Objects • DLRL objects are (native) language objects with: – data members and methods • Only the data members are relevant to data distribution; they can be: – attributes, i. e. , values – relations, i. e. , reference other objects – plain local data members (i. e. , not known or involved in data distribution) are also supported • DLRL classes can be organised by inheritance • DLRL objects maps to one or more DCPS Topics 56
Tutorial on DDS Douglas C. Schmidt DLRL Object Examples 57
Tutorial on DDS Douglas C. Schmidt DLRL Radar Example Track tracks a_radar x : real y : real comments [*] : string w : integer * Radar 0. . 1 TRACK_TOPIC Class Oid x y radar Track 1 100 200 11 Track 3 D 2 102 201 11 COMMENTS_TOPIC Track 3 D Class Oid index comments Track T 3 D_TOPIC 1 0 a comment Track z : real 1 1 another comment RADAR_TOPIC Oid z Oid 2 300 11 RADAR_TRACKS_TOPIC R_Oid index T_Class T_Oid 11 58 0 Track 1 11 1 Track 3 D 2
Tutorial on DDS Douglas C. Schmidt Evaluating DDS Pros • Low overhead & efficient use of transport bandwidth – e. g. , less features/overhead than CORBA in the main request path • Dynamically scalable & highly flexible – e. g. , can receive notification about all sorts of meta-events, such as new topics, publishers, subscribers, etc. • Supports one-to-one, one-to-many, many-to-one, & many-to-many communication • Large number of configuration parameters & Qo. S policies that give developers extensive control of message transmission & reception Cons • DDS is not well suited to request-reply services, file transfer, & transaction processing • The data-distribution paradigm is not optimized for sending a request & waiting for a reply • Implementations don’t yet cover the entire spec • Lack of interoperability between different vendor’s implementations 59
Tutorial on DDS Douglas C. Schmidt Comparing CORBA with DDS Distributed object • Client/server • Remote method calls • Reliable transport Best for • Remote command processing • File transfer • Synchronous transactions Distributed data • Publish/subscribe • Multicast data • Configurable Qo. S Best for • Quick dissemination to many nodes • Dynamic nets • Flexible delivery requirements DDS & CORBA address different needs Complex systems often need both… 60
Скачать презентацию Overview of the OMG Data Distribution Service Douglas
beef63aca2203e591ee15ccb4dd3644b.ppt