Community Grids Architecture Discussion Internal Discussion August 20 2004 Geoffrey Fox Community Grids Lab Indiana University [email protected].
Download ReportTranscript Community Grids Architecture Discussion Internal Discussion August 20 2004 Geoffrey Fox Community Grids Lab Indiana University [email protected].
Community Grids Architecture Discussion
Internal Discussion August 20 2004 Geoffrey Fox Community Grids Lab Indiana University [email protected]
Philosophy of Web Service Grids
• Much of Distributed Computing was built by natural extensions of computing models developed for sequential machines • This leads to the distributed object (DO) model represented by Java and CORBA – RPC (Remote Procedure Call) or RMI (Remote Method Invocation) for Java • Key people think this is not a good idea as it scales badly and ties distributed entities together too tightly – Distributed Objects Replaced by Services • Note CORBA organization and proposed infrastructure – and Java was considered too complicated in both was considered as “tightly coupled to Sun” – So there were other reasons to discard • Thus replace distributed objects by services connected by “ one-way ” messages and not by request-response messages
Web services
• Web Services principles. build loosely-coupled, distributed applications, based on the SOA • Web Services interact by exchanging messages in SOAP format • The contracts for the message exchanges that implement those interactions are described via WSDL interfaces.
Databases Devices Programs Computational resources
SOAP messages Humans
Importance of SOAP
• SOAP defines a very obvious message structure with a header and a body • The header contains information used by the “ Internet operating system ” – Destination, Source, Routing, Context, Sequence Number … • The message body is only used by the application and will never be looked at by “operating system” except to encrypt, compress it etc.
• Much discussion in field revolves around what is in header!
– e.g. WSRF adds a lot to header
Web Services
• Java is very powerful partly due to its many “frameworks” that generalize libraries e.g.
– Java Media Framework – Java Database Connectivity JDBC • Web Services have a correspondingly collections of specifications that represent critical features of the distributed operating systems for “Grids of Simple Services” – Some 60 active WS-* specifications for areas such as –
a.
Core Infrastructure Specifications
–
b.
Service Discovery
– –
c.
d.
Security Messaging
– – – –
e.
f.
g.
h.
Notification Workflow and Coordination Characteristics Metadata and State
WS-I Interoperability
• Critical underpinning of Grids and Web Services is the gradually growing set of specifications in the Web Service Interoperability Profiles • Web Services Interoperability (WS-I) Interoperability Profile 1.0a." http://www.ws-i.org
. gives us XSD, WSDL1.1, SOAP1.1, UDDI in basic profile and parts of WS-Security in their first security profile.
• We imagine the “60 Specifications” being checked out and evolved in the cauldron of the real world and occasionally best practice identifies a new specification to be added to WS-I which gradually increases in scope –
Note only 4.5 out of 60 specifications have “made it” in this definition
•
Web Services Grids and WS-I+
WS-I
Interoperability support Grids doesn’t cover all the capabilities need to •
WS-I+
is designed to minimal extension of WS-I to support “most current” Grids: it adds support for – Enhanced SOAP Addressing (WS-Addressing) – Fault tolerant (reliable) messaging – Workflow as in IBM-Microsoft standard BPEL • Security and Notification get added soon best practice and support will probably – There are Web Service frameworks here but various IBM v Microsoft v Globus differences to be resolved • Portlet-based User Interfaces could be added • UK OMII Open Middleware Infrastructure Institute is adopting this approach to support UK e-Science program – Currently UK e-Science largely either uses GT2 (as in EDG) or Simple Web Services for “database Grids” – http://www.omii.ac.uk/
Application Specific Grids Generally Useful Services and Grids Workflow WSFL/BPEL Service Management (“Context etc.”) Service Discovery (UDDI) / Information Service Internet Transport
Protocol Service Interfaces WSDL Base Hosting Environment Protocol HTTP FTP DNS … Presentation XDR … Session SSH … Transport TCP UDP … Network IP … Data Link / Physical
Layered Architecture for Web Services and Grids
Higher Level Services Service Context Service Internet Bit level Internet
Working up from the Bottom
• We have the classic (CISCO, Juniper ….) Internet routing the flood of ordinary packets • Web Services build the “Service Internet” with – Fault Tolerance (WS-RM not TCP) – Security (WS-Security not IPSec/SSL etc.) – Information Services (UDDI/WS-Context not DNS/Configuration files) – At message/web service level and not packet/IP address level • Software-based Service Internet useful as computers “fast” • Familiar from Peer-to-peer networks and built as a software overlay network defining Grid (analogy is VPN) • On top of “Service Internet”, one supports dynamic context or the “shared memory” supporting groups (from 2 to more) of services
Dynamic light-weight Peer-to-peer Collaboration Training Grid
Enterprise Grid Students Information Grid R1 R2 Teacher Campus Grid Compute Grid
4 Overlay Networks With a 5 th superimposed
Consequences of Rule of the Millisecond
• Useful to remember critical time scales – 1) 0.000001 ms – 2) 0.001 to 0.01 ms – CPU does a calculation – MPI latency – 3) 1 to 10 ms – wake-up a thread or process – 4) 10 to 1000 ms – Internet delay • 4) implies geographically distributed metacomputing can’t in general compete with parallel systems (OK for some cases) • 3) << 4) implies RPC not a critical programming abstraction as it ties distributed entities together and gains a time that is typically only 1% of inevitable network delay – However many service interactions are at their heart RPC but implemented differently at times e.g. asynchronously • 2) says MPI is not relevant for a distributed environment as low latency cannot be exploited • Even more serious than using RMI/RPC, current Object paradigms also lead to mixed up services with unclear boundaries and autonomy • Web Services are only interesting model for services today
Linking Modules
Closely coupled Java/Python … Coarse Grain Service Model Module B Module A Method Calls .001 to 1 millisecond Service B Messages Service A 0.1 to 1000 millisecond latency
From method based to RPC to message based to event-based
“Listener” Subscribe to Events
Service B
Message Queue in the Sky Publisher Post Events
Service A
What is a Simple Service?
• Take any system – it has service multiple functionalities – We can implement each functionality as an independent distributed – Or we can bundle multiple functionalities in a single service • Whether functionality is an independent service or one of many method calls into a “ • Simple services glob of software ”, we can always make them as Web services by converting interface to WSDL are gotten by taking functionalities and making as small as possible subject to “rule of millisecond” – Distributed services incur messaging overhead of one (local) to 100’s (far apart) of milliseconds to use message rather than method call – Use scripting or compiled integration of functionalities ONLY when require <1 millisecond interaction latency • Apache web site has many projects that are multiple functionalities presented as ( Java) globs file access .. services and NOT (Java) Simple Services – Makes it hard to integrate sharing common security, user profile,
What is a Grid I?
• You won’t find a clear description of what is Grid and how does differ from a collection of Web Services – I see no essential reason that Grid Services have different requirements than Web Services – Geoffrey Fox, David Walker,
e-Science Gap Analysis
, June 30 2003. Report UKeS-2003-01, http://www.nesc.ac.uk/technical_papers/UKeS-2003-01/index.html
. – Notice “service-building model” is like programming language – very personal!
• Grids were once defined as “ Internet Scale Distributed Computing ” but this isn’t good as Grids depend as much if not more on data as well as simulations • So Grids can be termed “ Internet Scale Distributed Simple Services ” and represent a way of collecting services together in same way that program (package) collects methods and objects together.
What is a Grid II?
• So we build collections of Web Services which we package as component Grids – Visualization Grid – Sensor Grid – Utility Computing Grid – Person (Community) Grid – Earthquake Simulation Grid – Control Room Grid – Crisis Management Grid • We build bigger Grids by composing component Grids using the Service Internet
Grids of Grids of Simple Services
• Link via methods messages streams • Services and Grids are linked by messages • Internally to service, functionalities are linked by methods • A simple service is the smallest Grid • We are familiar with method-linked hierarchy Lines of Code Methods Objects Programs Packages Methods Services Component Grids CPUs Databases Clusters MPPs Federated Databases Compute Resource Grids Data Resource Grids Sensor Sensor Nets Overlay and Compose Grids of Grids
Flood CIGrid Flood Services and Filters
…
Electricity CIGrid
…
Gas CIGrid Gas Services and Filters Collaboration Grid Sensor Grid Portals GIS Grid Registry Security Data Access/Storage
Core Grid Services
Notification Workflow Physical Network Visualization Grid Compute Grid Metadata Messaging
Critical Infrastructure (CI) Grids built as Grids of Grids
Field Trip Data Repositories Federated Databases Database Database Sensors Streaming Data Sensor Grid Database Grid Research Education SERVOGrid
Data Filter Services
Compute Grid Research Simulations ?
Discovery Services
GIS Grid Analysis and Visualization Portal
Customization Services From Research to Education
Education Grid Computer Farm
Geoscience Research and Education Grids
NaradaBrokering
Audio/Video Conferencing Client
Computer Modem
NaradaBrokering Broker Network
Minicomputer Firewall Stream Server-enhanced Messaging Workstation Peers PDA Server Web Service B Peers Queues Laptop computer
Audio/Video Conferencing Client NB supports messages and streams
Current NaradaBrokering Features
Multiple transport support In publish-subscribe Paradigm with different Protocols on each link Subscription Formats Reliable delivery Ordered delivery Recovery and Replay Security Message Payload options Messaging Related Compliance Grid Application Support Transport protocols supported include TCP, Parallel TCP streams, UDP, Multicast, SSL, HTTP and HTTPS.
Communications through authenticating proxies/firewalls & NATs. Network QoS based Routing Subscription can be Strings, Integers, XPath queries, Regular Expressions, SQL and tag=value pairs.
Robust and exactly-once delivery of messages in presence of failures Producer Order and Total Order over a message type Time Ordered delivery using Grid-wide NTP based absolute time Recovery from failures and disconnects.
Replay of events/messages at any time.
Message-level WS-Security compatible security Compression and Decompression of payloads Fragmentation a nd Coalescing of payloads Java Message Service (JMS) 1.0.2b compliant Support for routing P2P JXTA interactions.
NaradaBrokering enhanced Grid-FTP. Bridge to the Globus TK3.
Web Service reliability Prototype implementation of WS-ReliableMessaging
IOI and CIE
• Let us study the two layers IOI ( Service I nternet O n the Bit I nternet) and CIE ( C ontext and I nformation E nvironment) • IOI is most “straightforward” as it is providing reasonably well understood capabilities at a new “level” • CIE is roughly the inter-service “shared memory” used to manage and control them at “distributed operating system level – Critical is “shared” (a database service) versus message based CIE
Application Specific Grids Generally Useful Services and Grids Workflow WSFL/BPEL Service Management (“Context etc.”) Service Discovery (UDDI) / Information Service Internet Transport
Protocol Service Interfaces WSDL Higher Level Services CIE IOI
NaradaBrokering and IOI
• “Software Overlay Network” features • Support for Multiple Transport protocols • Support for multiple delivery mechanisms – Reliable Delivery – Exactly-once Delivery – Ordered Delivery – Optional Delivery optimization modules for different modes • Compression/Decompression of payloads with optional module • Coalescing/Fragmentation of payloads with optional module • NTP Time Service • Security Service with optional module • Virtual Private Grid
Virtualizing Communication
Communication specified in terms of user goal Service and Quality of – not in choice of port number and protocol Bit Internet Protocols have become firewall will not support ……… overloaded e.g. MUST use UDP for A/V latency requirements but CAN’t use UDP as A given “ Service Internet ” communication can involve multiple transport protocols and multiple destinations – the latter possibly determined dynamically NB Brokers
Fast Link A Satellite UDP Firewall HTTP B 1
NB Broker Software Multicast Client Filtering
Hand-Held Protocol Dial-up Filter B 3 B 2
NaradaBrokering and IOI
• Optimization of Messaging Environment (Broker set up etc.) using CIE • Support of WS-FlexibleRepresentation (see later) in transport of Streams • Performance Monitoring • Performance optimized routing with optional module • Ad-hoc Network Support • Broker Discovery --Find nearest broker • Topic discovery • SOAP Support and integration with WSIF, Axis, WSE, gSOAP, Java WSDP • Support for WS-Reliability, WS-ReliableMessaging and their Federation • Congestion control and other optimizing message delivery strategies
Performance Monitoring
Every broker incorporates a Monitoring service monitors links originating from the node.
that Every link measures and exposes a set of metrics
•
Average delays, jitters, loss rates, throughput.
Individual links can disable measurements for individual or the entire set of metrics.
Measurement intervals can also be varied Monitoring Service, returns measured metrics to Performance Aggregator .
Broker Node Link Data Monitoring Service Aggregates info from nodes in a certain domain Link Data Performance Aggregation Service Broker Node Control Message Exchange
NaradaBrokering Service Integration
S1 P1 P2 S2 Proxy Messaging Handler Messaging S1 S2 Notification S1 S2 S?
Service P?
Proxy
Any Transport NB Transport Standard SOAP Transport Internal to Service: SOAP Handlers/Extensions/Plug-ins Java (JAX RPC) .NET Indigo and special cases: PDA's gSOAP, Axis C++
Mechanisms for Reliable Messaging I
Service B Service A
M(n) M(n+1) • There are essentially sequence numbers on each message • Unreliable transmission detected by non-arrival of a message with a particular sequence number • Remember this is “just some TCP reliability” built at application level • One can either use ACK’s – Receiver (service B) positively acknowledges messages when received – Service A fully responsible for reliability • Or NAK’s – Service B is partially responsible and tracks message numbers – sends a NAK if sequence number missing
Mechanisms for Reliable Messaging II
• Each message has a retransmission time; messages are retransmitted if ACK’s not received in time – Uses some increasing time delay if retransmit fails • Note need to be informed (eventually) that OK to throw away messages at sender; pure NAK insufficient • Note this is reliability from final end-point to beginning end-point: TCP reliability is for each link and has different grain size and less flexible reliability mechanisms • There are several efficiency issues – Divide messages into groups and sequence within groups – Do not ACK each message but rather sequences of messages • NAK based system attractive if high latency (some mobile devices) on messaging from receiver back to sender
Custom Message Reliability
2 second PDA reply latency!
Different endpoints may well need different reliability schemes. Another reason to use application layer.
Need to define easy to use “standard reliability profiles
Wireless Optimized WS-RM Filter 2 Narada Broker WS-RM WS-Reliability Filter 1
NaradaBrokering and Fault Tolerance
• As well as reliable messaging, NaradaBrokering supports performance based dynamic routing – Choose both route and protocol (UDP, Parallel TCP ..) • It will also support automatic fail-over among replicated services subscribing to same message stream • Provides scriptable control of streams for custom management schemes • Saves ALL messages in fault tolerant storage for either session replay or recovery • Will support reliable BitTorrent P2P file swapping model (better than GridFTP?)
GridFTP plus NaradaBrokering
Mirror Mirror on the wall Who is the fastest most reliable of them all?
Web Services !!!
• Application layer “Internet” allows one to optimize message streams and the cost of “startup time”, Web Services can deliver the fastest possible interconnections with or without reliable messaging • Typical results from Grossman (UIC) comparing Slow SOAP over TCP with binary and UDP transport (latter gains a factor of 1000 ) Record Count 10000 50000 150000 375000 1000000 5000000 MB 0.93 4.65 13.9 34.9 93 465 µ 2.04 8.21 26.4 75.4 278 σ/µ 6.45% 1.57% 0.30% 0.25% 0.11% 2.23% WS-DMX/ASCII MB µ σ/µ 0.5 2.4 7.2 18 48 242 1.47 1.79 2.09 3.08 3.88 8.45 0.61% 0.50% 0.62% 0.29% 1.73% 6.92% MB WS-DMX/Binary µ σ/µ 0.28 1.4 4.2 10.5 28 140 1.45 1.63 1.94 2.11 3.32 0.38% 0.27% 0.85% 1.11% 0.25% 8.12%
SOAP Tortoise and UDP Hare II
• Mechanism only works for streams messages – sets of related • SOAP header in streams is constant except for sequence number (Message ID), time-stamp ..
• One needs two types of new Web Service Specification • “ WS-StreamNegotiation ” to define how one can use WS-Policy to send messages at start of a stream to define the methodology for treating remaining messages in stream • “ WS-FlexibleRepresentation ” to define new encodings of messages
SOAP Tortoise and UDP Hare III
• Then use “WS-StreamNegotiation” to negotiate stream in Tortoise SOAP – ASCII XML over HTTP and TCP – – Deposit basic SOAP header through connection – it is part of context for stream (linking of 2 services) – Agree on firewall penetration, reliability mechanism, binary representation and fast transport protocol – Naturally transport UDP plus WS-RM • Use “WS-FlexibleRepresentation” to define encoding of a Time stamp if needed Fast transport (On a different port) with messages just having “FlexibleRepresentationContextToken”, Sequence Number, – RTP packets have essentially this structure – Could add stream termination status • Can monitor and control with original negotiation stream • Can generate different streams optimized for different end-points
CIE: Common Service Information and Metadata
• WS-RF and WS-GAF approach state with different approaches to contextualization – supplying a common “context” (Shared token or more generally (resource) metadata) • One can supports such a common context either as pool of messages or as message-based access to a “database” (Context Service) • We define a collection of services sharing information as a Gaggle • Two services linked by a stream are perhaps simplest example of a Gaggle
CIE II
• There is small amount of core shared information that must be supported in the gaggle. This includes dynamic service metadata and the equivalent of configuration information.
– We can put any “light-weight” information in the CIE even if it is very application dependent • There are various metadata mechanisms suitable for “light weight dynamic situations”. WS-MetadataExchange, WS Context and WSRF represent variants that can be supported.
– UDDI, Globus MDS are “bigger” solutions • The CIE stops short of supporting workflow; that would be built on top of CIE in analogous fashion to way that WS CAF supports transactions as an extension of WS-Context. • Note that there is a tension between storing metadata in messages and services computing . – This is shared versus distributed memory debate in parallel
CIE III
• We view the metadata as equivalent to some shared memory for the services and whatever the actual implementation build a logical interface that is message based interactions with a simple database .
• There is a scripting environment that allows one to interact with administrative aspects of services and the support environment – it programs the Gaggle .
• Dennis Gannon notes that Linda ( JavaSpaces ) is an interesting model for the CIE at the Gaggle level. • Notification Service is one simple CIE information tool
CIE in NaradaBrokering I
• Scripting Management based on HPSearch technology • Topic discovery information repository • Broker discovery information repository • NaradaBrokering metadata information repository • Support of WS-FlexibleRepresentation information and negotiation services • CIE Portal allowing access to all metadata, management of deployment, firewall tunnels, performance info, error logs etc.
– User and “programmatic” interfaces to set “defaults” • Security Access and Authentication interface
– Recovery
CIE in NaradaBrokering II
• Support of fault tolerant message storage, discovery and access – Replay or time difference sensitive recovery – Instant replay or replay of real-time streams – Support of finite state change architecture with major (complete) and minor (partial) update events • Support of replicated services • Other core service fault tolerance mechanisms such as scalable heartbeat • Support of multiple topic subscription types including string, integer, XPath, Regular Expressions and
CIE in NaradaBrokering III
• Fault tolerant light weight metadata database (the core metadata technology) • Support of WS-Discovery masquerading as dynamic UDDI • Support of WS-DM to manage services • Support of WSRF • Support of WS-Context • Support for WS-MetadataExchange • Support of session management aspects of XGSP, JXTA Peer Groups etc.
• GridFTP • FTHPIS
Web Service Metadata and State I
• • • • The Semantic Grid and Semantic Web are important frameworks for metadata but handicapped by lack of “compelling” tools
RDF
Resource Description Framework (W3C) Set of recommendations expanded from original February 1999 standard http://www.w3.org/RDF/ and the heart of the Semantic Web and Grid http://www.semanticgrid.org
DAML+OIL
combining DAML (Darpa Agent Markup Language) and OIL (Ontology Inference Layer) (W3C) Note December 2001 http://www.w3.org/TR/daml+oil reference
OWL
Web Ontology Language (W3C) Recommendation February 2004 http://www.w3.org/TR/2004/REC-owl features-20040210/ More later!
•
Web Service Metadata and State II
WS-DistributedManagement
Web Services Distributed Management Framework with MUWS and MOWS below (OASIS) http://www.oasis • • • open.org/committees/tc_home.php?wg_abbrev=wsdm • Management includes issues like monitoring quality of service, enforcing service level agreements, controlling tasks and managing life-cycles.
WSDM-MUWS
Using Web Services (OASIS) V0.5 Committee Draft April 2004 http://www.oasis-open.org/committees/download.php/6234/cd-wsdm muws-0.5.pdf
Web Services Distributed Management: Management
WSDM-MOWS
Web Services Distributed Management: Management of Web Services (OASIS) V0.5 Committee Draft April 2004 http://www.oasis-open.org/committees/download.php/6255/cd-wsdm mows-0.5-20040402.pdf
WS-MetadataExchange
extensively Web Services Metadata Exchange (BEA,IBM, Microsoft, SAP) March 2004 http://www 106.ibm.com/developerworks/library/specification/ws-mex/ – Describes how metadata can be exchanged between services rather than by looking it up in registries like UDDI or higher level metadata catalogs; the old OGSI standard used such service-resident metadata
Web Service Metadata and State III
• • • •
WS-RF
Web Services Resource Framework including
WS ResourceProperties, WS-ResourceLifetime, WS RenewableReferences, WS-ServiceGroup,
and
WS-BaseFaults
(OASIS) http://www.oasis open.org/committees/tc_home.php?wg_abbrev=wsrf with Oasis TC set up April 2004 and V1.1 Framework March 2004 http://www 106.ibm.com/developerworks/library/ws-resource/ws modelingresources.pdf
– Uses rich metadata to define stateful interactions – its use of SOAP header creates interoperability problems
ASAP
Asynchronous Service Access Protocol (OASIS) http://www.oasis-open.org/committees/tc_home.php?wg_abbrev=asap with V1.0 working draft G June 2004 http://www.oasis open.org/committees/download.php/7151/wd-asap-spec-01g.pdf
WS-GAF
Web Service Grid Application Framework (Arjuna, Newcastle University) http://www.neresc.ac.uk/ws-gaf/ – Uses WS-Context to provide “opaque” (don’t say much) stateful interactions
Metadata Catastrophe I
• We keep finding places where metadata can be transmitted to and from services • WS-Addressing and WS-RF specify metadata in SOAP header of messages • WS-Context similarly specifies both SOAP header and WS Context context services as location of (temporary) metadata • We have registries like UDDI of service data • WS-MetadataExchange covers metadata stored in services – Service metadata is very common and often not explicitly called out e.g. WebDAV as in Apache Slide stores file metadata in addition to versioning information • In addition, we have major source of one or more (federated) catalogs • I think this confused situation will need to be addressed by some new dynamic metadata model
Metadata Catastrophe II
• There are large long term metadata catalogs associated with major applications/services – These are likely to remain as now based on traditional major database technology like Oracle MySQLK and DB2 • There are small but broadly available metadata catalogs – Globus MDS and EDG RGMA roughly address these – Semantic Grid enriched Service catalogs as in UDDI • We need to implement UDDI in a distributed (federated) fashion and work around its non-intuitive schema but this seems straightforward • All the problems occur for local and highly dynamic data where key issues are: – Consistency: If metadata stored in messages flowing around, how do we ensure consistency if it ever changes – Where is it: How do we decide where to look it up?
• My intuition is that best solution is highly dynamic lightweight database – doesn’t really fit any proposal yet!
Metadata and Semantic Grid
• Can store in one catalog, multiple catalog s or in each service – Not clear how a coherent approach will develop • Specialized metadata services like UDDI and MDS (Globus) – Nobody likes UDDI – MDS uses old fashioned LDAP – RGMA is MDS with a relational database backend • Some basic XML database (Oracle, Xindice …) • “By hand” as in current SERVOGrid Portal which is roughly same as using service stored SDE’s ( Service Data Elements ) as in OGSI • Semantic Web (Darpa) produced a lot of metadata tools aimed at annotating and searching/reasoning about metadata enhanced web pages – Semantic Grid uses for enriching Web Services – Implies interesting programming model with traditional analysis (compiler) augmented by meta-data annotation
Four Metadata Architectures
System or Federated Registry or Metadata Catalog Database Database1 Grid or Domain Specific Metadata Catalogs Database2 Web Service Ports Database3 SDE1 SDE2 Service
M M
SDE1 SDE2 Service
M
SDE1 SDE2 Service SDE1 SDE2 Service SDE1 SDE2 Service
M
Individual Services
M M M
Messages
M M
SDE1 SDE2 Service
M M
SDE1 SDE2 Service
M
• •
Stateful Interactions
There are (at least) four approaches to specifying state
–
OGSI use factories to generate separate services for each session in standard distributed object fashion
–
Globus GT-4 and WSRF use metadata of a resource to identify state associated with particular session
– –
WS-GAF uses WS-Context to provide abstract context defining state. Has strength and weakness that reveals less about nature of session WS-I+ “Pure Web Service” leaves state specification the application – e.g. put a context in the SOAP body I think we should smile and write a great metadata service hiding all these different models for state and metadata
Explicit and Implicit Factories
• Stateful interactions are typified by amazon.com where messages carry correlation information allowing multiple messages to be linked together – Amazon preserves state in this fashion which is in fact preserved in its database permanently • Stateful services have state that can be queried outside a particular interaction • Also note difference between implicit and explicit factories – Some claim that implicit factories scale as each service manages its own instances and so do not need to worry about registering instances and lifetime management
F A C T O R Y
Implicit Factory
1 2 3 4
Hidden instances Explicit instances
1 2 3 4
Explicit Factory
F A C T O R Y
Web Service Notification I
• • •
WS-Eventing
Web Services Eventing (BEA, Microsoft, TIBCO) January 2004
http://msdn.microsoft.com/library/default.asp?url=/libr ary/en-us/dnglobspec/html/WS-Eventing.asp
WS-Notification
Framework for Web Services Notification with
WS-Topics , WS-BaseNotification ,
and
BrokeredNotification WS-
(OASIS) OASIS Web Services Notification TC Set up March 2004
http://www.oasis open.org/committees/tc_home.php?wg_abbrev=wsn http://www-
and
106.ibm.com/developerworks/library/specification/ws notification/ JMS
Java Message Service V1.1 March 2002
http://java.sun.com/products/jms/docs.html
Notification Architecture
• Point-to-Point
Service B Service A Subscribe Publish
• Or Brokered
Subscribe Broker Publish Service B
Queues Messages Supports creation and subscription of topics
Service A
• Note that MOM ( Message Oriented Middleware ) uses brokered messaging for ALL transmission and not just “special” notification messages
Classic Publish-Subscribe
Notification Service
Web Service Notification II
•
WS-Eventing
is quite similar to WS-BaseNotification and provides service to service notification •
WS-Notification
is similar to CORBA event service and adds brokers to mediate notification which has several advantages – Don’t need queues and lists of subscribers on each service – Solution scales to any number of publishers/subscribers • JMS well known successful non Web Service brokered notification system • Topics defined in
WS-Topics
contextualization can also provide • Expect this area to clarify reasonably soon
• • • • • • • •
FTHPIS: CIE for Applications I
CIE is designed to support metadata needed by the “galactic distributed operating system” controlling services It can used for cases where an application naturally is supported by a Gaggle services with a modest amount of metadata A new approach to Configuration files Maintaining user profiles and preferences Maintaining application specific metadata such as GIS metadata Maintain “Grading System” for e-learning Maintaining information regarding sessions and also the state of entities in these sessions in GlobalMMCS Enable playback/replay capabilities in GlobalMMCS or other NaradaBrokering applications needing fault-tolerant event storage
• •
FTHPIS: CIE for Applications II
Three key features of FTHPIS are – Supported in Peer-to-peer or central server role – Fault Tolerance – Context (database) is lightweight and either files or real databases Fault Tolerance includes – WS-ReliableMessaging – Replicated Storage of Messages to support reliable messaging and permanent storage for replay. Also supports performance as data locality (caching) – Fault Tolerance metadata stored in CIE (FTHPIS) – NaradaBrokering support of replicated services where these services are discovery, access, storage etc for each FTHPIS (include caching support – location of nearest copy)
P2P and NaradaBrokering I
Server/Broker-free version to support based Community Grids using P2P “immediate deployment” of NB versions of Grid applications Initially:
•
Use a broker free If successful, add version of NB and brokers in the Grid sky file-based Web services to achieve better performance (if broker has better network link than clients) Service providers and supercomputer/national grid centers could sell such Grid Farm services Grid Farm in the Sky (clouds)
NaradaBrokers
P2P
P2P and NaradaBrokering II
Global Information System could start with multiple FTHPIS style lightweight Fault Tolerant Metadata Catalogs – subscribe multiple instances of metadata service to the MetadataCatalogN topics – publish queries to these replicated subscriber topics Then one could add DHT (Distributed Hash Table) approach (used in latest JXTA) in NB. NB nodes will have Ids that can determine where a specific content would be stored.
•
Provides scalable location of content over LARGE numbers of end-points with limited query capabilities Flooding to bridge gap between DHT and FTHPIS where queries are broadcast to multiple gaggles and FTHPIS is used within each Gaggle
P2P and NaradaBrokering III
GridTorrent: Merge NB-enhanced GridFTP and P2P BitTorrent http://bitconjurer.org/BitTorrent/ to provide WSRM fault tolerant Parallel TCP P2P or Grid file transfer
•
BitTorrent supports fragmented distributed files which are
•
natural WSRM and NB architecture Don’t really want GridFTP server ; prefer to use fault tolerant GridTorrent metadata service (implemented as a Gaggle FTPHIS)
GridTorrent file transfer automatically provides distributed fault tolerant caching – a better RLS?
P2P and NaradaBrokering IV
We have proposed building a portlet based portal to the IOI CIE and FTHPIS
In the spirit of P2P, we can propose a model where one exposes distributed files AND streams in same way as Napster or Gnutella – each resource (file or stream) is displayed in a “metadata enhanced” ls/Windows Explorer fashion
•
Metadata includes updating thumbnail for streams as in GlobalMMCS or AccessGrid Note that “unit of resource storage” could be individual end-point (client or server) or Gaggles or even better hierarchically endpoints
Gaggles
Gaggles of Gaggles