SlideShare a Scribd company logo
ARCHSTONE: Intelligent Network Services for Advanced Application Workflows
•  USC/ISI                            •  BNL
      •  Tom Lehman                      •  Dimitrios Katramatos
      •  Xi Yang                         •  Sushant Sharma
•    ESnet                               •  Dantong Yu
      •  Chin Guok
      •  Eric Pouyoul
      •  Inder Monga
      •  Vangelis Chaniotakis
      •  Bharath Ramaprasad (UMass)
•    UNM
      •  Nasir Ghani
      •  Feng Gu
      •  Kaile Liang
ARCHSTONE: Intelligent Network Services for Advanced Application Workflows
•  Network "Service Plane" formalization
      •  Composable Network Service architecture
      •  ARCHSTONE Network Service Interface as client entry point
•    Extensions to Topology and Provisioning Schemas to enable:
      •  multi-layer topologies
      •  multi-point topologies
      •  requests in the form of a "service-topology"
      •  vendor specific features
      •  technology specific features
      •  node level constraints
•    MX-TCE (Multi-Dimensional Topology Computation Engine)
      •  Computation Process and Algorithms
•    Enable a New class of Network Services referred to as "Intelligent
     Network Services"
      •  clients can ask the network "what is possible?" questions
      •  can ask for "topologies" instead of just point-to-point circuits
•  Advanced Network Service Interface
      •  "Request Topology" and "Service Topology" concepts
      •  Common Network Resource Description schema
      •  Network Service Plane access point
•    Multi-Dimensional Topology Computation Element (MX-TCE)
      •  High Performance computation with flexible application of constraints
•    Use OSCARSv6 as base infrastructure and development environment


                         request


 Network                                    Network
 Service                                    Provider
Requester                                    Agent
                                                         OSCARSv6
                           reply
                                                                     MX-TCE




                         Network
                         Resource
                        Description
Network Services
                     Interface



          Network Service Plane

    Composite Service (S1 = S2 + S3)



Composite Service     Composite Service
(S2 = AS1 + AS2)      (S3 = AS3 + AS4)


Atomic    Atomic      Atomic     Atomic
Service   Service     Service    Service
 (AS1)     (AS2)       (AS3)      (AS4)




          Multi-Layer Network Data Plane
Topology Service to determine       Security Service (e.g.
resources and orientation           encryption) to ensure data
                                    integrity

Resource Computation Service*       Store and Forward Service to
to determine possible resources     enable caching capability in the
based on multi-dimensional          network
constraints            (*MX-TCE)

                                    Measurement Service to
Connection Service to specify
                                    enable collection of usage data
data plane connectivity
                                    and performance stats


                                    Monitoring Service to ensure
Protection Service to enable
                                    proper support using SOPs for
resiliency through redundancy
                                    production service



Restoration Service to facilitate
recovery
•  Extensions to OSCARS v0.6
•  Added features for:
      •  multi-layer topologies
      •  multi-point topologies
      •  requests in the form of a
         "service-topology"
      •  vendor specific features
      •  technology specific features
      •  node level constraints
•    Result is a schema
     "Superset" to what
     OSCARSv0.6 now uses
      •  schema with ARCHSTONE
         extensions will be backward
         compatible with current
         OSCARS operations
•    Uses perfSONAR Topology
     and Lookup Services
•  MX-TCE role in OSCARS
       •  perform basic path computation for current OSCARS service
       •  standalone Topology Computation element to: answer "what is                            Client
          possible?" questions for clients to subsequently make request for                    Application
          OSCARS services
       •  Advanced multi-layer and multi-point computations

                                                                   What topologies are
                                                                 available next week for
                                                                       3 hour duration?

Current OSCARS Service:                                                                      Here's Two
"Please try and reserve a 5
Gbps VLAN circuit between
A and B at 15:00 on January
16, 2012 for 3 hour
duration"
                                                                                           MX-TCE
  Client
Application


                                                                                  can perform basic
                                                                                  path computations as
                                                                                  needed to respond to
                                                                                  current OSCARS
                                                                                  point-to-point VLAN
                                                                                  service
java web service, C++ multi-threaded core
•  Unified API/NSI support for P2P, Multi-Point, Multi-Layer,
     schedule and co-scheduling requests under extended NML
     schema
•    Implemented OSCARS PCE API to become swappable
     OSCARS module
     •  support existing OSCARS PCE capability as a single TcePCE
     •  support co-scheduling via optionalConstraint extension
•  Multiple path and topology computation workflows
     •  kicked off based on request types: P2P, MP, MLN/MRN, coSheduling
        etc. and combinations
     •  support concurrent requests through multi-threading
•  Modularized differentiated algorithm execution driven by
     workflows
•    Transform computation results into
     •  provisioning friendly path object depending on path control scenarios
     •  NSI and OSCARS compliant reply messages
ARCHSTONE: Intelligent Network Services for Advanced Application Workflows
•  ARCHSTONE Extensions                      •  Layer Decisions
   •  Multi-Layer Topology Representations      •  driven by resource constraints or
   •  Multi-Layer Topology Computations            client requests for specific
   •  Multi-Layer Provisioning                     performance characteristics (i.e.
                                                   low latency, low jitter, etc)
•  Prototype Deployment on Production Networks
   •  ESnet and Internet2
   •  Real-time reservations being processed to provide answers to "what is
      possible?" questions that client can then use to make requests on the
      operational networks
   •  VNOD project will discuss a specific use case for this
                                                          Client
                                                        Application



           coScheduleRequest:
           start time: Wed, 08 Feb 2012 05:00:00 GMT
           end time: Wed, 08 Feb 2012 07:00:00 GMT
           min bandwidth: 1 Gbps
           max number of path options: 3
           format for reply: bandwidth availability graph


           OSCARS                  MX-TCE




                   OSCARS


                                               real-time reservations feed
                                                from production networks
•  archstone.east.isi.edu
     •  Architecture and Design Documents
     •  MX-TCE Software
     •  Extensions to OSCARS Topology and Request
        Schemas
     •  Example topology descriptions, service requests,
        service topologies (responses)
     •  ANI Testbed configuration and use
•    OSCARSv0.6 project
     •  code.google.com/p/oscars-idc/
•    OSCARS ARCHSTONE Branch
     •  oscars.es.net/repos/oscars/branches/archstone/
•  Virtual Network On Demand (VNOD) is a project funded by DOE ASCR
•  Objectives:
      •  To provide on-demand, end-to-end virtual network topologies dedicated
          to servicing communities of users at multiple end-sites
      •  To facilitate the creation and management of such virtual network
          domains
            -  VNOD GUI based front-end
      •  To provide a platform for applying various optimization schemes with
          diverse objectives
            -  Successfully schedule multiple, fixed and flexible, end-to-end data
               transfer requests between multiple end sites and hosts
•    Currently working on developing technologies to address the
     integration and co-scheduling across Network and Middleware level
     resources
      •  This includes co-scheduling algorithms, protocols, and workflows which
          can operate in this "distributed heterogeneous multi-resource
          environment"
      •  Leveraging TeraPaths and ESCPS for end-host and site
          communications and configurations
•    Future work will include integration with application specific
     workflows and systems
•    Utilizing ARCHSTONE Intelligent Network Services as part of co-
     scheduling workflows
End Site                        End Site



                                                            End Site



ESDC         Transit WAN
IDC

                                        Transit WAN




                  Transit WAN

                                                      End Site




                            End Site
Topology           Community
                        Topology                                 distributed auxiliary
                                            Community                                                      Service           Membership
                         Service
                                            Membership                  services                                               service
                                              service
             Service                                                                           Service
            Discovery                                           Monitoring                    Discovery                               Monitoring
                                                                 Service                                                               Service




           ViNet
          request                  Resource Scheduler (RS)                                        Resource Scheduler (RS)
        (application                (selects and schedules                                         (selects and schedules
        requirement                        resources)                                                     resources)
           profile)
                                   VNOD                                                          VNOD
                                                                                                                                            remote RSs
        remote RSs                  ViNet Controller (VNDC)                                        ViNet Controller (VNDC)
                                     (creates and manages                                           (creates and manages
        remote VNDCs                        ViNets)                                                        ViNets)

                                                                                                                                          remote VNDCs
                                                                virtual networking layer


                                                              virtual end-to-end path layer
remote ESCPS                                                                                                                                       remote ESCPS
   negotiation           ESCPS                                                                                          ESCPS                      Negotiation

                                                      OSCARS                              OSCARS
                                                 with ARCHSTONE                      with ARCHSTONE




                                                                   Transit WANs
                        End-Site LAN                                                                                          End-Site LAN
ARCHSTONE: Intelligent Network Services for Advanced Application Workflows
•  VNOD workflow:
    •     User requests a virtual domain setup via the front-end
         -   Request can come from an application via the API
         -   End-system resource availability (e.g., storage system bandwidth)
             can also be submitted for co-scheduling
    •     The VNOD back-end:
         -   Collects bandwidth availability information from domains involved in
             virtual path establishment
             •  Interfaces with ARCHSTONE for WAN domains and TeraPaths/
                ESCPS for end-sites
         -   Runs scheduling algorithm(s)
         -   Makes appropriate reservation requests based on algorithm output
             by engaging ARCHSTONE and TeraPaths/ESCPS
+
              End Site
                A



+
                   Transit WAN W1



                                    Transit WAN W2
or


     +

                                                 End Site
         or                                        B



                         +
                                    +
(a)   (b)




(c)   (d)
Available Bandwidth (Gb/s)

                               Step 1
   10
                                                         Step 3
    8
               Step 0        Request #2
    6
                                            Step 2
    4                                                             Request #1
            Request #3
    2
                                          Request #0
        0        1       2        3        4         5      6      7 Time (secs)
•  Implementation in progress
•  Increasingly difficult scheduling problems:
  •    Single request/single path (TeraPaths/StorNet)
  •    Multiple requests/single path/single WAN domain
  •    Multiple requests/multiple paths/single WAN domain
  •    Multiple requests/multiple paths/multiple WAN domains
•  Co-design with ARCHSTONE to attack
  scheduling complexities
  •  Hierarchical scheduling (end-sites/WAN)
  •  Tighter integration and interoperation
•  The architecture adopted by ARCHSTONE, OSCARS, VNOD:
      •  Centralized at the Intra-Domain level for resource management and service
         provisioning
      •  Distributed at the Inter-Domain level for resource management and service
         provisioning
      •  External topology distribution systems must limit the amount dynamic data
         exported (scalability and stability issues)
      •  Resource identification for real-time service provision can only be done by local
         domain systems
      •  Multi-domain service provision will require chain or tree mode protocols which
         include real-time negotiation/multi-phase commit features
•    "Intelligent Network Services" is the key capability that needs to be developed
     next to support co-scheduling across network, middleware, application
     domains – Network API needed to make service available to workflow engines
•    OpenFlow/Software Defined Networking offers a set of network capabilities
     which can enhance these "Intelligent Network Services"
      •  but the "Intelligent Network Services" and co-scheduling technologies are the
         distinct and key value added feature set that we are addressing
ARCHSTONE: Intelligent Network Services for Advanced Application Workflows

More Related Content

PDF
QoS based Admission Control using Multipath Scheduler for IP over Satellite N...
PDF
Authentication of Secure Data Transmission In Wireless Routing
PDF
DETERMINING THE NETWORK THROUGHPUT AND FLOW RATE USING GSR AND AAL2R
PDF
Ab25144148
PDF
TOPOLOGY DISCOVERY IN MULTI-HOP CLUSTERING VEHICULAR NETWORKS
PDF
ECA MODEL BASED QOS AODV ROUTING FOR MANETS
PDF
A New Efficient Cache Replacement Strategy for Named Data Networking
PPT
Session18 Madduri
QoS based Admission Control using Multipath Scheduler for IP over Satellite N...
Authentication of Secure Data Transmission In Wireless Routing
DETERMINING THE NETWORK THROUGHPUT AND FLOW RATE USING GSR AND AAL2R
Ab25144148
TOPOLOGY DISCOVERY IN MULTI-HOP CLUSTERING VEHICULAR NETWORKS
ECA MODEL BASED QOS AODV ROUTING FOR MANETS
A New Efficient Cache Replacement Strategy for Named Data Networking
Session18 Madduri

What's hot (16)

PDF
AN EFFECTIVE CONTROL OF HELLO PROCESS FOR ROUTING PROTOCOL IN MANETS
PPT
Session 49 - Semantic metadata management practical
PDF
Dual stack approach ipv4 ipv6
PPTX
Qo s oriented distributed routing protocols : anna university 2nd review ppt
PDF
International Journal of Computational Engineering Research(IJCER)
PDF
Session9part2 Servers Detailed
DOCX
ORCHESTRATING BULK DATA TRANSFERS ACROSS GEO-DISTRIBUTED DATACENTERS
DOCX
cost effective resource allocation of overlay routing relay nodes
PDF
Fpga implementation of scalable queue manager
PDF
Ho2413421346
PDF
Load aware and load balancing using aomdv routing in manet
PDF
MuMHR: Multi-path, Multi-hop Hierarchical Routing
 
PDF
E010322531
PDF
Solving QoS multicast routing problem using ACO algorithm
PDF
Hadoop World 2011: Hadoop Gateway - Konstantin Schvako, eBay
AN EFFECTIVE CONTROL OF HELLO PROCESS FOR ROUTING PROTOCOL IN MANETS
Session 49 - Semantic metadata management practical
Dual stack approach ipv4 ipv6
Qo s oriented distributed routing protocols : anna university 2nd review ppt
International Journal of Computational Engineering Research(IJCER)
Session9part2 Servers Detailed
ORCHESTRATING BULK DATA TRANSFERS ACROSS GEO-DISTRIBUTED DATACENTERS
cost effective resource allocation of overlay routing relay nodes
Fpga implementation of scalable queue manager
Ho2413421346
Load aware and load balancing using aomdv routing in manet
MuMHR: Multi-path, Multi-hop Hierarchical Routing
 
E010322531
Solving QoS multicast routing problem using ACO algorithm
Hadoop World 2011: Hadoop Gateway - Konstantin Schvako, eBay
Ad

Similar to ARCHSTONE: Intelligent Network Services for Advanced Application Workflows (20)

PDF
Pervasive Computing
PDF
MobiSys Group Presentation
PPTX
MPLS/SDN Intersections Next Generation Access Networks at MPLS & Ethernet Wor...
PPTX
Scaling Diameter for LTE
PDF
Future Cities Conference´13 / Peter Steenkiste - "The eXpressive Internet Arc...
PDF
Cisco open network environment
PDF
OSS Presentation Keynote by Hal Stern
PPTX
FlowER Erlang Openflow Controller
PDF
Open source and standards - unleashing the potential for innovation of cloud ...
KEY
Scalable Architectures - Taming the Twitter Firehose
PDF
Presentazione laurea 1.2 matteo concas
PPTX
OpenFlow Beyond the Data Centre at IP Expo
PDF
Presentacio Fi2 Cat Sept 2008
PPTX
#lspe: Dynamic Scaling
PDF
Alcatellucentsdn2013
PDF
20130325 openstack-meetup
PPTX
Keynote -金耀辉--network service in open stack cloud-osap2012_jinyh_v4
PDF
Flexible Data Centre Fabric - FabricPath/TRILL, OTV, LISP and VXLAN
PDF
Networking is NOT Free: Lessons in Network Design
PDF
数据中心网络研究:机遇与挑战
Pervasive Computing
MobiSys Group Presentation
MPLS/SDN Intersections Next Generation Access Networks at MPLS & Ethernet Wor...
Scaling Diameter for LTE
Future Cities Conference´13 / Peter Steenkiste - "The eXpressive Internet Arc...
Cisco open network environment
OSS Presentation Keynote by Hal Stern
FlowER Erlang Openflow Controller
Open source and standards - unleashing the potential for innovation of cloud ...
Scalable Architectures - Taming the Twitter Firehose
Presentazione laurea 1.2 matteo concas
OpenFlow Beyond the Data Centre at IP Expo
Presentacio Fi2 Cat Sept 2008
#lspe: Dynamic Scaling
Alcatellucentsdn2013
20130325 openstack-meetup
Keynote -金耀辉--network service in open stack cloud-osap2012_jinyh_v4
Flexible Data Centre Fabric - FabricPath/TRILL, OTV, LISP and VXLAN
Networking is NOT Free: Lessons in Network Design
数据中心网络研究:机遇与挑战
Ad

More from Ed Dodds (20)

PDF
Updated Policy Brief: Cooperatives Bring Fiber Internet Access to Rural America
PDF
ILSR 2019 12 rural coop policy brief update page 8
PDF
Maximizing information and communications technologies for development in fai...
PDF
Iris Ritter interconnection map
PDF
Inoversity - Bob Metcalfe
PDF
Distributed Ledger Technology
PDF
UCX: An Open Source Framework for HPC Network APIs and Beyond
PDF
Digital Inclusion and Meaningful Broadband Adoption Initiatives Colin Rhinesm...
ODP
Jetstream
PDF
Innovation Accelerators Report
PDF
Work.
PDF
Strategy for American Innovation
PDF
Collaboration with NSFCloud
PDF
AppImpact: A Framework for Mobile Technology in Behavioral Healthcare
PDF
Report to the President and Congress Ensuring Leadership in Federally Funded ...
PPTX
Data Act Federal Register Notice Public Summary of Responses
PDF
Gloriad.flo con.2014.01
PDF
2014 COMPENDIUM Edition of National Research and Education Networks in Europe
PDF
New Westminster Keynote - Norman Jacknis
PDF
HIMSS Innovation Pathways Summary
Updated Policy Brief: Cooperatives Bring Fiber Internet Access to Rural America
ILSR 2019 12 rural coop policy brief update page 8
Maximizing information and communications technologies for development in fai...
Iris Ritter interconnection map
Inoversity - Bob Metcalfe
Distributed Ledger Technology
UCX: An Open Source Framework for HPC Network APIs and Beyond
Digital Inclusion and Meaningful Broadband Adoption Initiatives Colin Rhinesm...
Jetstream
Innovation Accelerators Report
Work.
Strategy for American Innovation
Collaboration with NSFCloud
AppImpact: A Framework for Mobile Technology in Behavioral Healthcare
Report to the President and Congress Ensuring Leadership in Federally Funded ...
Data Act Federal Register Notice Public Summary of Responses
Gloriad.flo con.2014.01
2014 COMPENDIUM Edition of National Research and Education Networks in Europe
New Westminster Keynote - Norman Jacknis
HIMSS Innovation Pathways Summary

Recently uploaded (20)

PDF
Electronic commerce courselecture one. Pdf
PPTX
20250228 LYD VKU AI Blended-Learning.pptx
PDF
Per capita expenditure prediction using model stacking based on satellite ima...
PDF
Review of recent advances in non-invasive hemoglobin estimation
PDF
Advanced methodologies resolving dimensionality complications for autism neur...
PPTX
A Presentation on Artificial Intelligence
PDF
Approach and Philosophy of On baking technology
PPT
“AI and Expert System Decision Support & Business Intelligence Systems”
PPTX
Cloud computing and distributed systems.
PDF
Modernizing your data center with Dell and AMD
PDF
Encapsulation_ Review paper, used for researhc scholars
PDF
KodekX | Application Modernization Development
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
PDF
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
PDF
Spectral efficient network and resource selection model in 5G networks
PDF
Machine learning based COVID-19 study performance prediction
PPTX
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
PPTX
Big Data Technologies - Introduction.pptx
PPTX
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
PDF
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows
Electronic commerce courselecture one. Pdf
20250228 LYD VKU AI Blended-Learning.pptx
Per capita expenditure prediction using model stacking based on satellite ima...
Review of recent advances in non-invasive hemoglobin estimation
Advanced methodologies resolving dimensionality complications for autism neur...
A Presentation on Artificial Intelligence
Approach and Philosophy of On baking technology
“AI and Expert System Decision Support & Business Intelligence Systems”
Cloud computing and distributed systems.
Modernizing your data center with Dell and AMD
Encapsulation_ Review paper, used for researhc scholars
KodekX | Application Modernization Development
Building Integrated photovoltaic BIPV_UPV.pdf
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
Spectral efficient network and resource selection model in 5G networks
Machine learning based COVID-19 study performance prediction
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
Big Data Technologies - Introduction.pptx
KOM of Painting work and Equipment Insulation REV00 update 25-dec.pptx
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows

ARCHSTONE: Intelligent Network Services for Advanced Application Workflows

  • 2. •  USC/ISI •  BNL •  Tom Lehman •  Dimitrios Katramatos •  Xi Yang •  Sushant Sharma •  ESnet •  Dantong Yu •  Chin Guok •  Eric Pouyoul •  Inder Monga •  Vangelis Chaniotakis •  Bharath Ramaprasad (UMass) •  UNM •  Nasir Ghani •  Feng Gu •  Kaile Liang
  • 4. •  Network "Service Plane" formalization •  Composable Network Service architecture •  ARCHSTONE Network Service Interface as client entry point •  Extensions to Topology and Provisioning Schemas to enable: •  multi-layer topologies •  multi-point topologies •  requests in the form of a "service-topology" •  vendor specific features •  technology specific features •  node level constraints •  MX-TCE (Multi-Dimensional Topology Computation Engine) •  Computation Process and Algorithms •  Enable a New class of Network Services referred to as "Intelligent Network Services" •  clients can ask the network "what is possible?" questions •  can ask for "topologies" instead of just point-to-point circuits
  • 5. •  Advanced Network Service Interface •  "Request Topology" and "Service Topology" concepts •  Common Network Resource Description schema •  Network Service Plane access point •  Multi-Dimensional Topology Computation Element (MX-TCE) •  High Performance computation with flexible application of constraints •  Use OSCARSv6 as base infrastructure and development environment request Network Network Service Provider Requester Agent OSCARSv6 reply MX-TCE Network Resource Description
  • 6. Network Services Interface Network Service Plane Composite Service (S1 = S2 + S3) Composite Service Composite Service (S2 = AS1 + AS2) (S3 = AS3 + AS4) Atomic Atomic Atomic Atomic Service Service Service Service (AS1) (AS2) (AS3) (AS4) Multi-Layer Network Data Plane
  • 7. Topology Service to determine Security Service (e.g. resources and orientation encryption) to ensure data integrity Resource Computation Service* Store and Forward Service to to determine possible resources enable caching capability in the based on multi-dimensional network constraints (*MX-TCE) Measurement Service to Connection Service to specify enable collection of usage data data plane connectivity and performance stats Monitoring Service to ensure Protection Service to enable proper support using SOPs for resiliency through redundancy production service Restoration Service to facilitate recovery
  • 8. •  Extensions to OSCARS v0.6 •  Added features for: •  multi-layer topologies •  multi-point topologies •  requests in the form of a "service-topology" •  vendor specific features •  technology specific features •  node level constraints •  Result is a schema "Superset" to what OSCARSv0.6 now uses •  schema with ARCHSTONE extensions will be backward compatible with current OSCARS operations •  Uses perfSONAR Topology and Lookup Services
  • 9. •  MX-TCE role in OSCARS •  perform basic path computation for current OSCARS service •  standalone Topology Computation element to: answer "what is Client possible?" questions for clients to subsequently make request for Application OSCARS services •  Advanced multi-layer and multi-point computations What topologies are available next week for 3 hour duration? Current OSCARS Service: Here's Two "Please try and reserve a 5 Gbps VLAN circuit between A and B at 15:00 on January 16, 2012 for 3 hour duration" MX-TCE Client Application can perform basic path computations as needed to respond to current OSCARS point-to-point VLAN service
  • 10. java web service, C++ multi-threaded core
  • 11. •  Unified API/NSI support for P2P, Multi-Point, Multi-Layer, schedule and co-scheduling requests under extended NML schema •  Implemented OSCARS PCE API to become swappable OSCARS module •  support existing OSCARS PCE capability as a single TcePCE •  support co-scheduling via optionalConstraint extension •  Multiple path and topology computation workflows •  kicked off based on request types: P2P, MP, MLN/MRN, coSheduling etc. and combinations •  support concurrent requests through multi-threading •  Modularized differentiated algorithm execution driven by workflows •  Transform computation results into •  provisioning friendly path object depending on path control scenarios •  NSI and OSCARS compliant reply messages
  • 13. •  ARCHSTONE Extensions •  Layer Decisions •  Multi-Layer Topology Representations •  driven by resource constraints or •  Multi-Layer Topology Computations client requests for specific •  Multi-Layer Provisioning performance characteristics (i.e. low latency, low jitter, etc)
  • 14. •  Prototype Deployment on Production Networks •  ESnet and Internet2 •  Real-time reservations being processed to provide answers to "what is possible?" questions that client can then use to make requests on the operational networks •  VNOD project will discuss a specific use case for this Client Application coScheduleRequest: start time: Wed, 08 Feb 2012 05:00:00 GMT end time: Wed, 08 Feb 2012 07:00:00 GMT min bandwidth: 1 Gbps max number of path options: 3 format for reply: bandwidth availability graph OSCARS MX-TCE OSCARS real-time reservations feed from production networks
  • 15. •  archstone.east.isi.edu •  Architecture and Design Documents •  MX-TCE Software •  Extensions to OSCARS Topology and Request Schemas •  Example topology descriptions, service requests, service topologies (responses) •  ANI Testbed configuration and use •  OSCARSv0.6 project •  code.google.com/p/oscars-idc/ •  OSCARS ARCHSTONE Branch •  oscars.es.net/repos/oscars/branches/archstone/
  • 16. •  Virtual Network On Demand (VNOD) is a project funded by DOE ASCR •  Objectives: •  To provide on-demand, end-to-end virtual network topologies dedicated to servicing communities of users at multiple end-sites •  To facilitate the creation and management of such virtual network domains -  VNOD GUI based front-end •  To provide a platform for applying various optimization schemes with diverse objectives -  Successfully schedule multiple, fixed and flexible, end-to-end data transfer requests between multiple end sites and hosts •  Currently working on developing technologies to address the integration and co-scheduling across Network and Middleware level resources •  This includes co-scheduling algorithms, protocols, and workflows which can operate in this "distributed heterogeneous multi-resource environment" •  Leveraging TeraPaths and ESCPS for end-host and site communications and configurations •  Future work will include integration with application specific workflows and systems •  Utilizing ARCHSTONE Intelligent Network Services as part of co- scheduling workflows
  • 17. End Site End Site End Site ESDC Transit WAN IDC Transit WAN Transit WAN End Site End Site
  • 18. Topology Community Topology distributed auxiliary Community Service Membership Service Membership services service service Service Service Discovery Monitoring Discovery Monitoring Service Service ViNet request Resource Scheduler (RS) Resource Scheduler (RS) (application (selects and schedules (selects and schedules requirement resources) resources) profile) VNOD VNOD remote RSs remote RSs ViNet Controller (VNDC) ViNet Controller (VNDC) (creates and manages (creates and manages remote VNDCs ViNets) ViNets) remote VNDCs virtual networking layer virtual end-to-end path layer remote ESCPS remote ESCPS negotiation ESCPS ESCPS Negotiation OSCARS OSCARS with ARCHSTONE with ARCHSTONE Transit WANs End-Site LAN End-Site LAN
  • 20. •  VNOD workflow: •  User requests a virtual domain setup via the front-end -  Request can come from an application via the API -  End-system resource availability (e.g., storage system bandwidth) can also be submitted for co-scheduling •  The VNOD back-end: -  Collects bandwidth availability information from domains involved in virtual path establishment •  Interfaces with ARCHSTONE for WAN domains and TeraPaths/ ESCPS for end-sites -  Runs scheduling algorithm(s) -  Makes appropriate reservation requests based on algorithm output by engaging ARCHSTONE and TeraPaths/ESCPS
  • 21. + End Site A + Transit WAN W1 Transit WAN W2 or + End Site or B + +
  • 22. (a) (b) (c) (d)
  • 23. Available Bandwidth (Gb/s) Step 1 10 Step 3 8 Step 0 Request #2 6 Step 2 4 Request #1 Request #3 2 Request #0 0 1 2 3 4 5 6 7 Time (secs)
  • 24. •  Implementation in progress •  Increasingly difficult scheduling problems: •  Single request/single path (TeraPaths/StorNet) •  Multiple requests/single path/single WAN domain •  Multiple requests/multiple paths/single WAN domain •  Multiple requests/multiple paths/multiple WAN domains •  Co-design with ARCHSTONE to attack scheduling complexities •  Hierarchical scheduling (end-sites/WAN) •  Tighter integration and interoperation
  • 25. •  The architecture adopted by ARCHSTONE, OSCARS, VNOD: •  Centralized at the Intra-Domain level for resource management and service provisioning •  Distributed at the Inter-Domain level for resource management and service provisioning •  External topology distribution systems must limit the amount dynamic data exported (scalability and stability issues) •  Resource identification for real-time service provision can only be done by local domain systems •  Multi-domain service provision will require chain or tree mode protocols which include real-time negotiation/multi-phase commit features •  "Intelligent Network Services" is the key capability that needs to be developed next to support co-scheduling across network, middleware, application domains – Network API needed to make service available to workflow engines •  OpenFlow/Software Defined Networking offers a set of network capabilities which can enhance these "Intelligent Network Services" •  but the "Intelligent Network Services" and co-scheduling technologies are the distinct and key value added feature set that we are addressing