Internet2%20End-to-End%20Performance%20Monitoring%20Initiative%20Update

1
Internet2 End-to-End Performance Monitoring
Initiative Update

• Eric L. Boyd and Jeff W. Boote, Internet2
• August 6, 2003

2
E2E piPES

• Project End-to-End Performance Initiative
  Performance Environment System (E2E piPES)
• Goal To allow end-users and network operators
  to determine performance capabilities, locate
  problems, and contact the right person to get a
  problem resolved.
• Approach Collaborative project combining the
  best work of many organizations, including
  DANTE/GEANT, NLANR/DAST, SLAC, UCL, etc.

3
BNI AMI

• Project Backbone Network Infrastructure Abilene
  Measurement Infrastructure
• Goal Instrument Next Generation Abilene network
  with extensive performance measurement
  capabilities
• Approach Collaborative project among E2E, BNI,
  and Indiana NOC.

4
E2E piPES Architecture
5
Intra-PMP-Module Protocol
6
Intra-PMP Module Components

• Domain Interface Web Service Interface to
  Request Performance Data
• Performance Measurement Controller (PMC)
  Schedules Tests
• Performance Measurement Point (PMP) Performs
  Tests, Stores Results in Database
• Source Initiates Test Request
• Target Accepts Test Request Starts Test

7
Domain Interface

• Request Interface
• Accepts External Result Requests
• Compares Requestor Role to Policy
• Rejects Request or Queries Response Interface
• Response Interface
• Accepts Result/Tool Requests
• Compares Requester Identity, Source Role to
  Policy
• Decides if Tool is Available
• Rejects Request or Supplies Capability

8
Initiator / Acceptor Performance Measurement
Controller

• Initiator PMC
• Supplies Capability, Identity, Tool
• Acceptor PMC
• Accepts/Rejects/Delays Request Based on Policy
• Contacts Target PMP to Initiate Test
• Accepts/Rejects Request Based on PMP Response

9
Target / Source Performance Measurement Point

• Source PMP
• Accepts/Rejects Requests to Start Test based on
  Identity
• Starts Tests
• Target PMP
• Accepts Test from Source PMP
• Stores Results Locally
• Sends Data to DB Gatekeeper

10
Database Gatekeper

• Accepts/Rejects Requests to Store Data based on
  Identity
• Accepts/Rejects Requests to Release Data based on
  Role, Identity
• Supplies Performance Data

11
E2E piPES Architecture v1.0
12
PMC Distributed Scheduling Algorithm (e.g. iperf)
13
piPEs Tool Deployment on Abilene

• All tests on IPv4 and IPv6
• OWAMP Deployed on 10/11 nodes (nms4)
• IPERF UDP In deployment beta on 2 nodes (nms1)
• IPERF TCP In deployment beta on 2 nodes (nms1)
• Traceroute In deployment beta on 2 nodes (nms4)
• Router Data Deployed on all 11 nodes (router
  interface)
• Flow Data Deployed on 10/11 nodes (nms3)

14
piPEs / AMI Rollout (Beta)
15
piPEs / AMI Rollout (Future)
16
piPEs / AMI Rollout (Future)
17
Performance Data as a Web Service

• We wish to create a web service that will allow
  other applications to request and receive test
  results.
• Possible Consumers
• Testing/Analysis Engine
• Web-based Display Engine
• High Energy Physics Routing
• Application Problem Detection

18
E2E piPEs Architecture (Grid)
19
Demo

• http//abilene.internet2.edu/observatory
• http//abilene.internet2.edu/owamp/status_map.cgi

20
Questions?

21
www.internet2.edu
22
Additional Slides to Answer Questions if Needed
23
piPEs / AMI Development Testing Environment
24
Abilene OWAMP deployment

• 2 overlapping full meshes (IPv4 IPv6)
• 11 measurement nodes 220 ongoing tests
• UDP singletons
• Rate 10 packets/second
• Packetsize (32 byte payload)
• Results are continuously streamed back to
  Measurement Portal for long-term archive and
  data dissemination (Near real-time)

25
piPEs / AMI Internal Resources

• Eric Boyd piPEs development, project
  coordination
• Jeff Boote OWAMP, piPEs development
• Prasad Calyam piPEs development
• Raj Swaminathan piPEs development
• Chris Heermann AMI
• Matt Zekauskas - AMI
• Susan Evett - Documentation
• Russ Hobby Campus Deployment
• George Brett Schema, Grid Service

26
piPEs External Resources

• Yee-Ting Li UCL (50)
• Paul Mealor UCL (up to 100)
• Warren Matthews SLAC
• Jim Ferguson et al NLANR / DAST
• Nicolas Simar et al DANTE
• Many early adopter volunteers

27
Testing / Analysis Engine

• Problem Encode Matt Z.s Brain
• Open Questions
• What are measurement results are acceptable for
  a given application (family)?
• What tools generate those results?
• How do you handle incomplete data?
• How do you rank multiple result-generators?
• What is the iterative decision tree to understand
  E2E problem?

28
Measurement Schema

• Measurement Types?
• e.g. BW, Latency, Loss, Jitter
• Measurement Units?
• e.g. Seconds vs. Microseconds
• Map Tools to Measurements?
• e.g. OWAMP -gt 1-way Latency
• Measurement Metadata?
• Database Table Design?

29
Access, Authentication, Authorization

• Roles for Access
• Standard End-User?
• Near Neighbor (Test Buddy)?
• NOC Staff / Network Engineer?
• Shibboleth for Implementation
• Each campus decides who in each role
• Others trust campus designation

30
Other RD Issues

• Design system to scale (eliminate
  centralization)?
• How to discover PMPs?
• How to insert On-Demand tests into
  Regularly-Scheduled test set
• Balance centralization and distributed database
  requirements