1 / 15

Paola Grosso (SLAC) and Warren Matthews (GA Tech)

Measurement and Fault-Finding Using MAGGIE and PIPES. Presented at the HENP SIG Internet2 Members Meeting, Indianapolis, October 2003. Paola Grosso (SLAC) and Warren Matthews (GA Tech). Background. There are a large number of tools available to help diagnose network problems.

truda
Download Presentation

Paola Grosso (SLAC) and Warren Matthews (GA Tech)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Measurement and Fault-Finding Using MAGGIE and PIPES.Presented at the HENP SIGInternet2 Members Meeting, Indianapolis, October 2003. Paola Grosso (SLAC) and Warren Matthews (GA Tech)

  2. Background • There are a large number of tools available to help diagnose network problems. • Many networks have a network monitoring project, often with a web page showing graphs. Internet2 Members Meeting, Indianapolis, October 2003.

  3. Toward a monitoring Infrastructure • Grid tools and data intensive science require a cross-domain monitoring infrastructure with standardized publishing. • Resource Brokers • DoE science community • Monitoring Infrastructure • IEPM-BW and MAGGIE (SLAC and DoE) • AMI and PIPES (Internet2) Internet2 Members Meeting, Indianapolis, October 2003.

  4. IEPM-BW (Only routes from SLAC crossing Abilene are shown) SLAC SLAC APAN RIKEN Stanford INFN-Roma APAN Geant INFN-Milan CENIC Abilene SEA CESnet NY ATL SNV HSTN SOX CLV IPLS UTDallas I2 UFL UMich Rice NCSA Internet2 Members Meeting, Indianapolis, October 2003.

  5. AMI STTL CHIN DNVR NYCM IPLS SNVA KSCY WASH ATLA LOSA HSTN Internet2 Members Meeting, Indianapolis, October 2003.

  6. Mysterious low throughput peak. Human intervention required. Internet2 Members Meeting, Indianapolis, October 2003.

  7. Measurements indicate a very tight distribution of delays across Abilene By utilizing other measurement projects, problem location can be narrowed. Graph contains 99.31% of the data. Tail (not shown) contains 0.69% of the data. Internet2 Members Meeting, Indianapolis, October 2003.

  8. Publishing • Unfeasible to look at (possibly hundreds) of graphs. • Publish the data in standard way • W3C Web Service, GGF OGSI Grid Service • Currently XMLRPC and SOAP servers Internet2 Members Meeting, Indianapolis, October 2003.

  9. Demo Client: /usr/local/bin/perl use SOAP::Lite; my $web_service = SOAP::Lite -> service('http://ndb1-blmt.abilene.ucaid.edu/aep/nmwg.wsdl') -> NetworkMeasurement('V4', ’SNVA', 'IPLS', '20031001000000', '20031002000000'); print $web_service->{resultSet}->{result}->{summaryData}->{minimum},"\n"; print $web_service->{resultSet}->{result}->{summaryData}->{average},"\n"; print $web_service->{resultSet}->{result}->{summaryData}->{maximum},"\n"; Response: 22.3 22.4 82.5 (Soap::lite extracts the data from the XML) Internet2 Members Meeting, Indianapolis, October 2003.

  10. Visualization • Numerous tools are being developed to assist with visualizing measurements • Monalisa, Advisor, Internet Detective, Anemos • Some will leverage 3rd party measurements via web services Internet2 Members Meeting, Indianapolis, October 2003.

  11. Fault Finding • Tools will (or should) provide some automated fault-finding as a starting point for human intervention • Use techniques such as Tony McGregor’s plateau concept for AMP and RIPE-TT rolling average to determine unusual behaviour. • Write to log Internet2 Members Meeting, Indianapolis, October 2003.

  12. Trouble Detection $ tail maggie.log 09/29/2003 14:58:47 (1:14) gnt4 0.51 Alarm (AThresh=38.33) 09/29/2003 16:25:45 (1:16) gnt4 3.83 Concern (CThresh=87.08) 09/29/2003 17:55:21 (1:17) gnt4 169.57 Within boundaries Status Date and Time Bin Node Throughput (iperf) Internet2 Members Meeting, Indianapolis, October 2003.

  13. Net RAT • In addition to quantitative measurements from ping, iperf etc • NetRAT will provide qualitative classification of nodes, hops and routes. • Cross-referencing data will identify hotspots (hop.performance = “poor”). Internet2 Members Meeting, Indianapolis, October 2003.

  14. Summary • Sites, Networks, Grids must interoperate to meet the needs of modern science and demanding applications. • A Monitoring Infrastructure consisting of measurements from many projects published in standard ways is required. Internet2 Members Meeting, Indianapolis, October 2003.

  15. Links • IEPM Home Page • Internet2 e2epi and PIPES home page • OWAMP data • GGF NMWG • Monalisa, Advisor,Internet Detective,Anemos • AMP Event Detection Internet2 Members Meeting, Indianapolis, October 2003.

More Related