MASSACHUSETTS INSTITUTE OF TECHNOLOGY HAYSTACK OBSERVATORY WESTFORD, MASSACHUSETTS 01886 Telephone: Fax: 978-692-4764 781-981-0590 12 July 2002 TO: Distribution FROM: Alan R. Whitney SUBJECT: 11 July 2002 e-VLBI telecon summary Attendees: Lee Foster, Bill Fink, Pat Gary, Kevin Kranacs, Paul Lang – GSFC Steve Bernstein, Lorraine Prior, Peter Schultz – Lincoln Laboratory Tom Lehman – ISI-E Richard Crowley, Hans Hinteregger, Arthur Niell, Mike Titus, Alan Whitney – Haystack Observatory This telecon is one of an ongoing series of telecons to prepare for gigabit/sec e-VLBI demonstrations between NASA GSFC and MIT Haystack Observatory using a combination of network facilities including all or part of Glownet, Bossnet, ISI-E, SuperNet, Max and GSFC/HECN. Status Reports The attached figures of the e-VLBI path have been updated to reflect current status and are pretty much self-explanatory; critical status items are indicated in red. In addition, the following comments are relevant: − Peter reported that there is still a problem with the OC-48 Bossnet service and LL is short of personnel to address the problem immediate. As a result, Bossnet will be maintained as a GigE link until at least August. The GigE link seems to be working properly and is suitable for the e-VLBI demonstration; however, the Bossnet GigE link is shared with MIT campus, so Bossnet use for e-VLBI needs to be scheduled (via Bossnet scheduling calendar) so that campus activity can be suspended while e-VLBI testing takes place. The Bossnet problem appears to be somewhere in the path between New York and Newark, but is not yet isolated. − Router configurations at G10, H3 and H7 have been updated, as promised at the last meeting, for e-VLBI use. − Test workstation at K6 is now named ‘clifford’ (not ‘dawg’). Alan will update on workstation specification list. − Lorraine reported that another LL demonstation using Glownet is scheduled for 22 July, during which time Glownet will be unavailable for e-VLBI. − Haystack target date is tentatively August for full demonstration. minutes15.doc 7/12/2002 1 Performance Testing Bill F. requested that all test workstations install a nuttcp server to facilitate ease of testing. He will distribute a note to everyone with details of the installation procedure (has been done). Kevin D. conducted several short tests from Haystack (‘evlibhay’) to GSFC/GGAO (‘pluto’), achieving ~350 Mbps throughput. Below is a summary of his results: Configuration MIT(evlbihay): ip 140.173.174.4 mtu 4470 (9000 max) mss 4418 /proc/sys/net/core/wmem_max 8388608 /proc/sys/net/core/rmem_max 8388608 /proc/sys/net/ipv4/tcp_timestamps 1 /proc/sys/net/ipv4/tcp_sack 1 /proc/sys/net/ipv4/tcp_wmem "4096 1048576 8388608" /proc/sys/net/ipv4/tcp_rmem "4096 1048576 8388608" nuttcp v3.1.9 Configuration GGAO(pluto): ip 206.196.178.53 nuttcp v3.2.1 Rtt: ~ 15 msec Throughput: ~ 350 Mb/s (Haystack to GGAO) Note: Only very short duration (1-15 secs) tests were done. The following was typical output: [kad@evlbihay bin]$ ./nuttcp -t -l4418 -n16384 -w2048 -ws2048 206.196.178.53 nuttcp-t: v3.1.9: socket nuttcp-t: buflen=4418, nbuf=16384, nstream=1, port=5001 tcp -> 206.196.178.53 nuttcp-t: connect to 206.196.178.53 nuttcp-t: send window size = 4194304, receive window size = 1048576 nuttcp-t: 69.031 MB in 1.64 real seconds = 43007.87 KB/sec = 352.3205 Mb/s nuttcp-t: 16384 I/O calls, msec/call = 0.10, calls/sec = 9968.33 nuttcp-t: 0.0user 0.3sys 0.01real 21% 0i+0d 0maxrss 0+0pf 0+0csw nuttcp-r: v3.2.1: socket nuttcp-r: buflen=4418, nbuf=16384, nstream=1, port=5001 tcp nuttcp-r: accept from 140.173.174.4 nuttcp-r: send window size = 16384, receive window size = 2097152 nuttcp-r: 69.031 MB in 1.68 real seconds = 41996.25 KB/sec = 344.0333 Mb/s nuttcp-r: 69.031 MB in 0.22 CPU seconds = 321309.09 KB/cpu sec nuttcp-r: 16467 I/O calls, msec/call = 0.10, calls/sec = 9783.16 nuttcp-r: 0.0user 0.2sys 0:01real 13% 0i+0d 0maxrss 1+1pf 0+0csw Kevin D. feels that there may some tuning issues on ‘pluto’, but he was unable to get a password to make tuning changes. Bill F. will work with Kevin on ‘pluto’ tuning issues. Tom also suggested trying a multi-stream TCP process to improve aggregate throughput rate (easily done with nuttcp). nuttcp is not yet ported to Windows. Lee reports that a summer student is currently working on this project. Tom will investigate possibility of putting nuttcp on MAX test workstations, but warned that these machines are not particularly high-performance machines so probably may not be able to draw useful performance conclusions. minutes15.doc 7/12/2002 2 Tom and Bill will coordinate increasing mtu size on paths between ISI-E and MAX. The issue is complicated by the use of ‘frame relay encapsulation’ on some of these links and links to Abilene. Action Items Bill F: Distribute information on nuttcp routine (has been done). Tom/Bill F.: Will coordinate installation of nuttcp on ISI-E test machines (Tom will be on vacation for next two weeks, but will be available via e-mail occasionally during this time). Bill will then do testing on ISI-E/GSFC link and report to group. Tom/Bill F.: Will coordinate mtu change on ISI-E/MAX links. Tom: Will investigate possibility of putting nuttcp on MAX test workstations. Kevin D./Bill F.: Will coordinate to allow Kevin access to ‘pluto’ (requires Kerberos account). Next telecon Next telecon will be Wed, 7 Aug 2002 at 2 pm. xc: Steve Bernstein, LL Jim Calvin, LL Lorraine Prior, LL Leslie Weiner, LL Herbert Durbeck, GSFC Lee Foster, GSFC Pat Gary, GSFC Chuck Kodak, GSFC Kevin Kranacs, GSFC Paul Lang, GSFC Aruna Muppalla, GSFC Bill Wildes, GSFC Dan Magorian, UMCP Tom Lehman, ISI Jerry Sobieski, Max Richard Crowley, Haystack Kevin Dudevoir, Haystack Hans Hinteregger, Haystack Arthur Niell, Haystack Joe Salah, Haystack minutes15.doc 7/12/2002 3 Westford Antenna Millstone RPE Bldg (Lorraine) Dedicated no scheduling GigE WS/ Mark 5 Summit 1i 1 A2 A1 GigE 8 Alpine A4 3804 A3 Dedicated no scheduling 28-1A 14 B1 Summit 15 WS/ Mark 5 100 Mb 28-1B B8 B4 B6 B7 Hub B10 B9 To HS switch B3 Corr Cntl Cmptr GlowNet Congress St Boston (~50 km) Amp E6 E3 GigE E2 E4A Test WS 3l (2 GigE + 1 OC-48) GlowNet Alpine 3804 E4 GigE Test WS Bossnet (Steve/Jim) ISI-E Scheduling on Bossnet calendar GigE Juniper M20 1310 E7 E8 GigE E5 (~100 km) D1 Shared with MIT traffic Cisco 1500 E1 D2 Lincoln Lab (Lorraine/Steve/Jim) Schedule to be coordinated with LL Cisco 1500 GigE B11 D3 WDM E5A WDM GigE 5i 13 (Shared with LL on scheduled basis) C3 16 GigE GigE Dell Poweredge 2500 (’evlibhay’) GigE B5 B2 LL C2 Haystack Correlator WS/ Mark 5 Schedule to be coordinated with LL C1 MIT Campus WDM F1 E9 Other users (no control) To be upgraded to OC-48 155x Amp Amp F2 Bossnet currently running GigE pending investigation and repair (’superglide’) (’evlbill’) From ISI-E Dell Poweredge 500SC (provided by Haystack) Figure 1: e-VLBI Path - Haystack to ISI-E evlbi011.cdr 12 July 02 ISI-E (Tom) Test WS Scheduling on Bossnet calendar Test WS G5 G7B G6 GE G6B (’kame’) G8 OC-48 WDM Summit 5i (’omega’) Tuner Bossnet GE G7 MAX @ ISI-E (Jerry) G6C ponder G1 OC-48 Switch GigE H2 GigE G3 G11 OC-48 Frame relay encapsulation H4 H3 Juniper M160 GigE H8 GigE H9 Primary e-VLBI path Needs configuration J3 Test WS? J2 Other users ( no control) Cisco 4912 Switch GigE Summit 5i H10 GigE J5 J4 WDM GigE on loan from GSFC H11 2 GigE l’s LuxN J6 J1 Need details Needs configuration H6 MAX/UMCP (Jerry/Dan) No jumbo frames OC-48 Juniper M160 Borrowed OC-48 interface - may not be able to keep G10 Needs configuration H7 No scheduling; sufficient capacity for e-VLBI Backup plan Juniper M40 G9 MAX performance node; need details Test WS? H5 H1 G12 GigE GigE Trans- G2 Summit 5i G4 LX (~14km) GSFC/Bldg 28 (Pat/Paul/Kevin/Bill F.) K4 K1 WDM Will try to schedule either K2 or K3 exclusively for e-VLBI experiment G4 Mac with Yellowdog Linux (’clifford’) GGAO/Bldg 201 (Pat/Chuck) GigE K2 GigE LX (~7.25km) 1310 K7 Summit 5i K3 K5A K5 Test WS Test WS K6 K6A GGAO Antenna Trailer (Bill W./Chuck) Dedicated - no scheduling necessary (~500m) GigE Summit L1 5i K5B Cisco 12016 K6B L2 GigE L3 L5 G4 Mac (’pluto’) SX Mark 5 L4 Test WS L6 Windows PC (’xly’) evlbi012.cdr Figure 2: e-VLBI Path - ISI-E to GSFC/GGAO 12 July 02