Anomaly Case Studies

August 4, 2006:Problems accessing S. African host 04 August '06' Denial of Service attack, followed by loss of power UNDERSTOOD.
July 25, 2006:Problems with BINP-SLAC link July '06 Power problems in Akademogorock region causes many services to be unavailable from SLAC to BINP -UNDERSTOOD.
May 18, 2006:Poor throughput for GLAST data transfers between SLAC and NRL Washinton Too small TCP windows set at default, firewall limiting throughput, application needs to apply ssh throughput patches, mandated application cannot use multiple streams or big windows, alternate path had port blocking. -UNDERSTOOD.
May 14, 2006:Throughput between SLAC and TaiwanTCP windows and number of parallel styreams needed optimizing. -UNDERSTOOD.
May 9, 2006:Large increase in RTT from South Africa to CERN April 2006 Change in routing traffic from S. Africa to CERN from using GEANT to going via Abilene and the U.S. after Abilene started exchanging traffic with S. African TENET -UNDERSTOOD.
February 1, 2006:Problems with Oracle between SLAC and ORNL, Jan 2006 Appears to be an application problem -UNDERSTOOD.
January 22, 2006:Effect of ESnet routing change between SLAC and SDSC San Diego Sunnyvale connectivity was interrupted by Level(3) maintenance. An alternate path was used. This study looks at the impact on the various IEPM metrics of the route change.UNDERSTOOD.
December 28, 2005:Problems with link to and from BNL Multiple paths seen from and to BNL showed step downs in pathchirp performanceBelieved to be related to the host at BNL - NOT RESOLVED.
September 10, 2005:Problems with Stanford's Connectivity Fan failure in Cisco 15454 causes route to CENIC/Sunnyvale to fail, and switch over to using ESnet.UNDERSTOOD.
August 16, 2005:Problems with link to and from BNLFiber cut causes OC48 outage and fail over to a T3 link_UNDERSTOOD_.
July 1, 2004:Fiber outage to Pakistan June 27th to July 8, 2005Fiber cut off coast of Karachi causes 12 day outage to Pakistan_UNDERSTOOD_.
May 31, 2005:New Route from SLAC to BINP Novosibirsk via GLORIADSee the effect of the new route in improving RTT and throughput_UNDERSTOOD_.
April 18, 2005:Loss of Connectivity between SLAC and BINP NovosibirskSome change caused the connectivity between SLAC and BINP to be blocked by firewalls in Novosibirsk.
September 2, 2004:Loss of Connectivity from SLAC for about 10 hours Due to CENIC maintenance SLAC lost conectivity to many sites from around midnight until 10:30am September 2, 2004. This is nicely illustrated in traceroutes taken each 10 minute interval to multiple sites from SLAC and in particular can be nicely visualized by means of a topology map. -UNDERSTOOD.
February 25, 2004: Network performance to Pakistan Shows Internet performance from SLAC to a few Pakistani sites, including: differences in performance depending on providers; identifying bottlenecks; effects of upgrade for NIIT; routing between Pakistani sites going via London. -UNDERSTOOD.
January 8, 2004: Effect of routing on RTTs and losses to 2 sites in Hangzhou, China Site with better Internet connectivity had much larger RTTs, but lower losses -UNDERSTOOD.
October 3, 2003: Throughput from SLAC to Caltech reduced by a factor of 5 CENIC's routing to Caltech was incorrect -RESOLVED.
July 23, 2003: Route to Sox (Southern Crossroads) has been changed, the bandwidth increased in factor 2 Abilene routing, traceroute analysis -RESOLVED.
June 12, 2003: Effect of change from Internet2 to ESnet for the routes from SLAC to Europe 2 problems, routing for IN2P3, incorrect iperf configuration for CESnet -RESOLVED.
June 10, 2003: One of our important partner Daresbury lab (UK) upgraded line to Internet to 1GBitper second Routing -RESOLVED.
June 6, 2003: Routing changes hit many sites in US Routing -RESOLVED.
May 22, 2003: SLAC to CERN RTT Anomalies Routing -RESOLVED.
May 20, 2003: Loss of connectivity between SLAC and Novosibirsk Misconfigured router -RESOLVED.
May 7, 2003: The unusual bandwidth drop to European sites (connected via Abilene).
May 1, 2003: The influence of strong LAN traffic on target hosts in monitoring results (NERSC,ANL) Effect of local NFS traffic- RESOLVED.
November 6, 2002: Change of Routing to NCSA and others Change of routing -RESOLVED
September 20, 2002: Change of Routing Network connectivity problems - reports from BWESLAC-UK, and SLAC-ANL- RESOLVED.
July 26, 2002: Network connectivity between SLAC and BINP, Novosibirsk following upgradeResults of effectiveness of network upgrade -RESOLVED.
May 30, 2002: Change of Routing between UK and US Routing Change -RESOLVED.
Mar 21, 2002: Poor throughput from SLAC to INFN,Rome Poor packet loss -RESOLVED.
Jan 4, 2002: A sudden change in bandwidth utilization Throughput from monitoring host suddenly dropped from several hundreds of Mbits/s to about 100Mbits/s due to turning on debugging flags -RESOLVED.
Jan 8, 2002: Problems with performance between SLAC and IN2P3, Jan '02 Only ~ 20Mbits/s throughput achievable on 30Mbps ATM VP bottleneck -RESOLVED.
Dec 13, 2001: Problems with performance between SLAC and CERN Large router buffers adversely affect interactive use -RESOLVED.
Sep 27, 2001: Nework connectivity between SLAC and RL, UK Poor performance from SLAC to Rutherford Lab, UK caused by firewall and small windows -RESOLVED.
Sep 19, 2001: Problems with links from SLAC to IHEP.SU High packet loss prevents ssh access, due to host problems -RESOLVED.
Sep 11, 2001: Impact of World Trade Center & Pentagon attacks on Internet performance How did the WTC collapse affect Internet connectivity for SLAC -RESOLVED.
May 3, 2001: Qbone Scavenger Service at SLAC Need to find hosts/sites to try with QBone Scavenger services -RESOLVED.
Mar 28, 2001: International Internet connectivity from SLAC Request for information on quality of International links -RESOLVED.
Mar 19, 2001: Comparing ihep.ac.cn with cas.ac.cn - March 2001 Request to understand the proposed effect of changing an ISP -RESOLVED.
Mar 1, 2001: Packet loss on Stanford-Pac Bell DSL links - Mar 01 Feb 25, 2001:Packet loss on Stanford-Pac Bell DSL links - Feb 01Problems with users DSL service due to bad pairs to home -RESOLVED.
Jan 27, 2001: Throughput between Tersk03 and Kangadata Problems in host to host throughput on LAN due to windows being to small -RESOLVED.
Jan 22, 2001: Network connectivity between SLAC and IN2P3, Jan 2001 Problems with throughput between SLAc and IN2P3 due to misconfigured ATM/PVC -RESOLVED.
Jan 12, 2001: Network connectivity between SLAC and Platform Computing Problems between SLAC and a commercial site due to poor routing -RESOLVED.
Dec 18, 2000: Network connectivity between SLAC and UT Dallas Multiple probelms with throughput between SLAC and UTDallas due to small windows and packet loss -RESOLVED WINDOW PROBLEM.
Dec 07, 2000: Network connectivity between SLAC and IN2P3, Lyon, France User question on suitability of network connections to BaBar tier A regional computing site -RESOLVED.
Dec 06, 2000: Network connectivity between SLAC and BINP, Novosibirsk Investigation of poor throughput from SLAC to Novosirbirsk, resulting in successful recommendation for link upgrade -RESOLVED.
Sep 11, 2000: Caltech to SLAC thruputPoor throughput from Caltech to SLAC while SLAC to Caltech was OK partially RESOLVED by moving from ESnet to Internet 2 -PARTIALLY RESOLVED.
Sep 03, 2000: SLAC to ILAN (Israel) thruput Poor performance between SLAC and Israel caused by ESnet upgrade to router using incorrect BGP configuration -RESOLVED.
May 16, 2000: Stanford to SLAC file transfer problems Poor throughput between SLAc and Stanford campus, required resetting an ATM interface -RESOLVED.
May 09, 2000: UK to US QoS Testing QoS did not appear to be working, the direction it was applied in was not congested -RESOLVED.
May 02, 2000: Throughput Performance between SLAC and CERN Big asymmetry in the throughputs from SLAC to IN2P2 and vice versa, partially due to Cisco web cache -NOT RESOLVED.
Apr 15, 2000: Troubleshooting VPN Disconnects Mainly some hints on tracking Microsoft VPN problems.
May 23, 1999: Ping Measurement Pathology Examples Pings from Linux to WNT on a LAN with normal RTT < 10 msec, manifest an unusual regularity in the sequence number separation of the pings with > 10msec. Probably an OS effect. -RESOLVED