dpk@BRL.ARPA (Doug Kingston) (07/17/87)
Internet performance has soared to some of the best I have see all year. I have been able to successfully access Berkeley and UCL is finally available again. I would like to thank whoever is responsible for the recent change of events and to suggest that perhaps a status update is in order. Putting away my UUCP, -Doug-
CERF@A.ISI.EDU (07/17/87)
Doug, Voodoo networking! (George Bush theme for the 88 campaign?) Maybe someone sacrificed a chicken over RFC 1009? Added some links? Your puzzlement points out that we lack any global view of the Internet and don't even have a good way to figure out when we should report changes we've made in some part of the system or to whom such changes should be reported. Sounds like a good INENG topic, to me. Vint
scohn@ALEXANDER.BBN.COM (Steve Cohn) (07/17/87)
Here's something that changed. BBN modified the metric used for calculating SPF routes in the Arpanet on Sunday, 12 July. A detailed description and analysis of how it works and a report on measured changes in performance in the Arpanet will be forthcoming eventually. In brief, this change in the metric leads to more stable routing behavior when the network is overloaded, thereby reducing the tendency for congestion to spread from overloaded paths to the rest of the network. This modification also tends to make satellite trunking more attractive. Other potential sources for improvement include recent topology changes including the addition of transcontinental VSAT trunks between MIT-77 and SRI-51. These trunks became operational about 1 July. The routing effort is still in the experimental/tuning stage. We expect that performance on many PSN-to-PSN flows will be improved by this modification. It is also possible that some PSN-to-PSN flows will receive degraded performance. We at BBN would like to hear both the good news and the bad news. Please send your observations about recent changes in internet/Arpanet performance to Fred Serr (fserr@bbn.com), Bob Pyle (rpyle@bbn.com) and myself (scohn@bbn.com) and I suppose to the list if your observation is of general interest. Stephen Cohn Director of Network Analysis BBN Communications
hinden@CCV.BBN.COM (Robert Hinden) (07/17/87)
Vint, Doug, Its not voodoo networking, but lots of hard work. There have been several changes on the Arpanet (new line and a software patch) which make things better. These changes are very recent. A more detailed report will be seen soon. Regards, Bob
SRA@XX.LCS.MIT.EDU (Rob Austein) (07/17/87)
The third transcontinental link (MIT77 <=> SRI51) came online about two weeks ago. I noticed a serious DECREASE in performance just after the link came up, but now I too am seeing a major performance improvement (it's been a long time since I could FTP from MIT to Stanford at noon on a Friday!). My guess is that there were some initial tuning problems after the new link came up; not too surprising given that this is a satellite bounce instead of a land line. Presumably somebody at BBN can give a more detailed report. --Rob
weltyc@NIC.NYSER.NET (Christopher A. Welty) (07/19/87)
Hard work indeed. I too have noticed significant improvement in ARPAnet performance. THose resposible should be congratulated! --- Christopher Welty - Asst. Director, RPI CS Labs weltyc@cs.rpi.edu ...!seismo!rpics!weltyc