Wed Nov 24 18:23:35 PST 1999 — I will be upgrading memory in thunder.sonic.net tonight to half a gigabyte. This will further improve performance for many multihomed websites. We are working now on deployment of a number of improvements for multihomed hosted sites including a really nice new stats package and a large new Alteon AceDirector 3 layer four switch. The L4 switch allows us to balance load between multiple web servers, which increases reliability and also gives us the ability to down servers for upgrades and maintenance transparently.

Multihomed customers already have enhanced quotas – an extra 50 megs of storage for each site, and an extra 200 megs per day of outbound served data transfers. We’ll continue to work to add value to this great service offering. -Dane, Kelsey and Scott

Wed Nov 24 18:19:17 PST 1999 — We are again seeing latency problems on our primary T3 circuit, and we’re working with PacBell and UUNet to isolate the trouble. PacBell will be testing the circuit tonight at 2am to see if they can find any problems, so we expect slower than usual connectivity as we run on backup links at that time. We’ll keep you posted as we continue to work on this. -Dane

Tue Nov 23 08:31:04 PST 1999 — At 3am last night, PacBell replaced the interface card which serves our UUNet T3 connection. This appears to have eliminated the latency troubles we’ve been seeing in the past few weeks. I’m currently seeing 17ms average latency over this circuit. We were previously seeing performance as bad as 120-160ms average. -Dane

Tue Nov 23 08:28:00 PST 1999 — Note that when you connect to the new news server the first time, you will probably be offered a few thousand new groups. Using most shell based newsreaders, you’ll be asked if you want to join each one, with a yY/nN option. Answering ‘N’ will skip all of the rest of them, answering ‘Y’ will join all of them. Lowercase n/y will allow you to work through each of them, but this can get really tiresome as there are so many. Most users will want to just hit ‘N’.

The new server has some article numbering issues as a result of the transition, as the numbers are close, but not exactly what was on the old server. Because of this, some groups may have different numbers of articles than the group listing indicates. For more discussions regarding the new server, see ‘news:sonic.net’. -Dane

Mon Nov 22 17:24:25 PST 1999 — We are swinging the CNAME for news.sonic.net from ultra.sonic.net over to our new server, typhoon.sonic.net. This huge new Usenet server blows the doors off old ultra with half a gig of ram and over 110 gigs of disk space. Typhoon is currently being fed two full T1s (3.0Mbps) worth of news 24×7 via satellite downlink on a new 1.2 meter dish on the roof of our data-center here in Santa Rosa. In addition, it has land-line based backup feeds and additional feeds to fill in any gaps that the high speed satellite feed might miss.

Over the next week, we will be adding additional feeds, both inbound and outbound, but right now, this system is running faster, larger and better than ultra, so we’ve decided to make the switch. If you notice any odd news performance, please post to the group news:sonic.net and let us know. Those of you who are using supernews.sonic.net for binaries will probably want to switch back to news.sonic.net for even better performance. -Dane, Scott and Eli

Fri Nov 19 16:24:08 PST 1999 — Sonic.net will be closed on Thursday the 25th of November for the Thanksgiving holiday. Technical support will be available 8am to 9pm on Friday, and 9am to 9pm during the weekend, but our sales and accounting offices will be closed Thursday through Sunday. If you have any pressing issues on Thursday, please email support@sonic.net and management will be looking for urgent tickets. Have a great holiday weekend! -Dane

Fri Nov 12 18:20:24 PST 1999 — System wide maintenance is scheduled for tonigh at 12:00am We will be working on thunder.sonic.net and bolt.sonic.net, both systems will be down for aprox. 15 to 20 minutes. we are also upgrading our MUX that serves T1s to our San Rafael, Sebastopol and Napa POPS this may cause connectivity issues that will last from 10 to 15 minutes. -Steve and Evan

Thu Nov 11 17:38:52 PST 1999 — An increase in the buffers has resolved the packet loss and latency troubles. Downtime was under 15 seconds, and this resulted in a very brief slowdown of Internet traffic during this time. We’re now seeing screaming fast performance, and we’re working to tune down the amount of traffic that’s still transitioning the T1 circuits. Customers should see an overall improvement in network responsiveness now versus the last few weeks. Thanks for your patience as we’ve applied this upgrade. -Dane, Eli, Scott and Co.

Thu Nov 11 16:17:42 PST 1999 — We’re still struggling with troubles on the primary T3 circuit here, and the next step in the troubleshooting is to change up a buffer count in the circuit provisioning setup. This will entail about three minutes of downtime for this circuit, and during this time, our two backup T1 circuits will be running very full, and probably quite slow. The alternative is to do this late at night, but the performance on the T3 is so poor right now that we’d like to make every effort to get this resolve ASAP. -Dane

Wed Nov 10 20:33:57 PST 1999 — The packet loss issue has been cleared, we determined that our SMDS/DSU needed additional configuration, and we’ve had some issues communicating with the device to configure it. We need support from the hardware Vendor to resolve these issues. In the short term, we’ve backed our access-class down to it’s previous level, and will turn it back up tomorrow morning. -Eli