Fri Jun 1 13:13:27 PDT 2001 — Sonic.net has always worked hard to make all of our services as fast and reliable as possible. Over the past year we have made some substantial upgrades to our network to provide truly world class services to our customers. The most noticeable upgrade so far has been the switch over from single servers to load balanced server pools for our critical services like web and email. We’ve also been very busy behind the scenes doing less visible upgrades too, such as clustering our Network Appliance NFS file servers for ultra high availability. Most recently we migrated away from a single border router configuration handling both of our T3’s we moved to a dual router setup to give us truly redundant upstream connectivity. The Cisco routers are sized such that either one is capable of handling both circuits in the event that one of the routers fail. The final outcome of the re-architecture of the network is that any single failure of a server or device on our network will not result in a loss of service. Where this is not possible or feasible, we’ve purchased cold spares so in the event of a failure, we will be able to restore services a quickly as possible.
We are currently focused on eliminating the last major point of failure in our network: our core switching. Currently all of our traffic must pass through our single Extreme Networks Black Diamond switch. Over the past few months this switch has failed a number of times. We have been working very closely with E.N. to resolve the issues that we’ve had with the switch but, due to the fact that neither E.N. nor or own team can account for the continued failures of the switch, E.N. has agreed upon our request to completely replace the $120+k switch. We are planning on replacing the switch Sun Jun 3 at 1:00AM. Due to the sheer size of the switch and the number of connections, it may take as long as 30 minutes before every connection is re-established. We will be keeping a close eye on the new switch; if it exhibits similar behavior to the existing switch we will replace it with a different switch, most likely from a different vendor.
Once we are satisfied that the new switch is stable we will integrate our second Extreme Networks switch into our core to provide us with a redundant core network. After this has been completed we will be able to wrap up the migration to an ‘active-active’ redundant configuration of our two Alteon AD3 Layer 4 switches and will move our NetApps to redundant Giga-Bit Ethernet. For those of you that are interested in more details we’ve posted a diagram of the proposed layout at www.sonic.net/network and we’ll be happy to answer any questions in news:sonic.net
-Kelsey, Dane, Scott, Eli, Nathan, Russ and Steve.