Dell Poweredge 1655MC (blades)

From Kathmann Labs

[edit] Dell Poweredge 1655MC (blades)

The Poweredge 1655MC was Dell's entry into the blade market. It consisted of 6 blades in a 3U rack configuration. Each blade could have up to two processors, two hard drives, two NICs, a RAID controller, and I forgot how much RAM. Each chassis could contain up to 4 fan assemblies, dual (redundant) KVM and management modules, and dual network switches. Believe it or not, this 3U system was louder than the IBM Bladecenter running more servers in a 7U rack configuration.

While the servers themselves weren't that bad, that's where it ends. The Dell network switches built into the back of the chassis were miserable to configure. They tried to match the Cisco CLI as much as possible, but there were certain differences that just made no sense. I lost all respect for the 1655MCs (and Dell in general) for two reasons over this product:

There are no options to connect FC directly to the blades. For this we decided to boot from SAN using iSCSI. No problems, the setup is relatively straight forward, however the Broadcom NICs don't have enough memory available to allow for a boot from SAN over iSCSI. Call our senior technical representative with Dell, who is, and has been, the most responsive person at Dell to date only to find out that even though other vendors using the same NICs in their machines can boot to SAN over iSCSI, Dell won't release the firmware updates to allow this. No reason, just won't release them.
During the time the Blaster and Nachi worms were spreading like crazy, the sites that were not properly patched and routed through us to get to the internet would generate enough traffic to completely shut down the Dell switches in the chassis. Even though none of the servers in the 1655MC were infected, and none of the other servers in the datacenter were infected, just the traffic as it traversed the WAN links through that site was enough to shut down the Dell switches. Called Dell about the issue. They recognized the issue, and didn't have a fix (although they were "working on one now"). Two years later when we decommissioned the 1655MCs, I decided to check on that problem to which they still didn't have a fix.