RE: The Cube
"....Each blade chassis design is out of date and needs replacement by the time you can buy it...." We still have some old HP p-class blades and some IBM HS20s in use, both over four years old and not retired yet. We will probably replace them with c-class at some point, but for now they do the job just fine.
"...they are all completely proprietary to ensure high margin...." Really? I thought it was so they could perfrom better in a more compact package with better power savings compared to rack servers. At least, that's what we see. If that saves us more money in the long run then I'm not to bothered if the vendor makes a smidge more margin.
"...thus appalling value for money...." Our saving point used to be around five blades in a chassis, at which the total cost (purchase, support, operational) became less than five racked servers. Currently, the saving point is at around four blades - if a "green-field" project requires more than four dual-socket servers then it's cheaper for us to go blades. Maybe you need to have a chat with your reseller about how much he's skinning you for!
"...and none of them play nicely within your existing data centre anyway...." What, they don't share they're sweets with the other servers? They interconnect via exactly the same cable types as racked servers, they use the same protocols, and they're easier to manage. In the case of HP, IBM, FSC and Dell, they use expanded sets of the same management tools that racked servers use, something Egenera definately can't claim, as PAN is for their frames only.
"....You may as well buy Verari or Egenera, at least these guys have done a little more than taking the case off 10 of their 1U servers...." I haven't any experience of the Verari kit so I'll take your word for it, but for the Egenera kit they certainly did more - they had to make a specially cut CD for each rack to boot off (not customer cutable), PAN couldn't manage more than one rack at a time, and they had storage bottlenecks through the virtualised fibre channel blocks because they spec'd two 2Gb FC cables for the whole rack. Egenera, though fascinating technology, just didn't meet our requirements, and that was before we started to look at branch offices where we needed just a small chassis to replace usually less than six racked servers. Given also that IBM and HP (and even Sun) could also put UNIX blades into the same chassis as x64 blades, allowing us to consolidate old UNIX apps onto new kit with much lower operational costs, and you may get an inkling as to why I smile at your post.
/still looking for pointing-and-laughing icon!
Biting the hand that feeds IT
