I migrated to the NEXTGEN servers because of the benefits listed everywhere. I built a server and realized that our processes (which were cpu intensive) were taking twice the time they used to on Firstgen.
The response from their side was 'You get "occasional" CPU bursts in Firstgen & that is the reason.
I decided to benchmark the cpu performance & found out that I was right about it.
I contacted the customer care again with this tests:
http://pastebin.com/k0uefXMh
Joe concluded by saying ' The test results were because of occassional cpu bursts' , even when I mentioned that burst was accounted for.
The test results: http://pastebin.com/aArbwAq2
This time I wanted conclusive results proving the customer care wrong & placed 3 firstgen servers with a cronjob running every minute to benchmark to check for cpu burst. Following were the results :
Server 1 -
Report : http://sebsauvage.net/paste/?d4c24ce3d92b4be6#pDtNzAhU8M3X/J1zLsmuYgIUHxwhYjmNuiic/rxbNng=
Only Total Time Report : http://sebsauvage.net/paste/?e902ee80be6a2c8d#GvW0Ab52F0iqNZr5dFMi7UzqX4Ux9NA54V1wGvHZnDc=
Server 2 -
Report: http://sebsauvage.net/paste/?c33c9030eb01e2af#Iid3w7rWgJ+ysDJZx2ULCzyItXBv04Q83tmGp066nTE=
Time Report : http://sebsauvage.net/paste/?a27a204039529abd#yTKoy3qZQUxbIsAOIyYWwlIgIHWnYnfbpkfDuiS/8YA=
Server 3 -
Report : http://sebsauvage.net/paste/?a32c096ad8f1968b#psOme81n2DPgYp8aK7i75v/WsETs6bmVLKaoHS0CVqY=
Only Time Report : http://sebsauvage.net/paste/?31a9d12316d9cbd2#LblvVwH+sGzQytOwU8DM9wtH5eN+87jAbUL7UnpNOH4=
All the results were consistent, proving me right. I went back to the customer care with this just to be terribly disappointed with their response. Here is the conversation : http://pastebin.com/cedtk1zL
Here is the benchmark script that ran every minute :
http://pastebin.com/UeskWw5M
Increasing the price two folds is one thing & not informing customers about it is another. The latter is the one I am more disappointed in.
vCPU allocations for next generation cloud servers were adjusted to provide more consistent performance, not to extract more profit. Rackspace is not making any more money per server than before. In fact, we recently reduced the price for 512Mb and 30Gb servers by an average of 30%. We take transparency seriously, so we published an article detailing the changes and explaining the logic behind it which you can read here: http://www.rackspace.com/knowledge_center/whitepaper/cpu-in-....
Erik Carlin, director of product management put it this way “"Our goal is to find the sweet spot that maximizes price performance while still providing a consistent experience." Customers with larger flavors are experiencing higher and more consistent performance with next generation cloud servers. Depending on the load running at a given time in neighbor instances, performance can be higher for small instances too. We are working with a third party on benchmarks which we expect to publish in the next few weeks.