The tests compared Sun UltraSPARC III hardware running Solaris 8 Zones vs. various Amazon AWS EC2 instances by performing a sort function using Java/Hadoop on a 10GB random text file. The results further show the ever falling price/performance ratio of Public Cloud vs. In-house Virtualization. You can download the complete copy of the results here
- Posted by: Kathy Bauman
- Posted on: June 04 2009 01:28 EDT
- Re: Freedom OSS Performance Lab : Sun Solaris vs. Amazon AWS Cloud by A D on June 04 2009 11:19 EDT
- Re: Freedom OSS Performance Lab : Sun Solaris vs. Amazon AWS Cloud by Tracy Nelson on June 04 2009 15:28 EDT
- Nice FUD piece! by Warren Strange on June 04 2009 13:51 EDT
- Sun Hardware Used in the Test [Correction] by Kathy Bauman on June 05 2009 12:05 EDT
the report is actually quiet useless. It is a single page XL sheet with no information related to price, duration of test and the actual setup and management overhead. Is there any other information related to the whole testing process.
Is there any other information related to the whole testing process.Doubtful. If there were, people would be pointing out that you can run Solaris 10 on a quad-core Opteron with 8G of memory and a 250G disk for under $500 (I know I am). Then they'd post some results showing numbers in the single-digit range, which would be more indicative of the kind of performance you get from contemporary hardware.
[Disclaimer: I work for Sun] All this shows is Moore's law at work. An UltraSPARC IIIi is a very old machine. Unfortunately this kind of FUD does manage to fool a significant number of people who don't bother to look deeper. BTW - Solaris 8 never supported zones. You can however run a Solaris 8 branded zone under Solaris 10.
Here's the actual hardware that was used, the spreadsheet has been updated accordingly: SPARC Enterprise T5220