So I started a new job and and part of my job is to analyze business cases for IT projects. My job is of an operational and financial capacity and I'm not too familiar with scientific research infrastructure (I'm more knowledgeable in ERP systems). So if a business case proposes to replace SGI Origin SMP Compute Servers (never even heard of them) with commodity based systems (Dell, HP, IBM, etc servers) and do not specify the benefits in quantifiable form (i.e. net increase in computing throughput) should I challenge the business case or do I just stfu cause the scientists know what they're doing.Oh, the current cluster is used to crunch mad numbers basically. [Edited on December 17, 2007 at 10:29 AM. Reason : .]
12/17/2007 10:25:48 AM
12/17/2007 3:16:46 PM
yeah i'd do the research before committing to the commodity hardware approach. everybody wants to be like google cus they are sexy, but it's not right for most i'd bet. it might be right if you have hardware budget constraints, but you're probably going to have increased maintenance costs
12/17/2007 3:25:29 PM
12/17/2007 4:45:19 PM
The current SGI is used for computational biology and chemistry. They want to replace the SGI cluster with commodity based system and retire the SGI to testing for a few years until it fully depreciates.
12/17/2007 5:48:43 PM
how old is the SGI cluster?
12/17/2007 5:51:13 PM
^ You know, it doesn't say how old in the business case. But one of the big business justification is that the commodity based system will be much cheaper maintenance.
12/18/2007 10:33:11 AM