Reader Feedback on Building a Legacy
July 11, 2011 Timothy Prickett Morgan
|
Timothy: Great article!!! In your statement about “Windows server variants” being almost 20 years old, does that include Microsoft SQL Server? Just wondering if that is considered “legacy” nowadays. But, since you define one of the characteristics of legacy as being “stable,” maybe SQL Server is really not legacy. Thank you again for a great article. –Doug Er, I suppose SQL Server is legacy, especially when you consider it is actually derived from Sybase‘s Sybase System and later SQL Server, which date from the late 1980s and early 1990s, respectively. TPM, it is always a privilege to read your work. As a technology investor, I try to follow the major trends. One question that I had when Intel presented its RISC migration slide was an “apples and oranges” issue. When originally presented by Kirk Skaugen, I thought he said that the $15 billion RISC/mainframe dollars were total system–not just the brain chip stuff–and that on an adjusted basis the dollars were about $1.5 billion? With this consideration, Advanced Micro Devices is an obvious target. In another area, the recently completed International Super Computing 2011 meetings in Hamburg, Germany were most interesting. Intel’s “commitment” using MIC chips seemed unusually firm? Also, Cray was notably missing from Intel’s MIC Partner list? –Ese97 Hey, Ese97: What Kirk Skaugen said was that if Intel could convert the RISC/mainframe systems revenue stream to X64 servers, it would generate $1.5 billion in additional revenues for Intel each year. The MIC co-processor, which I covered in detail here in my other job over at The Register, is a very tricky problem for Intel. The co-processor will have at least 50 X64 cores on a single device, with cache memory and GDDR5 graphics memory to support application code dispatched from X64 processors, and will have a considerable amount of oomph for certain kinds of work. Price it like a lot of server chips in a small box, then you have people buying it instead of lots of Xeons. Keep workloads restricted from it, then no one but supercomputer customers uses it. I think Cray and Intel were in-cahoots on a MIC co-processor project with DARPA for the “Cascades” range of supers due next year, but that contract was rejiggered a while back and I think the MIC co-processor might have been what fell off the table. But no one ever confirmed my suspicion. –TPM Your online magazine. I like it better than all of the other printed IT magazines. Also, it is more convenient. Bless you. –Herman Well, thank you very much. We work very hard to be the best. –TPM RELATED STORY
|
