 | |||
|
|
||
| |||
| |||
| |
|
Home > Data Center News > Server purchasing decisions in 2008 reveal virtualization's imprint |
| |
|
|
|
|
| Data Center News: |
|
||
|
|||
|
|||
|
|||
|
|||
|
|||
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
At the same time, the survey indicated that revolutionary change hasn't overtaken data centers: HP is still the favorite x86 server vendor, a good portion of data centers still favor rack servers over blade servers, and the vast majority use Microsoft Windows as their primary operating system. Users also said they have increased server purchases due to normal capacity increases, and their purchase decisions are based mainly on technical support and price. Data was culled from IT professionals whose data centers ranged in size from small to massive. More than half (51%) of respondents have less than 100 servers in their data centers, but 36% have between 100 and 1,000 servers, and 13% have more than 1,000 servers. Blade buysAs for the types of servers being used, about 42% of respondents said they planned on spending more money in 2008 on single- or dual-core rackmount servers than in 2007, and about 35% of respondents planned to spend more on blade servers this year. At the same time, a large number of respondents (40%) said they do not use blade servers at all. Further, 48% of users surveyed said they will deploy future virtualization implementations on blade servers -- about the same number that said they will use rack servers; a reversal from respondents' initial virtualization deployments. About 48% of respondents said they used rack servers and about 36% used blades for their first virtualization deployments. "The virtualization numbers basically say that virtualization happens on both blades and rackmounts somewhat independently of server form factor," said Illuminata Inc. analyst Gordon Haff. "That surprises me a little; virtualization tends to require richer configurations, which tend to be a better match with rackmounts. But the difference may well be in the noise." The most important factor in the choice of blades for respondents was first performance (42%), followed by price (19%), management features (17.5%), and finally power consumption (12%). Oddly enough, the same priorities don't apply to other x86 servers. Respondents said the most important factors in choosing a primary server vendor are technical support and service (35%), followed by product features and functions (22%). Only 16% base their decisions on price first. When it comes to vendors, Hewlett-Packard took the favorite spot when it comes to preference for blade servers; 43% said HP is their primary blade vendor, followed by IBM (27%), Dell (20%) and Sun Microsystems (6%). SMP and RAMAccording to the survey findings, there is an upward trend for large symmetric multiprocessing (SMP) servers, probably driven by virtualization, Haff said. Respondents were asked how their 2008 purchases of SMP systems (those with more than 16 cores) compared with purchases in 2007, and about 19% said that their spending had increased, 16% spent about the same and only about 5% decreased their spending on SMP machines. That said, 60% of respondents don't use SMP servers at all. As for memory, 48% of respondents said that the typical amount of RAM in their commodity servers this year was between 2 GB and 8 GB, while 21% provision between 8 and 16 GB. Only 16% put 16 – 32 GB of RAM in x86 servers, and the percentages dwindle to single digits for more than 32 GB. Haff said he was struck by the number of respondents who provision memory capacity in the 64-to-128 GB range (only 5%) for commodity servers and that only 1.5% said they provision servers with 2 GB of memory or less. This is surprising because virtualization requires lots of memory, and many respondents are virtualizing. Processor, OS choices With OSes, about 91% of respondents said they have installed Windows Server 2003, followed by Red Hat Linux (48%), Solaris (42%), AIX (30%), HP-UX (26%), Windows Server 2008 (23%), SUSE (21%), and z/OS or other mainframe OSes (18%). Meanwhile, the high-level reason users purchase new servers remains the same. Users who bought new hardware in 2008 were asked to name three drivers for their purchases, and more than half (57%) said it was due to normal increase in capacity. About 44% said they needed new hardware to upgrade old servers, and about 42% said they needed new servers to support new applications. For SearchDataCenter.com's entire survey report, click here.Let us know what you think about the story; email Bridget Botelho, News Writer. And check out our server blog Server Farming.
|
|
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
 |  |  | |  | |  | |  | |  | |  | |  | |  | |
|
|
About Us | Contact Us | For Advertisers | For Business Partners | Site Index | RSS |
|
|
||
|
|||||||
