Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Advanced Micro Devices - Moderated (AMD) -- Ignore unavailable to you. Want to Upgrade?

To: tecate78732 who wrote (257565)	12/26/2008 9:57:19 AM
From: pgerassi	Read Replies (2) \| Respond to of 275872

Intelbuyer: Double EE's do not work in IT. They are in the R&D or production divisions. IT is typically CS or MBA. Besides they are not referred to as "swap outs", but "upgrades". You don't replace a mission critical server to add memory, disks or more I/O channels. You may add a new one once the original is maxed out, but they are not usually maxed out at initial purchase. These are servers that cost $500K and up. Besides what do you do when your blade server with 3 or 4 blades are installed needs more processing power? They don't swap out the blade chassis with the 3 or 4 blades in it and replace it with a blade chassis with 5 blades in it. They just add a blade as needed until the chassis is filled. That is what mainframes are, a chassis with one or more cabinets. YOu have a main CPU chassis, but you can extend that chassis with a extender CPU chassis. You start out with 4 or so processor cards in the man chassis and add more cards as your needs grow. When the main chassis is filled, you add the extender CPU chassis and keep going. Only when you have filled both chassis, do you get a new mainframe. But usually by then the OEM (IBM typically) usually has come out with a more powerful processor card. So say you replace the Power5 cards with Power6 cards and get another boost. That is usually much cheaper than buying a brand new mainframe. The difference between the mainframe and a blade server is that its a single image across all cards in the mainframe while its separate images in each blade in the chassis. Of course you could work for GE, GM or Exxon with the PC tech support division or one of the engineering divisions and never run into a blade server or mainframe. You likely wouldn't even be allowed in the room with the mission critical mainframes and support equipment. When the servers go for $1 million and up, buying new is an expensive proposition. When they are $10K or so, that isn't a big deal. Its just like cars. You can work for the big three and not see many engine swaps. But if you worked in their racing division, engine swaps are very common. You may not do engine swaps on a $20K car, but you do on a >$500K racer. Pete

To: tecate78732 who wrote (257565)	12/27/2008 2:33:19 AM
From: rudedog	Read Replies (2) \| Respond to of 275872

CPU upgrades are not all that uncommon in blade installations - but they are not done by replacing the CPUs on-site. Instead, blades with new processors are swapped into a rack, and the blades removed have their firmware and CPUs upgraded offsite - the process continues until all the blades are upgraded. This can be done without taking the main systems down if the system has sufficient redundancy. This was also common in older systems (pre-blade), but without the ease of upgrade. ProLiant (both pre and post HP) had the ability to shut down and 'hot swap' a processor (and it's related power module if power characteristics changed) without taking the server down - this was true for higher-end 2, 4 and 8 way boxes. I was involved in planning of several upgrades of installations with more than 1000 servers, which are not small in any sense. Swapping of a processor could be done in under 10 minutes, start to finish, by an experienced technician. Arrangements to maintain refreshed hardware can be a part of an enterprise support agreement. OEM vendors work out the cost of upgrades over the life of the contract pretty carefully, often making that compliance a part of design acceptance. Obviously a silicon vendor who maintains socket compatibility presents a better value both to the OEM and to the customer. This is very common in GSA contracts, Telco companies, and other segments where long term business planning is tightly integrated with technology roadmaps.