Silicon Graphics, Inc. (SGI) Message Board

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 : Silicon Graphics, Inc. (SGI)

SGI 88.96

-2.3%

Dec 17 3:59 PM EST

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next

To: Charles Tutt who wrote (1402)	5/11/1997 7:07:00 AM
From: Jeff Maresh	of 14451

Charles - The application was an Oracle database comprised of about 500 GB of data contained in around 40 tables and 130+ indexes. It was used for decision support so it was pretty much a read-only. The hardware was an Ultra Enterprise 3000, 6-way, 4GB of memory and a Sun RAID box for mass storage. Original testing was done on bigger SparcCenters and the customer was lead to believe by Sun that the UE3000 would offer a certain level of performance defined by for example, comparative transaction execution times for Oracle SQL scripts that it did not deliver. The OS and production DBA folks that tuned the system are very good. My responsibilities are primarily app DBA work performance tuning Oracle. When everything was up and running, the OS folks were wondering where all the performance was. The Sun techies came in, checked out the system, and concluded that the system was indeed optimized and that was a good as it would get. Their recommendation: Buy a UE4000 and you will get the performance you want. There were several conclusions reached here. One is that at points the I/O to DASD was the bottleneck. But the biggest problem over the life of a concurrent long-running queries (ie. 4-12 hours) was how fast data moved around the buss. The idea of Oracle parallel query is to break up a complex query into a bunch of smaller queries and each processor works on a chunk of the problem. We had the processors. The remaining conclusion was that the benchmark numbers published by Sun although honest numbers, did not translate directly to application performance as well as they had in the past. A set of comparative numbers was developed for migrating Oracle SC apps to UE. Ie. figure that I/O bandwidth of X stated by Sun really translates to 0.44X for our application given a particular configuration. >"New maxed out UltraSparc II I worked on recently was a dog." >Please expound. What bothered you about it? In what context? What was the >application? What was the configuration, both hardware and software? Blanket >condemnations such as this aren't very effective -- we want enough details to make >up our own minds. Inquiring minds want to know! The general statement that I am making and again strictly opinion, is that Sun is still the Chevette of the industry. Its perceived to be a pretty good value but still Chevette performance relative to competitors. HP and DEC stuff is definitely higher performance but you pay for it. So I see a window of opportunity here for Origin at close to Sun's price to move into its market as Sun customers continue to be marginally satisfied with their products. A lot of the decision support machines do not belong to often "brand bigoted" IS groups, but rather are chosen, developed, and maintained by some other user department. The several Sun shops I deal with were really anticipating something more than what they got from the UE line. Regards Jeff

Report TOU Violation

Share This Post

Public Reply Prvt Reply Mark as Last Read File

Previous 10 Next 10 Previous Next