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 : All About Sun Microsystems -- Ignore unavailable to you. Want to Upgrade?

To: cheryl williamson who wrote (27725)	2/14/2000 3:30:00 PM
From: Steve Lee	Read Replies (1) \| Respond to of 64865

Re: No, I've never seen an application crash Solaris Have you ever seen Solaris? <G> If the customer is sensible enough to not put crap on their NT servers, then the customer doesn't have to suffer. It's a question of choice. Those customers that don't want to suffer don't have to - but those who do like a bit of suffering have the chance.

To: cheryl williamson who wrote (27725)	2/14/2000 3:46:00 PM
From: rudedog	Read Replies (2) \| Respond to of 64865

Cherryl - re: If M$FT can't certify & test 3rd party drivers ahead of time, it's their problem. Exactly right, which was the whole reason for the driver signing and DLL isolation implemented with Windows 2000 - which by the way had a number of the OEMs and IHVs up in arms, since they have an additional test burden, and then have to submit to a multi-week certification cycle with MSFT before they get a signed driver. I was not just referring to my personal experience - a technology group I work with monitored more than 10,000 instrumented NT servers in various places in wall street and the insurance business over an 18 month period and logged every event which caused the system to go down, either directly or because the OS needed to be cycled. As you point out, NT does go down sometimes... and with a sample that size, there were lots of events to look at. The work is not public so I don't want to go too deeply into the numbers, but that sample produced just over 140,000 down events - or about 1 every six weeks, per server - with an average time out of service of 23 minutes... generating availability of better than 99.5%. Of those 140,000 events, there were only about 250 where it was "possible" that an application had caused the failure - in every other case, the problem was traced to driver or DLL problems, or to incorrect handling of queues or failure to release resources or I/O. Before I paint too rosy of a picture, the biggest offender among software vendors was microsoft itself, with about 2/3 of the failures... but still, that work does not portray an OS which is especially vulnerable to being trashed by an application, since only about .2% of the failures could be assigned to that root cause. I also have some Solaris numbers for similar work done by the same group. The sample was smaller, only about 1200 machines, but the overall performance was not statistically different and was actually slightly lower in terms of availability.