Re: How much advisable is to migrate from oracle server to linux server 0-100?

From: Yong Huang <yong321_at_yahoo.com>
Date: Wed, 2 Jul 2014 10:09:56 -0700
Message-ID: <1404320996.16640.YahooMailNeo_at_web184804.mail.gq1.yahoo.com>



Suppose people's skills is not a concern. Troubleshooting a problem on Linux is generally easier than on Windows. This is for a typical, average, competent support person. Windows appears easy on the surface but tracing at the OS level needs some experience in systems programming. You don't nede that for Linux.

> At least last time someone paid me to measure OS overhead on a pure Oracle

> server the rank was UNIX > Linux >> Windows comparing throughput per CPU.

Would love to read that report, if it's published.

In 1998, there was a popular article on the fledgling Internet that compared UNIX with Windows mostly about performance. I vaguely remember the author strongly supported UNIX although he may be a little biased.

> I've never run the numbers to see what percentage of bugs are
registered

> on each platform,

One way I can think of to check this is to search

"226 - Linux"
"912 - Microsoft"
"23 - Oracle Solaris"
"59 - HP-UX"

etc. on MOS (including quotes) and count the bug notes. The platform ID's are not the same as in v$(db_)transportable_platform, but OPatch platform numbers. Note 337288.1 has some of them. But you still need the strings like the above to do a search. And each one I listed may be for one specific architecture or endianness, so you have to check more bugs to gather all strings for that OS (e.g. add Linux 32-bit and 64-bit together).

Alternatively, we can search on Google and immediately get the estimated count. Since that relies on people posting Oracle's bug notes to the Internet (which BTWis not right), the count is much lower than actual. But the order of the counts of various OS'es may still be good.

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Jul 02 2014 - 19:09:56 CEST

Original text of this message