Jeremy Zawodny wrote:
>
> On Thu, Jul 19, 2001 at 04:21:05PM -0700, Mike Wexler wrote:
> >
> > We currently have a dedicate server for MySQL. The server is a dual
> > pentium III (1Ghz) with 2GB of RAM in it. It also has 2 18GB 10,000
> > RPM drives in it arranged in a Raid 1 configuration (mirror).
> > Sometime in the next 3-6 months we will be maxing out its
> > capacity. (We were maxed out a few days ago, but we added 1GB of RAM
> > and cached some query results). The system is currently running
> > RedHat Linux 6.2.
> >
> > While there are some non-optimal queries and maybe some variable tuning
> > that we can and should do, we will need to upgrade at some point and its
> > not obvious to me what the upgrade path is.
>
> Until we have more of an idea where your system is stressed, it's hard
> to say. Are the CPUs maxed? I/O channels? RAM?
At 1GB the RAM was stressed. Paged like crazy. I don't have enough
experience under load since the upgrade to tell where the current stress
point is, but it does appear to be much better balanced now.
>
> > The axes of expansion I see are:
> > 1) CPU speed (2 GHz processors?)
> > 2) # of CPUs (quad processor, 8 processors?)
> > 3) Multiple machines (replication)
> > 4) More memory (current system maxes out at 4GB)
> > 5) Different CPUs (SPARC, Alpha, IBM Power, HP-PA, Itanium)
> > 6) Faster disks (15,000 RPM)
> > 7) More disks (striping, different databases/tables on different disks,
> > MySQL striping)
> > 8) Switch some high contention tables to InnoDB, BDB or Gemini to avoid
> > lock contention
> > 9) Optimize server variables
> >
> > Which approach or combination of approaches is likely to double
> > (quadruple?) our throughput at the best price performance? I have
> > attached some info to help characterize our usage.
>
> Replication. You can do it with less expensive hardware. You'll get
> good performance and probably be able to scale farther wit it. Of
> course, you'll want to look at #9 before spending any money. And try
> to get an idea of where your contention for resources is today.
I already increased table_cache from 128 to 2048. Which helped. And last
night I increase key_buffer from 16MB to 64MB. Maybe it should be even
larger?
If I use replication, I guess I should have all the updates go to a
master server and distribute the queries to the slaves?
>
> Jeremy
> --
> Jeremy D. Zawodny, <[EMAIL PROTECTED]>
> Technical Yahoo - Yahoo Finance
> Desk: (408) 349-7878 Fax: (408) 349-5454 Cell: (408) 685-5936 <-- NEW
>
> MySQL 3.23.29: up 33 days, processed 263,569,017 queries (89/sec. avg)
---------------------------------------------------------------------
Before posting, please check:
http://www.mysql.com/manual.php (the manual)
http://lists.mysql.com/ (the list archive)
To request this thread, e-mail <[EMAIL PROTECTED]>
To unsubscribe, e-mail <[EMAIL PROTECTED]>
Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
