> My advice is that in the end, hardware does not matter that much. If
> it solves a
> problem, it solves it until your data outgrows it again, and
> eventually you hit
> your budget limitations and cannot afford an upgrade. Of course, there
are
> certain common sense rules that need to be followed, eg. if you have a 20
GB
> database, you need at least a 20 GB disk, but otherwise, if your
> application is
> good, it runs well on modest hardware, and if not, doubling the datais
likely
> to kill it no matter how powerful hardware you use on it. A good
> case in point
> was an earlier post from an 8-CPU Irix user.
>
> I would suggest you focus on making the application good. It would be
wise to
> invest a portion of the hardware budget into purchasing a MySQL
> support contract
> or consulting services.
Done that - and very pleased with the result. The application is already,
for reasons given below, as well-tuned as I can make it.
> Regarding Windows 2000 - I am curious why MySQL is an option, but
> Linux is not.
> They kind of go together, almost the same as MS-SQL and Windows, or
> Oracle and
> Solaris. Is this a dedicated MySQL machine? If yes, I cannot think of one
> technical reason to run Windows on it, and I've tried hard in the
> past. If you
> were using Oracle or especially MS-SQL, it would make sense. But if
you've
> decided that MySQL is it for your database, I would really have a hard
time
> coming up with any reasonalbe justification for Windows even if Microsoft
or
> somebody else was going to pay me big money for it.
This is not a unique system - this is a large scale example of a general
system. It is the central database for a number of surrounding specialised
hardware units, all of which are controlled by Windows PCs. Normally there
are three or four such systems, and our current MySQL/Windows solution is
very good. The order I am trying to meet now has about 24 surrounding
systems instead of 4. But it is otherwise identical. We have, for example,
24/7 support staff who will have to support end user staff if anything goes
wrong. It is hard enough getting them up to speed on the Windows platform -
adding the Linux platform for them to learn would be an extra burden. And I
would find it difficult to be truly expert on several databases. And the
testing burden of the same software against several database back-ends
would be considerable.
And to the suggestion of using MSSQL: this giant project would support the
cost of MSSQL but the smaller systems wouldn't. Aside from the fact that I
have already coded some MySQL specific code (which is, of course,
reversible), smaller systems would not absorb the cost of MSSQL with
replication. (I didn't mention that we actually have a replication pair of
the machines described, and MSSQL with replication is much more expensive
than without).
We are currently retreating from an unsatisfactory Linux development with
an outside contractor. Nobody's fault, but our knowledge of Linux and his
understanding of our requirements didn't overlap enough. Our management are
understandably reluctant to introduce an OS of which we have little
knowledge when we are working hard to increase our knowledge base of our
primary platform, which is (for better or worse) Windows.
Thanks to all who have commented.
Alec
--
MySQL General Mailing List
For list archives: http://lists.mysql.com/mysql
To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
