Oracle FAQ Your Portal to the Oracle Knowledge Grid

Home -> Community -> Mailing Lists -> Oracle-L -> Re: Oracle's relationships with expert DBAs (and the rest of us mere mortals)

Re: Oracle's relationships with expert DBAs (and the rest of us mere mortals)

From: Nuno Souto <>
Date: Thu, 1 Jun 2006 14:03:30 +0800
Message-ID: <>

Quoting MVE <>:

> I still don't see any value in going Linux even for a shop on a tight budget
> --
> this is just a trend. They shoot themselves in the foot by doing so and
> will
> be paying the price in the next 4 years.

I have to disagree with this view that Linux is expensive in the long run. Mainly because I've been Linux only for the last 14 months (and mostly Linux for 3 years now) in our shop and I see no evidence whatsoever of increased running costs.

Quite the opposite, in fact.

We used to be Solaris/Sun. In common with just about every other proprietary solution, Sun kept changing their hardware/software platform every year, at a cost: to us. These were physical architecture changes, not just slight upgrades! That meant complete retest of not only Solaris-to-app but app-to-everythingelse and Solaris-to-everythingelse. Expensive!

We now use IBM or DELL blades whichever are cheapest when needed, Apple raids or whatever else is cheap and reliable, RHEL3upd5 on them - soon to be RHEL4 - and Oracle 9ir2 - soon to be (I hope...) 10gr2.

Hardware and system costs are a pittance compared to how much it'd cost us to change the whole server and disk architecture every time Sun said so. Or IBM AIX. Or M$/Intel, or any other proprietary architecture for that matter.

The hardware, if and when IBM or Dell releases a new blade model and we need one, is directly compatible to what we have and our software (all of it, not just the OS!) will run with little - if any - need for re-testing.

Maintenance once things are running is minimal if ever needed. Monitoring is the same across the board and easy as pie, using standard monitoring tools (Nagios and Sitescope as well as a few custom scripts).

The only thing that costs us considerable money is the amount of testing we have to do to make sure Oracle runs fine with Linux when we get new releases of each. But that to me is a problem of Oracle, not Linux! And it is manageable and doesn't cost the earth - although it *IS* a cost.

The other big cost we have is in setting up and installing new boxes: the number of patches and hoops we have to go through is above average compared to others. But, that is an INITIAL cost. Not a recurring maintenance cost!

And once again it's mostly because of our choice to use Oracle: no matter how much they have bleated over the last 5 years or so of how friendly they are to Linux, it's not true.

Then again: have you seen what happens when a new version of Windows comes out? It has ALWAYS broken the current version of Oracle.

So: 6 of one, half a dozen of the other.

And you folks want me to believe that the problem with Linux is the long term running costs? What I'm seeing quite frankly, is exactly the opposite!

Nuno Souto
from sunny Sydney
Received on Thu Jun 01 2006 - 01:03:30 CDT

Original text of this message