Re: switch from file system to raw device

From: Hans Forbrich <forbrich_at_yahoo.net>
Date: Thu, 16 Sep 2004 14:28:06 GMT
Message-ID: <Wph2d.43585$XP3.11245@edtnps84>

Igor wrote:

>> That has changed a lot with RMAN and certainly seems not true in your
>> case.

> We thought about EBU in the first place, later RMAN but considered it to
> be an additional complexity and of course, who will backup RMAN's
> database? Another RMAN? We (me and the dba) in fact have seen RMAN in
> class course but never used or tested it intensively at home. We are very
> comfortable with the bash shell script backup/recovery "software". In fact
> we test recovery procedure once in a while with dba and operator.

If you want to use external RMAN repositories, then Yes.

Do you have 2 Oracle databases around? Then you can have 2 RMAN repositories backing each other up at no additional cost other than setting up an extra schema and possibly an extra tablespace. (Of course, a whole raw device for this tablespace is somewhat silly.

The story gets even better in 9i ...

>> Another part of the effort is the monitoring and sizing of the
>> tablespaces. With raw, such conveniences as autoextend are challenged.

> This is talking about Oracle 8 or further. Oracle 8 came into this company
> later, we started off with 7.3.4 where, as far as I recall, no autoextend
> capability is offered. But at last, even autoextend has to be monitored
> and sized, right? One day even autoextend will come to the end of the
> disk, won't it?
>

I keep forgetting in this thread that your organization is deliberately staying at an older version.

If you happen to progess to newer versions, which is likely if you want to have SAP support, you will note that a lot of effort has been put forth in Oracle to eliminate administration or at least 'up front admin skill'.

For example, with autoextend, you simply guess at a 'reasonable' number. Accuracy at setup time is no longer that critical as it can be resolved when needed. And that generally means lower initial cost as it's no longer necessary to size for 18 or 24 months - transferring cost from startup capex to maintenance capex pr even opex (capital expenditure/operational expendture) which tend to be hidden from senior management.

>> In a world of ever-increasing point-and-click administration, a la
>> Windows, reading about your experience is refreshing.  And a reminder to
>> me that not everyone wants the easy way out.

> As I said, I don't find it more complicated dealing raw devices as I am
> doing nothing else for several years now. But you're right, I am near at
> frustration having clicked a button, and things work, but have not really
> understood what happened. In a situation of malfunction many of those
> "point-and-click admins" seem not to be able to turn things around and fix
> problems (I know someone like this). };-)

I agree. My hat's off to your ability.

But yout skill set, and that of a number of contributers in this forum, is becoming more rare. We are, of necessity, being forced into consulting roles to troubleshoot behind the point-and-clickers - it seems with the dramatic reduction in computing-related prices, management no longer can afford to have advanced specialists on staff.

>> And then there is the question of getting enough raw devices when the
>> entire system can be handled by one 146GB SCSI drive.  And the 400GB
>> SATA drive is causing some serious rethinking ... who would have thought
>> that 3 disk drives for under USD$1500 (eg. www.dirtcheapdrives.com)
>> could provide over a TB of storage.

> This is, when your company support this kind of choices.

If your organization is concerned about cost, value, profit and loss .... eventually management will ask how to reduce costs. Compare the price of 1 TB using IBM FAStT SAN to the SATA price above.

>
> But, returning to the technical issue, I am doing some tests today, to get
> things going. I will sure post my experience, just give me some time ...

Looking forward to the information. Received on Thu Sep 16 2004 - 09:28:06 CDT