Re: Using Orcale Parallel Server to archieve nearly zero down time

On Mon, 18 Oct 1999 00:57:02 +0800, "Jason Wong" <jaswong_at_mail.com> wrote:

>I am currently looking for a high availability database solution for one of
>system. In my mind, I am planning to implement SUN Enterprise 4000 as the
>Servers, Oracle8i Parallel Server as the database engine, Veritas Cluster
>Server as the cluster manager and Clariion 5700 dual process RAID as the
>data storage.
>
>I want to have two instances of Oracle8i, that share the same database,
>running in 2 separate servers actively. So, when one of them shut down, the
>other can take up the task immediately. I expect the down time to be less
>than 1 minute. However, some vendor told me that even with Oracle Parallel
>Server, two instances of Oracle8i that share the same database cannot be
>running actively at the same time. This is true because at any one time the
>Cluster Server will allow only one server to mount on the disk volume. Hence
>only one Oracle8i instance can access the database. When the primary
>Oracle8i is down, the secondary has to take 15 to 20 minute to initiate the
>database.
>
>Is this true? Has anybody tried a similar setup? Please give me some advise
>and tips.
>
>Thanks in advance!
>
>JasonWong
>
>
>

This largely depends on the capabilities of the underlying cluster hard- and software. In the old days (5 years ago) I was administering a database on a VAX/VMS LAVC cluster (means no shared hardware, only DECNET). Disks where 'shared' and visibly by both systems, and Oracle Parallel Server running on both systems on ONE database. Synchronization between the instances is done by Oracle. You do need to setup a redo log thread on BOTH systems. They will both process their own redo log, so you will need both redo logs to recover. I am not familiar with this particular SUN hardware, but no doubt it would be possible to accomplish a similar setup.

Hth,

Sybrand Bakker, Oracle DBA Received on Sun Oct 17 1999 - 15:45:28 CDT