Re: Help! Oracle server clustering

Pete Sharman wrote:
>
> In article <afd7ib$65t$04$2_at_news.t-online.com>, "shrinad_at_ora-india.com" says...
> >
> >forget all the cluster shit !!!
>
> Hmm, is that a technical term? ;)
>
> >
> >go for standby database
> >
> >easy and really unbreakable
>
> Yes, but it is used for something completely different from "the cluster shit".
> From the tone of the OP, I think this is something they don't have a great
> understanding of yet, so it's important not to confuse them into thinking they
> are used for equivalent issues.
>
> >
> >put the server in different places, propagate the logfiles over network
> >
> >this is the best solution, use the standby server for testing, reporting,
> >make a second standby database with delayed log apply, copy a test database
> >from the standby databases, make your backups on the standby
> >
> >no serious shop should ever be without standby
> >
> >
> >"Thao" <thuthaot_at_hotmail.com> schrieb im Newsbeitrag
> >news:1b9e9988.0206260809.4efe5df4_at_posting.google.com...
> >> I am looking to have my 8i oracle server run on NT to two compaq Win2K
> >> servers that will back-up for each other. I dont' know waht method I
> >> use use and what are the advantages/disadvantages?
> >>
> >> 1. Clustering server or server-mirroring?
> >> 2. what tools do I need?
> >> 3. Does Oracle 8i support clustering?
> >> 4. Any suggestions on using Oracle on Win2K clustering on Fault
> >> Tolerance Features?
> >>
> >> Thanks for any of your feedbacks
> >> thao
> >
> >
>
> HTH. Additions and corrections welcome.
>
> Pete
>
> SELECT standard_disclaimer, witty_remark FROM company_requirements;

The biggest problem is that people don't know what a real cluster is.

NT is not a cluster. It it a fail-over only solution (sys-a dies sys-b wakes up, tries to recover and then continues)

A real cluster will allow 100% availability of the application, database, files on all node concurrently.

For the real definition read up on OpenVMS clusters -- this is where it all started.

Try running 9iRAC on Tru64 TruCluster or OpenVMS Cluster. A common file system requiring only one install of the software, updates, patches, one set of files to backup, and if configured correctly could lose any piece of hardware and the application/database would still be 100% available.

You can rip a disk, network cable, powercord from either system and the cluster continues to function. (Generally configured in an N+1 configuration - say you need 3 nodes to handle all of the processing, you would get 4 so that if any one node crashed, you still have enough capacity to run your shop.-- try that with NT) and if configured correctly the nodes and mirrored disk would be in different buildings: Bankers Trust (or whoever they are today) had nodes in NYC and nodes in NJ).

And yes, I am still an OpenVMS biggot.

-- 
Regards,

Michael Austin            OpenVMS User since June 1984
First DBA Source, Inc.    Registered Linux User #261163
Sr. Consultant            http://www.firstdbasource.com
                          http://www.firstdbasource.com/donation.html
704-947-1089 (Office)     704-236-4377 (Mobile)