RE: 64 node Oracle RAC Cluster (The reality of...)

Bill Bridge is involved in writing it. Whether it will be fast is an open question, because even Bill can't change the laws of physics.

But it will be a minimal moving part. "Everything Bill Bridge starts ends in checkmate."

Associating the anticipated reliability of ASM with the word "instance" in order to confuse the reliability of ASM with the complexity of a full featured RDBMS instance is an interesting debating trick, but it is no more cogent than insisting a progressive income tax is "fair."

IF in fact ASM "instances" prove to be less reliable than OS add-ons like volume managers, then ASM will certainly fail to be adopted. The code path of ASM should in fact prove to be less complex than a full featured CFS. The functional requirements are a tiny subset of a CFS, and only precisely controlled "clients" will be able to request well defined services from ASM.

You are probably uniquely positioned to know the rich set of features and functionality that a good CFS provides.

Now as for "a given CFS" that is intentionally *not* pointed at your product. The more reliable the CFS, the better the argument for the simplicity of a single ORACLE_HOME for each vintage/variety of the code you have.

I don't understand your entire sentence about raw, nor the meaning of "old school" in that context. I'm pretty sure that if shared raw is broken, then the grid is going to fall apart. If you're talking about performance, that is an entirely different conversation.

I hope I've answered your question "can you tell me how having a separate instance specifically for ASM in addition to your production instances is considered less moving parts?"

To recap, an ASM instance should be quite a simple small thing. For imagery, an ASM instance is maybe a hummingbird egg and an RDBMS instance is an Ostrich egg.

Regards,

mwf

-----Original Message-----
From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org]On Behalf Of Kevin Closson Sent: Wednesday, June 22, 2005 4:34 PM
To: oracle-l_at_freelists.org
Subject: RE: 64 node Oracle RAC Cluster (The reality of...)

 >
>1) Just because a CFS is supported doesn't mean it is the most
>reliable service of an OS. If a given vintage of ASM or
>straight shared raw has fewer "moving parts" (shall we say
>less code path?) than a given CFS,

 can you tell me how having a separate instance specifically for ASM in addition to your production instances is considered less moving parts? As far as code path, raw versus direct IO CFS comparisons are old school. ASM has, um, quite a bit of overhead and comms when manipulating files (not to be confused with manipulating the contents of files).

--
http://www.freelists.org/webpage/oracle-l


--
http://www.freelists.org/webpage/oracle-l