Re: A beginners guide to implementing ASM

From: Mladen Gogala <gogala.mladen_at_gmail.com>
Date: Wed, 3 Mar 2021 21:10:52 -0500
Message-ID: <d3d66dbf-d732-f82b-0434-75bb84eed1ff_at_gmail.com>


That may depend on your database load. I tried it with an OLTP database. I am not sure about the secret sauce in the Oracle's filter driver. Generic description of the Linux filter drivers can be found here:

https://www.msystechnologies.com/blog/understanding-device-mapper-and-filter-driver/

My assumption is that FD uses OS caching to speed things up. Since it's supposed to use the same driver as the udev rules, that's the only way it can be significantly faster, which was, again, not what I observed. If it uses Linux vfs cache, it will probably be significantly faster for some read intensive loads. Filter driver is an interface to the real disk driver, just like the SCSI driver  generated by the udev rules. There is an additional software interrupt in both cases. However, you gave a good idea. I will create an ACFS file system using both FD and udev rules  and will run bonnie++ on top of that. That will be sort of objective measure. Sort of, but not not quite. Many things depend on the type of the database load.

On 3/3/21 8:41 PM, Jared Still wrote:
> We will have to agree to disagree on that point.  I have measured it,
> and FD is faster than udev rules or ASMLib
> Jared

-- 
Mladen Gogala
Database Consultant
Tel: (347) 321-1217
https://dbwhisperer.wordpress.com

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Mar 04 2021 - 03:10:52 CET

Original text of this message