Re: SSD Storage
From: Niall Litchfield <niall.litchfield_at_gmail.com>
Date: Wed, 17 Oct 2012 13:27:04 +0100
Message-ID: <CABe10sYQ66CDUnPUiehYUnm00c4gvu+-z3C-FYwENw26VPH4bQ_at_mail.gmail.com>
You'll hate this..
It depends.
Date: Wed, 17 Oct 2012 13:27:04 +0100
Message-ID: <CABe10sYQ66CDUnPUiehYUnm00c4gvu+-z3C-FYwENw26VPH4bQ_at_mail.gmail.com>
You'll hate this..
It depends.
OK so the smart answer to my smart alec answer is "On what does it depend?" I think there are several concerns.
- What are the current bottlenecks in the system? If they aren't database disk I/O speeding up database disk I/O is unlikely to help.
- What's the real-life life expectancy of the kit your vendor is talking about - IIRC for example the lifetime of exadata flash parts is 3 years, for a write acceleration technology that probably doesn't matter, for data storage it does.
- How does your vendor deal with the fall off in write performance over time, i.e how do they guarantee the same performance in a year's time when you've loaded up the flash with data? EG http://storagemojo.com/2012/06/07/the-ssd-write-cliff-in-real-life/
- Are they selling flash or SSD, there's a difference
- If you currently use storage level snapshots/clones etc are those features there?
Personally I was convinced enough about flash to apply for a job with Violin (they didn't want me :) ) but that doesn't mean that flash is a panacea, or in all cases has enterprise levels of reliability as yet - it will shortly and many suppliers already are up to the mark.
Others on the list are better qualified than me to comment.
On Wed, Oct 17, 2012 at 12:32 PM, Howard Latham <howard.latham_at_gmail.com>
wrote:
> I've got a vendor going on about SSD Storage for database - Are these the
> way to go as far as performance goes?
> --
> Howard A. Latham
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>
-- Niall Litchfield Oracle DBA http://www.orawin.info -- http://www.freelists.org/webpage/oracle-lReceived on Wed Oct 17 2012 - 14:27:04 CEST