On 1/13/2010 3:31 PM, Bob McConnell wrote:
> Emanuel Machado wrote:
>> Another issue to consider with SSDs is that they are based on Flash
>> technology. Each flash cell can only be written on about 10,000 to
>> 100,000 times or so (*), so if you're using extensive read/write on
>> your server you will be impacted. SSD manufacturers go around this
>> issue by giving some intelligence to the drive controllers, so that
>> they minimize the per-cell usage (which means moving things around a
>> bit internally, transparently to you), so in many cases you will not
>> see any impact. However, I would be careful on what I run on it, and
>> what services are enabled, maybe having another disk around for write
>> intensive apps.
>
> No, you can write (append) as often as you like. It is the erase cycles
> that are limited. So the chip life depends on how often those files get
> deleted.
But every time you append to a file the inode info is updated and the
free space list may need to be rewritten so it dies when erase/write
count is exceeded on the filesystem metadata. If you don't turn off
atime updates, you'll rewrite even on file reads.
--
Les Mikesell
lesmikesell@xxxxxxxxx
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
