10-25-2010 02:21 PM - last edited on 05-27-2013 12:21 AM by maikoro
For a new database server I'm considering using 4x 60GB SSD drives in raid 5. Normally I never use raid five but seeing that controllers have improve much over the last couple of years I'm tempted to use it.
The second reason I'm considering raid five is the fact that this server will see mostly read action. At least 75% read but probably even more. That's a lot of read but at what point does raid five make more sense?
And then there is the financial motive. SSD drives are still rather expensive. 4x60 raid 5 equals 6x60 in raid 10.
Some other questions that keep me occupied: Will the extra parity data writing have a big impact on the life span of the drives.
Do I need to over provisioning my storage to reduce wear? I expect this server will have a production run of 5 years. Initial DB size will be around 80 GB but this could grow at least 20 GB per year. All tough at some point some data will probably be archived.
Transaction logs, OS and backups will be stored on traditional disks. Hardware will be a Proliant DL380 with 16 disks and two P410 controllers.
P.S. this thread has been moved from Servers > Planning to Storage > planning - Hp Forum Moderator