Discussion in 'OT Technology' started by piratepenguin, May 29, 2008.
- Describe your RAID.
- Describe what it means.
srsly, raid = data reliability and increased i/o performance.
i don't really need it though. jbod is fine for my needs.
Kills bugs dead!
To me, its safety of all the crap i've accumulated over the years.
Already had an ex "accidentaly" delete a hard drive and got most of it recovered... it was scary
It means, "FUCK. I should be running Solaris with ZFS on this box. RAID on small servers is so lame."
data protection without loss or downtime
Downtime is impossible to prevent.
Redundant Array of Inexpensive Disks
Some of it is.
raid is just one piece in a highly available solution. in looking for a solution to maximize uptime, you want to remove any single points of failure.
- redundant disks through RAID
- redundant controllers to disk
- redundant power supplies connected to separate PDUs which are connected to separate UPSs, either of which across the board can handle a failure (this is one reason why the Sun 880/890 sucked, it had 3 power supplies, so you couldn't connect it to 2 power sources without a special PDU in between)
- redundant servers in a cluster or an equal or better fault tolerant solution like load balancing across multiple hosts
my preference is a Sun T-series machine with 2 Emulex fiber cards connected to redundant Brocade fiber switches (that are configured as 2 isolated fabrics) connected to Hitachi USP storage and running Veritas Storage Foundation HA (VxVM, VxFS and VCS). i guess DMP for redundant paths. i can go either way for root disk mirroring, Disk Suite/Sun Volume Manager or VxVM encapsulated root (it isn't so bad if you do it right and properly create all of the slices).
redundancy, if one drive fails, i won't lose all my pirated softwares and movies