Scott Lowe had an interesting post on Friday about creating metaLUNs that has attracted some interesting comments, which is where the best parts are.
Given today's computing power, there is no need for stupid software contraptions like metaLUNs, they are a side effect of a dull architecture that puts the onus on the administrator to figure out what to do. It's easy - just figure it out!
3PAR InServ arrays automatically wide stripe data over all the drives in the system, without headaches or "weird science". It's very simple for administrators and the performance is excellent - without requiring an army of professional services people, the way things tend to work with arrays from the Ever Mounting Costs company.
And now for something completly different: Human Tetris!
Kudos to the Clariion for supporting 'wide stripes'; however, the level of complexity required leaves me asking, 'isn't there a better, simpler way?'
Posted by: Vaughn Stewart | October 26, 2009 at 03:38 PM
Yes, but some people like a good challenge with negative consequences.
Posted by: marc farley | October 26, 2009 at 04:20 PM
EMC complexity reminds me of Cisco complexity, I guess it's fitting that they are partners. Kind of strange to have VMware in there though they like to take the complexity out of (most) things it seems.
Posted by: nate | October 27, 2009 at 03:51 PM
Now if there were just a big (probably yellow) "Rebalance array" button on the 3PAR. I love the wide striping, but am not such a big fan of manually rebalancing when adding additional storage.
Posted by: Jason Jensen | October 28, 2009 at 01:00 PM
Oh, and watching that human tetris video was probably the best 6 minutes of my day. Thank you Japan and thank you Marc. :)
Posted by: Jason Jensen | October 28, 2009 at 01:14 PM
Hey thanks Jason. How are you at funky handstands?
As for the magic "Rebalance array" button - that would certainly be the ultimate, but for now DO (Dynamic Optimization) is the tool we have. It's a decent tool, but only redistributes the data from one volume at a time. Question: how much data would you want to rebalance all at once?
Posted by: marc farley | October 28, 2009 at 05:00 PM
Well, we added a new cage to our T800 that had 32TB of SATA disk in it and wanted to rebalance our existing 80TB out to include the new disks. DO worked, but it was definitely not automated and not the speediest process ever.
Posted by: Jason Jensen | November 04, 2009 at 02:03 PM