3PAR

HP realised EVA wasn’t cutting it for enterprise market, and hence bought 3PAR to get in there. 3PAR is quite different to EMC/HDS in the way it operates and their architecture. Below are some comparison points between them (I gathered these about a year ago and I don’t remember where I had got them from, so I am giving a blanket acknowledgement to its original author):

Main difference between EMC and 3PAR:

Data Processing: EMC is monolithic – all ios are queued, small ios wait for larger ones to finish, whereas in 3par, control information and data are queued and processed separately – control data in Control Processor and Memory whereas data in 3PAR ASIC and Memory.

Sparing: – EMC reserves spare disks in each raid group. In 3PAR, all drives are carved up in chunklets – data chunklets and spare chunklets – these are distributed across all hard drives. When a hard drive fails, its rebuild time is shorter. In 3PAR, raidlet groups are distributed across multiple shelves. Loss of a single shelf doesn’t cause data outage.

Controller failure: In traditional arrays, in case of a failure of a controller, write through mode is enabled which hits the performance because write-cache mirroring is not available any more. In 3PARs with 4 or more controllers, when one controller dies, its partner will continue write-cache mirroring with remaining controllers.

CLI execution – They each have a good suite of commands but their output are quite different. EMC symcli is more versatile and can return the output in XML format which is extremely easy to parse to prepare various reports. 3PAR report is very rigid, and plain text type. Parsing it in scripts is not straightforward and can be more error prone.

Having said all above, I am still biased towards EMC/HDS. Just a personal choice. Don’t hang me for it. ‘Nough of it, hopefully the links below might prove useful to you:

* 3PAR CLI Reference
3PAR How To Operations
3PAR Architecture in nutshell
3PAR Remote Copy Concepts