Skip to main content
Start a Conversation

Unsolved

This post is more than 5 years old

dg0ldsmith

15 Posts

91916

May 15th, 2007 01:00

MD1000 Redundant EMM ?

I'm trying to figure out what the benefit is of having a redundant EMM in an MD1000 storage array. The documentation says the redundant EMM will maintain enclosure functions ... but if the primary EMM fails, and that is what connects the MD1000 to the PERC5 card in a server, how is it still connected? Can the primary EMM "fail" and still provide electrical connectivity to the RAID controller? The MD3000 is probably the product I want to consider for redundancy and clustering options but I'm just trying to make sense of the MD1000 "redundancy". Thanks, Dave

tommo666

1.2K Posts

May 15th, 2007 15:00

From my manuals:

If an EMM fails then the surviving EMM then assumes enclosure management, which includes monitoring and control of the audible alarm, enclosure LEDs, power supply/cooling fan modules, and fans.

Failover does not include providing connectivity to the drives controlled by the failed EMM

dg0ldsmith

15 Posts

May 15th, 2007 15:00


@tommo666 wrote:
From my manuals:

If an EMM fails then the surviving EMM then assumes enclosure management, which includes monitoring and control of the audible alarm, enclosure LEDs, power supply/cooling fan modules, and fans.

Failover does not include providing connectivity to the drives controlled by the failed EMM





Right, so what good is it to have something in the storage array maintaining management and monitoring functions of the storage array if my server can't access the drives?

tommo666

1.2K Posts

May 15th, 2007 18:00

I can only think that because you can split the backplane and make 2 arrays. They decided that if it had a 2nd EMM in unified mode it could take over some fuctions on failure. If you lost all sensors then alarms would sound and multiple failures reported and the box could power down. I suppose this way at least the box will stay up even if you can't access it.
 
The MD3000 can provide high availability redundancy. It's basically an MD1000 chassis but with raid contollers instead of EMM's.

Dell-Jeff G

156 Posts

May 16th, 2007 12:00

The MD1000 is also used as an expansion disk enclousre and in this case, the EMMs are fully redundant, even for data paths.  When the MD1000 (in unified mode) is connected to a PERC 5/E, you can only connect 1 EMM.  As tommo666 statedm if you split the backplane, then both EMMs are used, but not for redundant data paths.
 

SSG_Bob

17 Posts

May 30th, 2007 21:00

The Enclosure Management function also controls power shutdown in the event of excessive thermal rise beyond critical limits as part of the safety features of the unit.   Without failover to the other EMM the whole unit would be powered down on EMM failure - resulting in the situation where there is no status to indicate what failed.
 
Loss of the functionality of the enclosure processor itself does not imply loss of your data path through the failed EMM module.  The data paths are lost during the part replacement action where by necessity one must disconnect the cabling and unplug the module.   Otherwise the redundant EMM will keep your system powered up and the SAS data paths remain operational.   That would seem to be a benefit.

SSG_Bob

17 Posts

May 30th, 2007 21:00

Dave,
 
The MD3000 is the solution that fully supports clustering and redundancy.   Dell has recently enabled expansion using MD1000 enclosures (with 2 EMMs) for up to 45 SAS drives with failover capability.
 
Bob

View More

View All

No Events found!

Top