The real advantage of the md3000i is that it allows vmware clients who have deployed standalone/esx standard deployments to move up to esx enterprise rather easily. The disks in the md3000i are identical to those in the PE2900's/1950's and MD1000's, so converting from standalone to enterprise is just a matter of having good ethernet connectivity, buying the md3000i chassis (only $8K after tax/shipping), and swapping drives and/or connecting existing md1000 units to it. Performance wise, the md3000i's are likely to be OK and probably will be sufficient for small enterprise deployments, but longer term and when site-to-site failover becomes a critical requirement, you'll need to replace it with a netapp or equallogic. We just bought one md3000i and will be testing it over the next few weeks.

We looked at lefthand too, and are actually quite hopeful about their longterm prospects especially with 10gb/s adapters coming out... As far as performance of the md3000i, the disks array itself wont be a problem..I think what it comes down to is a) you are limited to only 4 gige interfaces which should be ok for now (some of the equallogic and netapps at the low end are limited to 2-3 active interfaces) and b) dell is extremely unlikely to provide any upgrades to processing power/controller network interfaces ever...they'll just roll out a new md4000i model or some such so the performance you get from each md3000i on day 1 is all you'll ever see. As far as replication goes, you can only hope that c) manual management of luns across several md3000i's isnt that much of a problem (becoming less of one definitly with the new storage vmotion capabilities in vmware 3.5) and d) lefthand or someone recognizes the need for a device to do iSCSI network raid w/ generic storage on the backend so that you dont have to buy lefthand VSA's or arrays, just a single box to do RAID and replication across several md3000i's and have one of these boxes at each of several datacenters. It's always possible that a few years down the line that vmware will do this for you somehow too. However, if not and if you just end up putting the array to lower and lower end uses, it isnt that much of a loss given the cost.

With EqualLogic, cache, network bandwidth, and CPU are added with each shelf of disks because of their architecture. Each array comes complete with controllers and the systems are virtualized as one pool of storage with all the contollers working together to distribute the workload. So, perfomance shouldn't be an issue - you can choose whatever RAID setting you feel is appropriate. I have several units in various RAID policies and the system automatically load balances my mailstores and logs to the appropriate array(s).

OK guys, we finally started testing our own md3000i yesterday.

good news: performance is very very reasonable, with just a single iscsi nic and a single interface on the md3000i - we easily achieved 80 MB/s writes and 105 MB/s reads with minimal cpu.

bad bad news: from what I can tell, until Dell formally announces vmware support and probably updates the firmware -- the current firmware may not be able to support DR/HA properly. When we create a vmfs on one node, all works fine on that node...but if you try to access the vmfs from another node, the md3000i blocks. I spoke briefly to a Dell tech who said he'd look into it, but he commented that the current cluster support is primarily windows which does active/failover on a per-lun basis rather than per vmdk which vmware does? I'm not knowledgeable about windows clustering, but we just havent been able to get the clustering to work properly here yet.

No problem. My EqualLogic replaced a CX500 and that is a better apples-to-apples comparison, not the md3000i. It seems this product is designed for test/dev or really small IT shops without the need for extra features.

I am curious to see if this is a "homegrown" Dell product or if its a re-branded LSI - may be interesting but not relevant.

Good news. Finally managed to figure out how to get vmware clustering to work on the md3000i with vmware esx 3.0.1 (w/ latest patches).

1) set LVM.disableSnapshotLun to 0 on each member of cluster (config -> advanced settings)

2) I tested with qlogic 4052C cards firmware 2.0.0.45, we downgraded from 3.0.x

3) You must find out which controller the md3000i has designated as primary for each vmfs volume.

Also manually input each of the ESX HBA ID's into a md3000i host group that owns the virtual disks and set host types to linux.

4) Point ESX Host 1 HBA 1 to md3000i controller1 nic 1's ip address for Dynamic Lun Scanning

5) Point ESX Host 1 HBA 2 to md3000i controller2 nic 1's ip address for Dynamic Lun Scanning

6) Point ESX Host 2 HBA 1 to md3000i controller1 nic 2's ip address for Dynamic Lun Scanning

7) Point ESX Host 2 HBA 2 to md3000i controller2 nic 2's ip address for Dynamic LUN Scanning

8) Initiate Storage Adapter Rescan on ESX Host 1

9) Your luns should show up, for each lun that has the second md3000i controller as its primary - deactivate the primary HBA path temporarily. You do not need to do this for luns which have the primary md3000i controller as primary.

10) Set all discovered luns to "last used path" rather than fixed

11) Go to storage screen and create your vmfs filesystems

12) now you can reactivate the primary hba path's you disabled, they should have switched to standby

13) For the rest of your esx hosts, you need scan the luns and do the same deactivate/activate primary hba's steps before any luns served from 2nd md3000i controller will show up.

Once all the luns are up on all hosts, you are set. Also, at this point, primitive loadbalancing to all 4 of the md3000i interfaces and both controllers should be good.