I don't mean to get overly technical, just want confirmation if this is a good idea, bad idea, or just an acceptable idea.I have setup a virtual Ex07 server, that is coexisting with a Ex03 server. Both of which are single domain, single instance servers whos public & private data stores are both on local storage. I am toying around with the idea of placing my heavy user storage (40ish mailboxes) on local storage in the virtual server, and creating another mailbox store in the FSG for my "light" users (100ish mailboxes), that would be attached via iSCSI. The reason I wanted to do this was to protect my local storage on the Hyper-V server for other VMs, yet give the best performance to my "heavy" users.The differene in HD LUN structure is: my local EX07storage is (8) 10k sas 3gb/s drives in a raid 50, and my iSCSI storage is (6)Sata 1.5gb/s drives in a raid5 with no drive contention. Whereas the EX03 server the private store is on it's own single scsi160 drive, and the public store is also on it's own single scsi160 drive.I've got the stores in place, and a couple people migrated to each one. Am I missing anything? Is this espcially bad in any scenario?Thanks!

RAID5 is a supported solution for Exchange 2007 as well you know, but I presume you are saying your iSCSI storage is deemed lower tier not because of the RAID type or that it is SATA disk based but because it is shared between other hosts? I also presume you have no available space left to put physical disk into your Hyper-V Host.Do you know your user IO profiles? You may find all 140 mailboxes (without taking into account mailboxsize and available space here - you never mentioned it) will still fit the IO profile of that RAID5 LUN on your iSCSI storage.Otherwise, even sticking with your original deduction, it appears it would work as long as you have ratified IO.Oliver

I only deemed the iSCSI lower tier because of the drive & raid speed, plus it is not a local hard disk to the VM. I am half full of drives in the Hyper-V server - and can add 8 more local SAS disks - which I have no problem doing. I was just thinking it would make sense for me to get the heavy IO users on local disks to the Exchange server, and put the less intensive users mailboxes on an iSCSI drive - that still has ample space. I also was thinking it made sense to split the groups simply because of the eseutil needing at least 110% free disk space - which is in essence wasted space that I could be using for VMs. If per say the originalprivate store of 100gbI need 110gb free + expansion room for mailboxes to grow. Essentially needing about 300gb of pre allocated local VM disk space just for the sake of having expansion & defrag room. We aren't a small shop - but that is valuable room to me that I can use for other VMs. Thats why I wanted to only keep the heavy IO users on the local disks - while putting the majority of the mailboxes on an iSCSI drive elsewhere.I do not know the IO profiles of either. Right now as it was - everythign was on a single SCSI160 drive - so I would imagine even the iSCSI drive has to be an improvment because I am adding another 5 spindles, and then even more so adding 7 spindles to the local SAS drive space.Thanks for the comments Oliver - I'm sure my plan will work, I just wanted to make sure someone didn't see a red light that I was seeing.

More resources: Storage Technology

iSCSI when configured correctly with best practice - will be just as fast as local disk, and can exceed IO as external storage has in general far greater controllers and capability than your normal on board RAID controller - plus you can spread IO over far greater disk.Use the Exchange 2007 Calculator to give you a guideline on your IO requirements, and then use your iSCSI storage or add local disk as needed.http://msexchangeteam.com/archive/2007/01/15/432207.aspxOliver

The one problem I'm having is that evidently, I am not using an enterprise class iSCSI server. It is just hung off an HP AIO unit that has 3TB of internal storage. There is something about the private store that will not automatically mount itself (or be persistent in the terms of iSCSI) I learned from an inside sales person that it is because the store does not have time to flush the buffers in the logs & also the cache of the HD before the iSCSI connection is severed upon shutdown. I have tried to manually unmount the store, stop the IS service, and then close the iSCSI connection and then rebooting the server, but no matter what I try I cannot get the mail server to mount the private store iSCSI drive. It shows up in disk management - and I have to go in and manually attach it, then scandisk it, and then I can mount the private store. The public store however does not have this problem? Should I be concerned at needing to do this process everytime the server restarts? (Granted it shouldn't be that often - but we all know we need to from time to time + accidents happen) Knock on wood. I have learned that an EMC or equallogic iSCSI array is app-smart and will flush all the logs and buffers before severing the iSCSI link, so that you can simply remount it elsewhere in the event of a disaster (or simply to take a snapshot). We don't have the budget for a $35k app-smart iSCSI device though.