Hello all This is a bit of an open question, looking to see if I can get some insight as to what people have been doing regarding Exchange 2010 DAG design and utilising either their exisiting Storage or buying new storage. I, and 2 colleagues, will be Designing and then upgrading our 50 000 user Exchange 2007 environment over the next 9 months. It is in a Cross Site failover environment using 12 Mailbox Servers in a SCC Active passive arrangement (6 Active MB servers, 6 passive, 2 nodes SCC's). We are using LeftHand Networks P4500 SAN nodes. My primary concern is storage. Is anyone out there intending to keep using LHN kit for exchnage 2010 (It is expensive for what it provides, in that with 2010 we would not intend to keep using the LHN replication technology. So we would be paying a lot of money for effectively 4TB of server storage per node with our current product. We have 20 nodes mind you). Could I ask for some examples of what people out there are looking at for their storage options with 2010? I am thinking of going with RAID'ed storaged and 2 copies of Database. This would be for a predicted storage usage of about 28TB by the time 2010 goes live. I am happy to take this Off forum if that is best... Tom

Tom, My adviswe is to work with Microsoft Infrestructure team for design part.Anil

Could I ask for some examples of what people out there are looking at for their storage options with 2010? I am thinking of going with RAID'ed storaged and 2 copies of Database. This would be for a predicted storage usage of about 28TB by the time 2010 goes live. We run a 5 node Exch2010 on Vmware with real-time replication on a Netapp storage via iScsi. We had considered fibre channel but the performance requirements just didn't call for it and so far we've been happy with both performance and recovery. However, we came to these solutions and conclusions by having our exchange techs, netapp support and vm support all site down and discuss options over a number of conference calls. With only 5k users it took some thinking. We have more TB than you're running so I think you should be looking for higher volume raids with built-in redundancy and start bringing in the big guns to help you determine some good rollout solutions.

Thanks New York Tech. Do you mind if I ask you a few more questions? Are you using the replication that comes with Exchange 2010? Or does the netapp solution offer you a replication mechanism. With LHN ISCSI I have replication built-in, but it accounts for half the cost of the node, so to turn it off and JUST use it as storage is ridiculously expensive. Though, it is EASy to use. Also, is your setup cross site? I will be data gathjering this coming 2 week period, determining my existing i/o, required i/o, user profile sizxing etc so will have a clearer pic of requirements soon enough. I am trying to determine if we are trying to do this on too tiny a budget, in that we have under 100k for new storage i.e we need to utilise our existing SAN storage with something newer and cheaper. Not going to be easy. Fibre channel would be over specc'ed for us also, I've used HP EVA's before and we just wont need it (as well as being soo pricey).

Microsoft's guidance is incredibly clear: get off the SAN, use local disks, and use the built-in replication in Exchange DAGs. If you've already invested in the SAN you might want to keep using it, but of course disks contain moving parts and they don't last forever. If you go with cheap storage, you could be looking at more than just 2 copies of the database. If you do that, you can even stop using RAID and just let databases fail over when a disk fails. http://msexchangeteam.com/archive/2010/03/29/454443.aspx It overall could be cheaper even though you'd be buying more Windows and Exchange licenses to support all those additional copies. (Microsoft just wants the money you are giving to your storage vendor, I guess!) I'm in kind of same situation with a similar sized environment (except we use 2007 CCR to replicate between sites...got off SAN-based replication a while ago.) The technical decision seems easy, it's more of a political thing. So we'll see. There's nothing stopping you from mixing your storage types, i.e. use your existing "paid for" storage for one server and have it replicate to servers that are using cheaper storage. The when it's time to renew your SAN hardware, replace it with the cheap stuff. As long as they all meet the I/O requirements (which to be honest, are pretty low in 2010) you're good.

Thanks very much for that CN9, very helpful. The place I work investied in the SAN just over 2 years ago so I still have 30 months of it left. Buy-back is an option but it truly does create soo much extra complexity in the implementation stages - always tight deadlines.... :-) If they are happy for them to slip then the problem goes away. It is indeed a political thing. Mixing up the storage was my probable choice, just need to work out the practicalities for implementation. Thanks for the input. Tom

Hi Tom, As someone who also manages P4500 SANs I agree with you that buying more Lefthand will not be the best investment. The *same* hardware (e.g. DL180/185s) can be bought cheaper without SAN/iQ and Exchange will provide the "Network RAID" via Database Availability Groups. Another option to consider (some testing may be worthwhile to validate this) is to look at using Network RAID 5 (improvements coming in SAN/IQ 9) to get more storage out of your Lefthand kit while retaining Network RAID, and/or move away from Lefthand multi-site replication to using DAGs. I have considered (and still considering for a remote site with vSphere also) Lefthand for Exchange 2010 but on it's own it doesn't make financial sense compared to DAS. Feel free to mail me off list, contact details via link below. SteveSteve Goodman Check out my Blog for more Exchange info or find me on Twitter

Thanks guys, And Steve, I may well takr you up on that offer. Will get some further work done on it and might drop you a line later. Tom

No. Microsofts guidance is not that at all. The guidance is the provision of cheap, low-cost mailboxes. "cn9" wrote in message news:8c2edb60-8b9c-4b05-8f57-563c0cc54409... Microsoft's guidance is incredibly clear: get off the SAN, use local disks, and use the built-in replication in Exchange DAGs. If you've already invested in the SAN you might want to keep using it, but of course disks contain moving parts and they don't last forever. If you go with cheap storage, you could be looking at more than just 2 copies of the database. If you do that, you can even stop using RAID and just let databases fail over when a disk fails. http://msexchangeteam.com/archive/2010/03/29/454443.aspx It overall could be cheaper even though you'd be buying more Windows and Exchange licenses to support all those additional copies. (Microsoft just wants the money you are giving to your storage vendor, I guess!) I'm in kind of same situation with a similar sized environment (except we use 2007 CCR to replicate between sites...got off SAN-based replication a while ago.) The technical decision seems easy, it's more of a political thing. So we'll see. There's nothing stopping you from mixing your storage types, i.e. use your existing "paid for" storage for one server and have it replicate to servers that are using cheaper storage. The when it's time to renew your SAN hardware, replace it with the cheap stuff. As long as they all meet the I/O requirements (which to be honest, are pretty low in 2010) you're good. Mark Arnold, Exchange MVP.

NetApp does offer a replication solution and it is called SnapMirror. In the Exchange 2007 timeframe nobody implemented SCR where they had a NetApp solution in place because of the incremental block level protection afforded by the NetApp platform. However, our advice for Exchange 2010 is generally to use the DAG where the latency parameters are appropriate. If you have servers at the remote location you would generally want to use the Exchange replication so as to reduce the number of troubleshooting points. NetApp do have customers who place their DR storage into vendors data centres but only rent the servers if a disaster happens. For this scenario they are delighted that SnapMirror exists as it reduces the TCO of the DR environment. There are a myriad of solutions out there and you choose the one most appropriate to your business requirements in terms of TCO, ROI, RPO and RTO. Its all a balancing act. "Millardus" wrote in message news:9d19ea7a-8a92-48a6-9c31-92ca12d0ec20... Thanks New York Tech. Do you mind if I ask you a few more questions? Are you using the replication that comes with Exchange 2010? Or does the netapp solution offer you a replication mechanism. With LHN ISCSI I have replication built-in, but it accounts for half the cost of the node, so to turn it off and JUST use it as storage is ridiculously expensive. Though, it is EASy to use. Also, is your setup cross site? I will be data gathjering this coming 2 week period, determining my existing i/o, required i/o, user profile sizxing etc so will have a clearer pic of requirements soon enough. I am trying to determine if we are trying to do this on too tiny a budget, in that we have under 100k for new storage i.e we need to utilise our existing SAN storage with something newer and cheaper. Not going to be easy. Fibre channel would be over specc'ed for us also, I've used HP EVA's before and we just wont need it (as well as being soo pricey). Mark Arnold, Exchange MVP.

No. Microsoft’s guidance is not that at all. The guidance is the provision of cheap, low-cost mailboxes. With the big cheap SAN? It's of course supported to use a SAN, but reading between the lines (or direct conversation) their idea of how to achieve this is clear.