We are in the middle of planning a migration from Exchange 2003 to Exchange 2010. We currently run Exchange 2003 clusters on NetApp storage and use NetApp SnapMirror technology to replicate Exchange data to DR. For folks already on Exchange 2010 in production, have you left Exchange running on your SAN or have you migrated to local storage? The only reservation we have with migrating to local storage is having to give up the ease of performing point in time restores from NetApp snapshots (full calendar restores from a certain point in time, restoring databases infected by a virus). Yes Exchange 2010 offers lagged copies, but from what I hear, restoring data from a lagged copy is pretty painful. Though restores do not happen regularly, we work in a time sensitive environment and when called upon need to recover data quickly. Thanks for your insight.

There is no simple answer to your question. If you're using NetApp for your storage and plan to continue to do so, then you should contact NetApp for guidance on configuring it. They have specific requirements, especially if you're going to use Snap Manager for Exchange to back up. There are cases to be made for local storage or SAN, especially the way NetApp does it. One key input to the formula is how many database copies you plan to retain. Another is whether you want to back up at all. Microsoft Online, for example, has multiple DAG copies and doesn't run backups, saving a substantial amount of money. In my opinion, you don't want to be expecting to depend on point-in-time backups for Exchange. For Exchange 2010, you should be considering multiple database copies. Ed Crowley MVP "There are seldom good technological solutions to behavioral problems."

Today NetApp has millions of seats on Exchange 2010. I can think of over a half-million in E2010 right off the top of my head in my local area, but then again, its my job. Anyway, sales over. Just about everyone who deployed Exchange 200x onto a FAS has remained with NetApp for 2010. One company I can think of and that youve heard of went to DAS because their MS consultants told them to. After six months they came back to NetApp because couldnt get the IOPS and found that they had latency problems to DR. When you deploy Exchange 2010 youre going to do it on the new DS4243 shelves with 2TB disks which gets you a lot of capacity and all the IOPS you require. Youre going to use deduplication within the arrays to get back all that space which was lost due to SIS going away. Youre going to use Thin Provisioning to get back all that LUN space that NTFS requires you to maintain. Youre going to reutilize those expensive FC/SAS disks for SQL or something else. Youre going to use the FlashCache module to beef up your capabilities. What you are, most likely, going to do is to move away from SnapMirror in favour of using the DAG. Only where your WAN has excessive latency will you want to continue with SnapMirror and what are you going to do with DAS in that case? Youre SOL there without intelligent array replication. One point to take away is that using a NetApp, or any other, array is part of an enterprise strategy. Far too many Exchange = DAS fanbois leap on the SAN is expensive bandwagon without understanding two things: 1) organizations need simple, scalable and reliable storage across the board, not Exchange, and 2) its pretty easy to prove that the Opex of DAS far exceeds the Opex of SAN, so long as you look at the entire picture. And if youve got a consultant just talking about your Exchange, youve got the wrong consultant. Regarding lags, Microsoft apparently doesnt use lagged copies any more. That should tell you something as to the general value of that particular bit of functionality. Microsoft will give you case studies that say how you can get Large, Low Cost Mailboxes with direct storage but the storage vendors can do the same with their storage, and more so. Download TR3842 and TR3867 to show the benefits from a technical perspective. I could go on for hours. "TGIF2011" wrote in message news:c45d1999-05a0-49fa-b936-79bca59c1652... We are in the middle of planning a migration from Exchange 2003 to Exchange 2010. We currently run Exchange 2003 clusters on NetApp storage and use NetApp SnapMirror technology to replicate Exchange data to DR. For folks already on Exchange 2010 in production, have you left Exchange running on your SAN or have you migrated to local storage? The only reservation we have with migrating to local storage is having to give up the ease of performing point in time restores from NetApp snapshots (full calendar restores from a certain point in time, restoring databases infected by a virus). Yes Exchange 2010 offers lagged copies, but from what I hear, restoring data from a lagged copy is pretty painful. Though restores do not happen regularly, we work in a time sensitive environment and when called upon need to recover data quickly. Thanks for your insight. Mark Arnold, Exchange MVP.

Thank you both for your responses. Ed, right now we take backups of Exchange every 4 hours and replicate to DR, so over a 24 hour period we keep 6 copies. We retain 5 days worth of copies, so there are 30 copies at a time. We use NetApp snapshots, which consume less space by only snapshotting the deltas. Since we are limited to a single copy of a mailbox database per Exchange server, preserving this setup require a lot of servers and Windows and Exchange licenses. We plan to have multiple copies of the Exchange databases, 2 in the production site for site resiliency, 1 in DR, but we'd also like to maintain point-in-time backups, for example in the case there was a virus outbreak and we needed to restore prior to the time of infection or if someone required a calendar restore and needed their calendar in a specific state. These are just a few examples. Mark, after speaking with my colleagues we are considering doing a hybrid solution. Since we are under the 250ms latency requirement, instead of SnapMirror we plan to use DAG replication. 3 member DAG 2 DAG members in the production site, connected to local storage. 1 DAG member in DR, connected to NetApp storage. For snapshots, we will utlilize SnapManager for Exchange on the DR DAG member. Will this solution work?

Sure. Thats what most people tend to do. SME6 (which you will need along with SDW 6.3) has a new feature called Copy Mode. You do your normal backups on the DR** copy which truncates the logs. You also do copy mode backups on one (or two) copies in the primary centre but you only keep one or two hanging around (so thats just a 10% or so additional storage). You mitigate that additional storage by using dedupe. That will prevent you needing to do a full reseed if you activate the DR copy. NetApp has a paper for that. Ask your local CSE or partner for the long-winded answer. Also, if youre going to continue using your FC/SAS disks you can get huge space savings by putting both copies of the local copy of the database into the same Flexvol and deduplicating it. Thus your two, 1TB databases will consume a single terabyte on disk. There will be inflation and deflation over a 24 hour (if you dedupe once a day) period but your local NetApp person will help you understand that. http://markarnold.blogspot.com/search/label/De-duplication will help a bit, or completely confuse you. As Ed says, Microsoft do their solution with four copies and no backups or lags. With NetApp you can do that if you want, but with half the physical storage and you get some emergency protection in snapshots if you need/want. Ive seen more than a few customers who, for lots of very convoluted reasons, want to maintain backups for a while for reasons as varied as my boss says so to compliance. I dont want to open the debate on whether its a legitimate reason or not, just that the customer is always right (except when he isnt) and well help him achieve the business goals he communicates to us. For that reason SnapVault is an epic solution and, with Exchange 2010, has some fantastic deduplication and availability benefits. **Remember that with DAG its not a DR copy, its a second/third copy of the storage in a different site. If you cant run Exchange in that remote data centre just like it was in your primary centre. Some people fall down on that minor point. "TGIF2011" wrote in message news:62845173-120a-4bf1-bd72-ac28a9c08bf4... Thank you both for your responses. Ed, right now we take backups of Exchange every 4 hours and replicate to DR, so over a 24 hour period we keep 6 copies. We retain 5 days worth of copies, so there are 30 copies at a time. We use NetApp snapshots, which consume less space by only snapshotting the deltas. Since we are limited to a single copy of a mailbox database per Exchange server, preserving this setup require a lot of servers and Windows and Exchange licenses. We plan to have multiple copies of the Exchange databases, 2 in the production site for site resiliency, 1 in DR, but we'd also like to maintain point-in-time backups, for example in the case there was a virus outbreak and we needed to restore prior to the time of infection or if someone required a calendar restore and needed their calendar in a specific state. These are just a few examples. Mark, after speaking with my colleagues we are considering doing a hybrid solution. Since we are under the 250ms latency requirement, instead of SnapMirror we plan to use DAG replication. 3 member DAG 2 DAG members in the production site, connected to local storage. 1 DAG member in DR, connected to NetApp storage. For snapshots, we will utlilize SnapManager for Exchange on the DR DAG member. Will this solution work? Mark Arnold, Exchange MVP.

Thanks Mark. For the two DAG members in production we wanted to use local storage. For the third DAG member in DR, we wanted to use NetApp storage with SnapManager for Exchange. In DR we would have the supporting infrastructure to run Exchange. 1) Since the active DAG member will be continously replicating to the DAG member in DR, is there any benefit to doing copy mode backups in the primary center? Can't we just get away with backing up the passive copy in DR? 2) With DAG in place in DR, why would a full reseed be required to activate the DR copy? Is a full reseed relevant for SnapMirror replication only?

Oh. OK, thats fine. 1) if you have local storage you can simply use normal backups with SME in DR and youre good to go. If you were to use NTAP on the primary site youd get all those benefits and be able to do the copy mode thing that (2) will continue with. One thing you have to think about is why, if Microsoft were happy with one copy of storage with 2003, they suddenly decided that you needed two copies in 2007. And after switching to 2010 you now needed four copies. Of a progressively more stable and reliable database. Why? Because MS couldnt rely on external influences not to mess things up. Basically Exchange couldnt rely on Windows not sucking so Exchange has 4 copies in 2 sites so that Exchange is available. MS dont have financial constraints in the same way you do and dont have the same level of Exchange availability that your customers demand. Trust me on this. 2) oh boy, youve been taken in. The DAG is, OMG, no panacea. Its not the 2nd coming. It doesnt have 44 written on its back. Its less flawed in SP1+ but nooo, its not perfect. Expect to re-seed and plan for it accordingly. If you want a customer/internal conversation on this then please feel free to engage me @ marnold at netapp dot com. Try do it through your account team though, but its not a show-stopper. "TGIF2011" wrote in message news:655ef27f-6483-4471-bfa3-4c50e7a1f11c... Thanks Mark. For the two DAG members in production we wanted to use local storage. For the third DAG member in DR, we wanted to use NetApp storage with SnapManager for Exchange. In DR we would have the supporting infrastructure to run Exchange. 1) Since the active DAG member will be continously replicating to the DAG member in DR, is there any benefit to doing copy mode backups in the primary center? Can't we just get away with backing up the passive copy in DR? 2) With DAG in place in DR, why would a full reseed be required to activate the DR copy? Is a full reseed relevant for SnapMirror replication only? Mark Arnold, Exchange MVP.

Hi TGIF2011, In my opinion, DAG as a HA, it is good for the online failover, could keep the email system online for users as possible as it could. If we want to backup the database and for the big disaster or other purpose, your choice seems great. Per my known, we use the backup for the passive node database, and could reduce the seed time. Some information for you: http://technet.microsoft.com/en-us/library/dd335158.aspx Regards! Gavin Please remember to click “Mark as Answer” on the post that helps you, and to click “Unmark as Answer” if a marked post does not actually answer your question. This can be beneficial to other community members reading the thread.