Tech ONTAP Articles

Virtualize Your Microsoft Applications

Tech_OnTap
12,032 Views

As you move toward the goal of 100% virtualization in your data center, careful attention to the virtualization of business critical Microsoft® applications—including Microsoft Exchange, Microsoft SQL Server®, and Microsoft SharePoint® Server—becomes essential.

To get from where you are today to an environment that delivers all the benefits of virtualization, including efficiency, improved availability, and decreased cost, you have to focus on virtualization of all layers of your infrastructure, including virtualization software, servers, networks, and storage.

Key elements of the joint Netapp, VMware, and Cisco solution.

Figure 1) Key elements of the joint NetApp, VMware, and Cisco solution.

That’s why NetApp joined forces with Cisco and VMware to create  a complete solution for virtualizing Microsoft applications. This architecture  combines the benefits of VMware® vSphere 4 virtual infrastructure, Cisco Nexus unified  fabric, and NetApp® unified storage hardware and software.

This flexible architecture allows you to virtualize a mixed  workload Microsoft application environment to deliver the full benefits of  server, network, and storage virtualization. We’ve tested the performance of  Microsoft applications on this solution to make sure there are no bottlenecks  and that all performance metrics are well within Microsoft’s published  parameters.

This article briefly describes the reasons for virtualizing  Microsoft applications and highlights the most important architecture and  deployment considerations to help you get started. For full details on the  joint solution, you can refer to the NetApp technical report “NetApp  Solutions Guide: Microsoft Exchange Server, SQL Server, and SharePoint Server  Mixed Workload on VMware vSphere 4, NetApp Unified Storage (FC,...

Why Virtualize Microsoft Applications?

The reasons for virtualizing Microsoft applications with this  solution are in large part the same reasons for virtualizing any application:

  • Reduced  costs. Upgrading to newer Microsoft server applications without  virtualization can require even more server hardware to support an application  that has already become excessively costly to run. VMware virtualization unlocks  the full power of your existing hardware by running multiple workloads on each  system. Increased utilization means less hardware and lower overall capital and  management costs.
  • Advanced  storage capabilities. You can deploy Microsoft Exchange, SQL Server, and SharePoint  on NetApp storage across any storage protocol, including FC, iSCSI, or NFS.  NetApp FAS and V-Series storage arrays have been fully tested and certified for  use in FC and IP-based VMware environments. By using NetApp storage efficiency  and intelligent caching capabilities, you save significantly on storage costs.  NetApp virtualizes storage by pooling available IOPS and capacity for on-demand  use by multiple applications similar to how VMware virtualizes server resources.
  • High  availability. VMware can provide high availability (HA) for Microsoft server  applications without the need for clustering at the virtual machine (VM) level.  VMs are no longer tied to the underlying server hardware and can be moved  across servers at any time with VMware VMotion®. VMware HA provides server  hardware fault tolerance for every VM and offers greater levels of availability  over solutions designed to protect just the server. NetApp active-active  storage configurations provide similar capabilities at the storage level.
  • Advanced  backup/recovery and DR. Backup and recovery for this solution are built using  integrated VMware, Microsoft, and NetApp technologies for advanced,  application-aware data protection. Deduplication-aware remote replication for  disaster recovery with NetApp SnapMirror® provides end-to-end data protection,  and the addition of VMware Site Recovery Manager can automate the entire recovery  process.
  • Enhanced  mobility. You have the option of easily and nondisruptively relocating  the virtual machines and/or storage used by your Microsoft applications for load  balancing, upgrades, and maintenance or to meet other organizational goals.

Despite these obvious benefits, there are two persistent  concerns about virtualizing critical Microsoft applications, but these concerns  have been addressed:

  • Performance. With the  release of VMware vSphere 4.0, VMware has increased performance to the point  where it is suitable for any mission-critical business application, as  demonstrated in a recent VMware white paper that describes Exchange performance  using various storage protocols. We made performance  validation a key part of the development of this solution to address this  lingering concern.
  • Support. There  are still widespread concerns about support for virtualized Microsoft apps. The  good news is that there are multiple ways to get necessary support. Microsoft  fully supports virtualization through its Server Virtualization Validation Program (SVVP).  You also qualify for direct support of virtualized Microsoft applications if  you contract for the Microsoft Services Premier Support Program. Support might also  be available through your server OEM vendor, VMware Global Support Services  (GSS), and Technical Support Alliance Network (TSANet).

Key Design Considerations

One of our key goals when architecting the joint solution was to  provide clear design guidelines and at the same time provide enough flexibility  so that you can create a solution that is tailored to meet the requirements of  your environment This section is structured around some of the key questions  you will want to ask yourself as you move your Microsoft applications to a  fully virtualized environment.

What storage protocol should I choose? One of the great things about  this solution—like all solutions that include NetApp storage—is that you have  the flexibility to choose whatever storage protocol makes sense for your  environment. We provide architecture guidelines for all protocols: FC, iSCSI, and NFS. A joint NetApp and VMware performance study demonstrates that all protocols perform within 10% of one another, so there is  no reason based on performance to choose one protocol over another.

If you already have Fibre Channel (FC) infrastructure, you can  continue to use it. If not, NFS and/or iSCSI can easily meet your storage  needs. I advise you to look at each protocol in terms of cost to you (capital  and operational), manageability, scalability, and flexibility and choose the  one that fits your needs best. (A few more specific guidelines are forthcoming  in the section on storage layout.)

What  NetApp software will I need? We strongly recommend the use  of a core set of four NetApp products:

  • Rapid  Cloning Utility (RCU). This free vCenter plug-in provides rapid, space-efficient  provisioning of virtual servers and desktops leveraging NetApp FlexClone®, data  store deduplication management, data store provisioning, resize, and destroy  operations.
  • Virtual  Storage Console (VSC). This free vCenter plug-in lets you manage and monitor NetApp  specific storage-side attributes that pertain to VMware directly from within  vCenter.
  • SnapManager® for Virtual Infrastructure. SMVI is an integrated data protection  solution. It provides backup and recovery for virtual machines and replication  for DR. It uses the NetApp Snapshot™ capability with the option to invoke VMware snapshots for VM backups.
  • SANScreen®  VMInsight. This vCenter plug-in provides monitoring and extensive  reporting on the virtual to physical storage mapping (vmdk, data stores, LUN,  storage fabric) to help in environment management and troubleshooting.

  You can learn much more about the first three software tools shown  above in a recent Tech OnTap article devoted to the subject.

In addition, you will want to install NetApp SnapDrive® and the  application-specific SnapManager product inside guest VMs that hosts an Exchange  Mailbox server, SQL Server, or SharePoint database and index server to provide application-consistent backup and granular restores of databases, logs, and so  on. (Backup and DR are covered in more detail later.)

What storage  layout should I use for different data components? The  storage layout you choose will depend in part on the storage protocol you have selected. Rather than trying to cover all possible storage layout and protocol options  here, I’ll simply focus on one of the most flexible IP-based storage layout  options. If you are deploying from scratch or your infrastructure will support this approach, the layout shown in Figure 2, combining NFS and iSCSI, is the  one I would suggest. For FC or iSCSI layouts, refer to TR-3785.  (The  approach in all cases—and the logic behind them—is similar in most respects.)

Storage layout using NFS data stores and iSCSI LUNs.

Figure 2) Storage layout using NFS data stores and iSCSI LUNs.

Here are the guidelines at a high level:

  • Guest file system alignment is very important for optimal  performance. NetApp TR-3747  provides best practices around file system alignment in virtualized environments.
  • Create the VMs on an NFS data store using NetApp RCU.
  • Host the virtual machine (VM) vswap and temp/page file on a  separate NFS data store on a different volume on the NetApp storage system. (Separation  of transient data allows faster completion of NetApp Snapshot copies and  achieves higher storage efficiency.)
  • Locate your application data (databases, logs, and so on) on iSCSI  raw device mapping (RDM) LUNs, directly created and connected inside the guest  VM using NetApp SnapDrive software (version 6.2 or higher must be installed  on guest OS).
  • Install application-specific SnapManager software inside the  guest VM for consistent backup and granular restore.

This approach is recommended over guest-connected LUNs using the Microsoft iSCSI software initiator because if you want to implement VMware  vCenter Site Recovery Manager for disaster recovery—now or at some point in the future—the failover/failback process is much simpler with application data on  iSCSI RDMs, and you’ll get better support from VMware. You also should put all data stores and RDM LUNs on the same storage system if you are going to use VMware  vCenter Site Recovery Manager.

To leverage the benefits associated with SnapDrive and (either iSCSI RDMs as recommended above or guest-connected RDMs using the iSCSI S/W  initiator), if you want to use application-specific SnapManager tools for backup of your Exchange, SQL Server, and/or SharePoint data, you must use RDMs (either FC RDMs, iSCSI RDMs as recommended above, or guest-connected LUNs using  the Microsoft iSCSI S/W initiator).

If, for some reason, you must configure your environment using VMFS or NFS data stores for application data, your best backup option is SMVI.  SMVI is capable of producing consistent backups for all three applications but with  some limitations. Currently, because of limitations in the VMware VSS Requestor  (VMware uses copy enumeration for shadow copy), SMVI cannot provide automatic transaction log truncation or backup verification. Both have to be done  manually. Also, the VMware VSS Requestor does not currently support application  consistency for VMs running Windows® Server 2008. Therefore this solution is  limited to scenarios where granular transaction-level restore is not required (for  example, point-in-time restore for SQL Servers), manual backup verification can  be performed after the backups, and alternate methods of transaction log  truncation are possible, for example, with SQL Server databases in simple  recovery model (SQL Server provides an automated method for log truncation).

How do I  perform application-consistent backup and recovery? The best  way to achieve application-consistent backups for Microsoft applications is to  install SnapDrive and the appropriate SnapManager product (SnapManager for  Microsoft Exchange, SnapManager for Microsoft SQL Server, SnapManager for  Microsoft SharePoint Server) inside the guest OS for each VM as needed. These  tools deliver the specific capabilities to provide application-consistent  backups, automated backup verification, and granular restores. For example, SnapManager for Exchange provides single mailbox recovery capabilities. You can  learn more about these SnapManager tools in a previous Tech OnTap article.

What’s the best way to implement DR? NetApp SMVI and application-specific SnapManager products can provide  replication and disaster recovery for VMs and hosted Microsoft apps. Fully  automated disaster recovery can be achieved using VMware vCenter Site Recovery  Manager in conjunction with these products. This solution provides complete failover workflow automation for complex environments as described in the Tech OnTap article Using VMware Site Recovery Manager to Simplify DR.

Combining NetApp SnapManager, SnapMirror, and VMware Site Recovery Manager creates a complete data protection solution for backup/recovery and disaster recovery.

Figure 3) Combining NetApp SnapManager, SnapMirror, and VMware Site Recovery Manager creates a complete data protection solution for backup/recovery and disaster recovery.

How do I implement multipathing? If you want your environment to be robust, you must implement multipathing. For an FC-based architecture, I would  recommend the Asymmetric Logical Unit Access (ALUA) protocol and round robin (RR) path selection policy. ALUA allows for the autonegotiation of paths  between SCSI target devices and target ports, enabling dynamic reconfiguration.  ALUA is enabled by default on ESX hosts. On NetApp storage arrays, ALUA should  be enabled on the initiator groups, resulting in a more dynamic, or plug-and-play-like,  SAN architecture. The RR path selection policy (PSP) provides path redundancy  and bandwidth aggregation. Note that there is no need for device-specific  module (DSM) inside the guest VM.

For iSCSI, vSphere introduced support for multiple TCP sessions  at the ESX host level for multipathing. You can have two vmkernel ports and use  round robin PSP to achieve plug-and-play multipathing. It provides multiple  active paths, and no DSM is required inside the guest VM. Also, both the traditional  and multiswitch trunking network designs can be used, as described in TR-3749.

For NFS, multipathing can be achieved for both traditional and  cross-stack switches. For details, see NetApp TR 3749.

When using Cisco Nexus 10 Gigabit Ethernet (10GbE), only two  10GbE ports are required on the ESX host. Cisco virtual port channeling (vPC) provides  redundancy, fault tolerance, and security.

Using Cisco Nexus vPC to connect ESX hosts and NetApp storage.

Figure 4) Using Cisco Nexus vPC to connect ESX hosts and NetApp storage.

Are there benefits to using deduplication and thin provisioning? One of  the benefits of this configuration is that no matter which protocol you choose,  you can take advantage of NetApp storage efficiency capabilities (FlexClone, deduplication,  and thin provisioning) to significantly reduce the amount of storage space you need.

Typical virtual environments have many copies of the same OS and  application binaries in different VMs, consuming large amounts of space on  expensive shared storage. By using NetApp storage efficiency capabilities, you  can achieve more than 50% storage savings on primary storage. Figure 5 illustrates  the 92% space savings we achieved while validating the joint solution.

Space savings due to combining NetApp storage efficiency techniques.

Figure 5) Space savings due to combining NetApp storage efficiency techniques.

How do I size my environment? Sizing your environment includes sizing both VMware data stores (containing  the guest OS, application binaries, VM page file, and vswap file) and LUNs  hosting application databases and logs. NetApp has developed sizing tools to  properly size your environment. Your NetApp systems engineer or reseller can  help you size your environment based on information gathered from your site:

  • Number of application servers to be virtualized
  • Number and type of Microsoft apps
  • Capacity requirements for different data components, including  expected growth rate
  • Performance requirements, including read/write and  random/sequential ratio
  • For SQL Server databases, number and type of databases (OLTP,  DSS, mixed)
  • For Exchange server, number and size of mailboxes, user profiles
  • For SharePoint server, number of users, space required per user,  user concurrency percentage
  • Backup/restore/DR requirements

How do I validate the performance of my virtualized Microsoft application environment? You can  use the same set of performance validation tools available from Microsoft and  third-party vendors that are used in physical environments. These tools can  help you determine if performance is within Microsoft guidelines. To test this  joint solution, we used the Microsoft Exchange Load Generation Tool, Microsoft  SQLIOSim utility, and AvePoint Sharepoint Test Environment Creator and Usage Simulator to validate performance. Several load tests were conducted for these  applications, all running at the same time. Performance validation methods and success criteria for each application are described in TR-3785.  Our tests validated that:

  • There are no CPU or memory bottlenecks within VMs or on ESX  hosts
  • No I/O, CPU, or disk bottlenecks exist on storage
  • All read and write latencies were well within published Microsoft  guidelines
  • No network bottlenecks occurred

Conclusion

As you march toward your goals of a 100% virtualized data center, I hope the information provided in this article is helpful in understanding the process of virtualizing Microsoft applications. This article only covers the high points of the joint solution for Microsoft application virtualization. You can get all the information you need to deploy  this solution in the detailed, 50-page solutions guide, which  covers all the configuration details based on the careful work done by NetApp,  VMware, and Cisco. The guide covers FC, iSCSI, and NFS implementations.

In addition to the various links embedded in this article, other  valuable resources include:

NetApp and VMware vSphere Storage Best Practices (TR-3749). Best practices to implement VMware with NetApp storage.
Using the Performance Acceleration Module with Exchange 2007 (TR-3767). This technical report describes how PAM can boost the  number of Exchange users you can support without adding spindles.
SnapManager guides:
   SnapManager 5.0 for Microsoft Exchange Best Practices  Guide (TR-3730)
   SnapManager for MOSS: Best  Practices Guide (TR-3776)
   Protecting Exchange Server 2007 with NetApp SnapManager for Exchange (TR-3598)
   SnapManager for Virtual Infrastructure Best Practices (TR-3737)
SRM
   NetApp and VMware vCenter SRM Best Practices (TR-3671)

Got opinions about virtualizing MS apps?
Ask questions, exchange ideas, and share your thoughts online in NetApp Communities.

Author Alt Text

Abhinav Joshi
Reference Architect for Server and Desktop Virtualization
NetApp

When Abhinav joined NetApp in 2008, he brought over nine years of experience with data center consolidation and virtualization to NetApp. His current responsibilities include developing scalable reference architectures and best practices for securely integrating NetApp virtualized storage and data protection solutions with VMware virtualization technologies and Cisco Unified Computing System and networking technologies—solving customer problems and helping save cost. Since joining NetApp, Abhinav has been an active author, having led and participated in the development of many of the solution guides referenced in this article.

Explore

Please Note:

All content posted on the NetApp Community is publicly searchable and viewable. Participation in the NetApp Community is voluntary.

In accordance with our Code of Conduct and Community Terms of Use, DO NOT post or attach the following:

  • Software files (compressed or uncompressed)
  • Files that require an End User License Agreement (EULA)
  • Confidential information
  • Personal data you do not want publicly available
  • Another’s personally identifiable information (PII)
  • Copyrighted materials without the permission of the copyright owner

Continued non-compliance may result in NetApp Community account restrictions or termination.

Replies

Hello,

Re this paragraph:

Locate your application data (databases, logs, and so on) on iSCSI  raw device mapping (RDM) LUNs, directly created and connected inside the guest  VM using NetApp SnapDrive software (version 6.2 or higher must be installed  on guest OS).

Can the LUNs be mapped directly to an in-guest initiator or must they be actual VMware RDM?

Yes, LUNs can be mapped directly to an in-guest initiator but using RDM would allow for support for advanced features such as MSCS, SRM , and Vmotion.  You can also manage the storage from ESX server instead of individually from the vm's.

Hi,

MSCS inside VMware guests is a very hairy topic & in fact it's not supported at all on iSCSI, no matter how disks are mapped: http://www.vmware.com/pdf/vsphere4/r40/vsp_40_mscs.pdf. Having said that, MSCS normally works just fine in both scenarios.

From my experience VMotion works OK with software initiator.

SRM is another hairy topic, because during failover it reverts target volumes to the latest, 'dirty' SnapMirror update. So for that reason any LUNs containing data should be failed over using relevant SnapManager products anyway.

What I've heard though is that the performance is much better when using iSCSI RDMs, rather than software initiators.

Regards,
Radek

The problem we are having when engineering this solution specifically  for SQL Server is how the system databases are replicated to the DR  site.  We have vSphere 4.0 Update 1 at both locations and have SRM  working well.  We are using NFS datastores for the OS/binaries and ESX  ISCSI Software Initiators RDM LUNs for the application data as detailed  in TR-3822.  The issue I have come across that I cannot get an answer to  is how to replicate my System databases (which reside on their own  LUN).  Because of the inability (I understand this is an MS limitation)  to take a snapshot of the System databases (this is instead backed up to  the Snapinfo LUN), I am unable to bring up those databases in a  quiesced state on the DR side via SRM.  I could manually take snapshots  and then SnapMirror them over to the DR site but this is not quiesced  and would potentially be unusable.  The answer NetApp has provided is to  restore the System databases once over to the DR site but because you  have a non-quiesced copy of those databases, you are put into a "chicken  or egg" situation.  You need the system databases online to perform a  restore of the system databases but since they are not stable, you  cannot start the SQL Services to perform the restore.

The couple  recommendations NetApp made were to do a repair of the system databases  on the DR side during a failover and then do a restore....I would think  this is far from an optimal DR solution and almost impossible to  script.  The other idea offered was to perform a database "pause"  nightly and take a snapshot and replicate that system LUN to the DR  site.  I am not sure of the effect of this database "pause" on the  entire SQL instance.  This also does not cleanly tie into SMSQL as you  would expect.

As  a result of these limitations, we are now considering a DoubleTake DR  solution for our SQL Servers which is a shame with all of this great SAN  replication and virutal infrastructure.

The questions I have are:

  1. What  are other folks doing with respect to failing over their SQL Server from  one vSphere environment to another?
  2. Why isn't this MAJOR gotcha  documented in any TR reports?  Has this issue not been raised by  anyone?
  3. Am I missing a more simplistic solution to this problem  (I hope this is the case) or do I really need to look at a third party  solution for my DR despite the investment made in NetApp and VMware?

Any  assistance you can provide would be very greatly appreciated.

Thanks,

Joe

Here's another post by Abhinav that discusses the SRM / SQL Server DR scenario in great detail . VMware SRM and NetApp

I thought this would be another good reference for those looking for SQL HA/DR.  Disaster Recovery solution for Microsoft SQL Server

Hi watan,

I've read through these and neither seem to address the issue of how you failover a guest from one site to another with respect to the system databases.  If you are bringing up the same guest in your DR site relying on NetApp SnapMirror replication, their is no way that I see to replicate over the system databases in a guaranteed quiesed state like you do all the user databases.

If you have any other recommendations, I would certainly appreciate it.

Thanks!

Joe

Guys,

This gap apparently has been just closed!

I can't find any further details at the moment, but literally this week on Partner Academy I saw a slide deck describing SRM integration with SnapDrive instance at the fail-over site. This in turn allows rolling back to a 'clean' SnapManager snapshot.

Cool, isn't it?

Regards,
Radek

If you can get me any level of detail on this, beers are on me if you are ever in the philadelphia area!!!

You'll need to use SnapManager for SQL for this and you'll need to restore from SnapManager backup sets on the DR site if you run into any consistency issues. 


As Radek mentioned, the next version of SnapDrive for Window 6.3 will have integration between SMVI, SMSQL and, SDW which will allow for a cleaner way to recover.  Currently scheduled to ship end of July.

The issue I have using SMSQL to perform a restore of the system databases is that if they do not come up clean, I cannot start the MSSQL services to perform the restore from the Snapinfo directory to get the System databases back to form.  The only thing I can think of would be to do a repair of SQL and rebuild the system databases then do a restore from the Snapinfo directory at this point.  Do you agree that is what needs to be done if the system databases do not come back clean?

Our recommendation is to host the System DBs on a separate LUN (e.g. S:\) and perform a daily verified backup of the System DBs (along with SnapMirror update) using SMSQL. We also recommend performing verified backups when major changes are made to the System DBs.

This will ensure that you have application consistent snapshot of the system DBs to fall back upon in case the SQL Server fails to recover at the DR site.

After SRM failover, in case the SQL Server in the guest VM fails to start because of corruption with System DBs, our recommendation is to leverage SnapDrive in the guest VM and restore the LUN containing System DBs from the last verified backup of System DBs. Once the System DB restore is complete, start the SQL Server as you would normally do and verify the availability and functionality of the production DBs.

Worst case, if the System DB recovery is unsuccessful, rebuild the system DBs using the procedure described in this Microsoft KB: http://msdn.microsoft.com/en-us/library/dd207003.aspx

We will update both the TR-3785 and 3822 with this recommendation. Thanks for bringing this concern to our attention.

Hope this helps.

Regards,

Abhinav

I still think you are missing a big piece to this.  Because SMSQL does not store the  snapshot of the System database LUN  in that separate volume as you  describe but instead only updates the Snapinfo directory, you do not  have the ability to restore on the destination side using SnapDrive.  To go one step further, since snapshots are not stored in this volume at all, there is no System database volume on the destination side to restore from....only the Snapinfo directory which requires functional system databases to restore from....hence the chicken and the egg problem.  The only solution I can think of is what I previously noted of repairing system databases and then restoring from the Snapinfo directory at that point which would make it a large challenge to automate this process for DR.

@jwhelan27

If the system DB is stored on it's own LUN which is on it's own separate volume, then SMSQL will not replicate System DBs to the DR site when it triggers a SnapMirror update to occur at the end of a backup job.  However, if the system DB is on a LUN that is stored in the volume that also contains the Production DBs (the DB with your data in it) then it will be included with the replication of that volume.  The snapshots on this volume will contain quiesced system DBs along with a quiesced Production DBs. This is because system DBs are quiesced at the same time that the Production DB is quiesced. The LUN with the system DBs is quiesced by snapdrive at the same time that the LUNs containing the Production DB is quiesced. So a quiesced system DB and LUN is captured in the same volume snapshot. The designs discussed here and in NetApp TR-3785 and

TR-3822 describe storing the system DB in a volume with the Production databases.

Hope this helps clarify.

The reason the system luns are separated out from the user db's is by following the best practice guide recommendations.  I was also told from Sourav at netapp that even if that configuration was implemented, the system db's will get replicated but still would not be in a quiesced state due to the nature of what they are.  If you tried to use that LUN without any kind of restore attempt from Snapinfo, there is still no guarantee that the system dbs are quiesced.  This only assists in replicating the dbs to your DR site.

Just as an update to this subject....I have been in communcation directly with folks at NetApp regarding this specific item.  So unfortunately it does look like I am correct that there is no way to replicate a guaranteed quiesced copy of the System LUN over to the DR site with the current NetApp tools.  The latest version of SDW 6.3 still in development will not be addressing this issue.  A standby SQL Server also has similiar limitations since the system database are not replicated in these scenarios either, it is on the customer to script a procedure for regularly backing up all system database info (security items, stored procedures, etc….) that would be necessary to properly bring up these user databases on the DR side. 

Just for clarification, if the system database is corrupt on the DR side, you cannot just use the SnapInfo data to restore it because of the “chicken and the egg problem”.  In order to restore from SnapInfo, you need the SQL Server instance running but in order to get the SQL Server instance running, you need to restore from SnapInfo.  The only real way around this is to rebuild the system database as documented in http://msdn.microsoft.com/en-us/library/dd207003.aspx or to do a repair from the SQL Server 2008 media.  Once you would have the database back to an “out of the box” state, you should then be able to perform a restore with the SnapInfo data.  The best recommendation NetApp engineer have at this time is to use SnapDrive to create a crash consistent snapshot of the system database LUN (database status unknown however), then mirror it.  If the system databases come up corrupted, then you must perform the previous steps I just noted.

Hope this helps anyone running into this similiar issue.  If anyone finds a better workaround to this issue that ties nicely into VMware Site Recovery Manager, I would love to hear from you.

Just for future reference, here is another excellent document that discusses the NetApp + SQL solution.

Accelerating Development of Microsoft SQL Applications in Heterogeneous Environments

Public