cancel
Showing results for 
Search instead for 
Did you mean: 

How do I migrate from MSCS to SFWHA?

Gonzalo_Gomez_G
Level 4
Partner Accredited

Hi, I'm in the middle of a proposal and the customer ask us to describe the way (step by step) to migrate their MSCS (2003) to SFWHA (VCS)? do you know if Symantec has a procedure? Any recommendation? The customer requires minimal impact and smooth process?

Thanks for your help.

Gonzalo Gomez

1 ACCEPTED SOLUTION

Accepted Solutions

Wally_Heim
Level 6
Employee

Hi Gonzalo,

The switch to SFW-HA is dependent on what is being clustered.

 

Fileshares may not be too much problem depending on how they are clustered is MSCS.  NTFS permissions will be not be changed by upgrading the physical disks to dynamic.  Depending on how share level permissions are being handled there might be some extra steps involed.  For the most part, I would guess that these are user home directories and the permissions are handled at the NTFS level and the share permissions are open to everyone FULL_CONTROL.  In this case the disks would need to be upgraded to dynamic and the fileshare configuration wizard is SFW-HA would need to be run to create the Fileshare service group(s).  The process is very simple and straight forward.  However, with 3000 shares there are bound to be some that are not user home shares and would need more attention. 

 

SQL is a MSCS aware application and installs differently in MSCS than in SFW-HA.  Because of this SQL will need to be reinstalled to work with SFW-HA.  The only tricky part with this is that each SQL instance needs to have its master database installed to a temporary drive that has the same path as what was used during production.  Then it is a matter of stopping SQL, upgrading the production disks to dynamic, swapping what drive is in the mount path to SQL and running the SFW-HA SQL configuration wizard to create the service group for SQL.  Since the master database is swapped to the production database, SQL will mount all user databases as normal.

 

Oracle is not installed differently in MSCS that I"m aware of.  Swtichting the disk to dynamic and running the SFW-HA Oracle configuration wizard should be all that is needed.

 

I'm not familiar with the HP applicatoin that they are using.  I'm not sure how it works.  This would require some investigation to be 100%.  In general, I would guess that this application is controlled in MSCS with a basic generic service type resource.  If this is the case, then in SFW I would create a basic Fileshare service group, remove the fileshare resource and replace it with GenericService resource to control the HP service.  Multiple GenericService resources maybe needed depending on the makeup of the HP services.  A RegRep resource might also be needed.

 

These are very basic and highlevel overview of what needs to be done.  Each cluster should be examined and a plan made for how best to migrate that cluster and its resources.  There might be special considerations needed for each cluster or for only a few of them.

 

From your PDFs, it looks like your customer will continue using mirrored volumes between the sites and not VVR or hardware replication.  If they have redundant links (with low latency) for heartbeats they might want to use Replication Data Clusters.  If the links do not meet the needs for heartbeats then a GCO configuration would work.

 

Thanks,

Wally

View solution in original post

7 REPLIES 7

Robert_Lindenbu
Level 3
Employee Accredited Certified

I don't think this is possible. For most applications (SQL Server being the one I am most familiar with), there are actually different binaries installed for standalone vs MSCS installs. SFWHA uses the "standalone" version of the application.

 

What applicaitons do you currently have in the cluster?

Wally_Heim
Level 6
Employee

Hi Gonzalo,

The steps needed are very dependent on what is being clustered.  As Dexter mentioned some applications, such as SQL and Exchange, are cluster aware and install differently in MSCS verses SFW-HA.  These applications would require a different set of steps and if you were migrating a non cluster aware application or if you clustering say a file share server.  Without knowing more details of what you are starting with it is almost impossible to provide you with a step by step process to get you to where you want to be.

Thanks,

Wally

Gonzalo_Gomez_G
Level 4
Partner Accredited

In advanced thanks for your help.

Description: Our costumer has six (6) MSCS; distributed One File Server Cluster MCSC, three SQL 2005 MSCS, One Oracle Cluster MSCS, and One HP Openview Application in MSCS Cluster, Total 12 servers.

The Data is managed by the Windows Disk and volume services. They want to migrate to SFW Ent or SFWHA; but they are concern about the  migration process between those MSCS and VCS; they ask things like what happen with shares when We migrate from MSCS to SFWHA because they manage like 3000 shares and ACLS and the don't want to recreate all these shares. Additional, the want to know before to decide if buy SFW or SFWHA if the migration or conversion process is to complex and the had to be too much time offline or if this is too risky.

I'm adding a .pdf file with a draw with the options they want to add (at these time they don't have any Symantec Storage foundation installed everything is based on MSCS) to these windows environment, We want to sell SFWHA but the customer only buy this option is they know in advanced what kind of changes We have to do in theirs environment.

Thanks,

Gonzalo Gomez

eu22106
Level 4

Gonzalo,

I'm very curious. What is the reason for migrating to SFWHA? Has SFWHA some features that MSCS does not have? What version of MSCS do they want to migrate from?

Ivo

Gonzalo_Gomez_G
Level 4
Partner Accredited

Hi, Ivo thanks for your interest.

At this time this customer has a very complex requierement and is to provide DR testing with production servers and migrate production to DR site and viceverza each two weeks all year long; they want to have tha maximun performance over the failover and failback process, the most automated, and if is possible the oportunity to test the DR site (like Fire Drill does) before they launch they hot DR test.

At this tiem they are doing this with MSCS and a extended SAN and replicated Storage but takes like 2 or sometimes 3 hours to have the 6 MSCS cluster from one site to the control in the DR site? the have to do a lot of task in the Storage and replication storage software to indicate the replica direction and has to do so many things with the MSCS volumen manager and configuration to have the DR site ready.

That's answer your question?

 

Thanks beforehads for your help,

Wally_Heim
Level 6
Employee

Hi Gonzalo,

The switch to SFW-HA is dependent on what is being clustered.

 

Fileshares may not be too much problem depending on how they are clustered is MSCS.  NTFS permissions will be not be changed by upgrading the physical disks to dynamic.  Depending on how share level permissions are being handled there might be some extra steps involed.  For the most part, I would guess that these are user home directories and the permissions are handled at the NTFS level and the share permissions are open to everyone FULL_CONTROL.  In this case the disks would need to be upgraded to dynamic and the fileshare configuration wizard is SFW-HA would need to be run to create the Fileshare service group(s).  The process is very simple and straight forward.  However, with 3000 shares there are bound to be some that are not user home shares and would need more attention. 

 

SQL is a MSCS aware application and installs differently in MSCS than in SFW-HA.  Because of this SQL will need to be reinstalled to work with SFW-HA.  The only tricky part with this is that each SQL instance needs to have its master database installed to a temporary drive that has the same path as what was used during production.  Then it is a matter of stopping SQL, upgrading the production disks to dynamic, swapping what drive is in the mount path to SQL and running the SFW-HA SQL configuration wizard to create the service group for SQL.  Since the master database is swapped to the production database, SQL will mount all user databases as normal.

 

Oracle is not installed differently in MSCS that I"m aware of.  Swtichting the disk to dynamic and running the SFW-HA Oracle configuration wizard should be all that is needed.

 

I'm not familiar with the HP applicatoin that they are using.  I'm not sure how it works.  This would require some investigation to be 100%.  In general, I would guess that this application is controlled in MSCS with a basic generic service type resource.  If this is the case, then in SFW I would create a basic Fileshare service group, remove the fileshare resource and replace it with GenericService resource to control the HP service.  Multiple GenericService resources maybe needed depending on the makeup of the HP services.  A RegRep resource might also be needed.

 

These are very basic and highlevel overview of what needs to be done.  Each cluster should be examined and a plan made for how best to migrate that cluster and its resources.  There might be special considerations needed for each cluster or for only a few of them.

 

From your PDFs, it looks like your customer will continue using mirrored volumes between the sites and not VVR or hardware replication.  If they have redundant links (with low latency) for heartbeats they might want to use Replication Data Clusters.  If the links do not meet the needs for heartbeats then a GCO configuration would work.

 

Thanks,

Wally

eu22106
Level 4

So the overall added value will be the DR and testing the DR?  Then there is indeed a good reason. Switching to a different site takes us about 2 minutes.....using mirrored diskes.
 
I think that if your shares are already on SAN you can import the disks to the Veritas server, then mirror the disks. (You need SAN on both locations) This would take hardly any time. Probably the resynch will take some time dependant on the size of the disks (but that can be done inflight.)

Ivo