Joel Oleson's Blog - SharePoint Land

SharePoint Backup Restore, High Availability, and Disaster Recovery

This post has been long comming.  I recently spent some time with Samer Sawaya of SharePoint Recycle bin fame :).  We have some good memories working through that one.  You can thank me later for the 2nd stage not getting cut, I fought hard for that.  We walked through his backup presentation a few weeks ago and remembered I hadn't done any posts on this topic.  Although this has been a hot topic in the past I haven't allocated much time getting up to speed.  I did spend some time with AvePoint at ITForum even got a bite to eat with Y.  Quest's acquisition of Imceda's, and hence SQL Litespeed, adds more to the Quest arsenal of SharePoint tools.  Partners aside, here's a good overview of backup/restore including tools, HA and DR.

What to Backup

1.     All Servers and all drives in the farm including system state: includes metabase, system state, home directories, web.config, install path, custom assemblies, binaries and code, customizations, site defs, list defs, IIS logs, evt logs, etc… (Doesn’t even need to be daily if this concerns you…)

2.     Databases – All your data and 99.97% of config is stored in SQL

3.     Index – The index will be propagated to the query servers, but you can’t recover it from there if you have system failure, drive, sector, or file corruption.  The two ways to backup your index is via Stsadm –o backup and with the central admin.  The two can be used interchangeably.

Tools and APIs

·         Native SQL Server Backup & Restore – Native SQL backup can be used to backup all databases (Note: Does not include index (even though you’ll see it can backup the search and SSP database.)  You can even use the jobs to schedule or create some interesting scripts to do multi threaded backups.

·         SQL 2005 Database Mirroring – mirrors of your data in another location for location, data, and hardware fault tolerance.  See the White paper: Using database mirroring for details.

·         SQL Log Shipping to a Standby farm or server – pretty much the same as with SPS 2003.  You can use this with SQL 2000 or SQL 2005 to a read only SQL box connected to a stand-by farm and fail back or read/write and log ship back.

·         SharePoint Backup and Restore (Central Admin) – UI based backup for farm including services like Index… includes full and differential backup. (Requires a full first, to perform any diffs.)

·         Stsadm Utility – stsadm –o backup/restore, stsadm –o export/import, the stsadm command is pretty powerful.  You can schedule this with a scheduled task to run nightly, failures will result in an error in the error log which you can capture in MOM or your monitoring environment…  Includes full, differential, partial backups of services or individual web apps, site collections, sites

·         VSS Writer – Snapshots can be completed to capture point in time backups of the databases.  (See note in gotchas below.)

·         Web delete event— The events are in the system and can be captured.  MS IT has a working prototype of a site and site collection archive on delete solution deployment they plan to share after they "bake it in".

·         SharePoint Designer—you can use SharePoint designer as an end user means of backing up a site or site collection.

·         Recycle Bin – one of the best new features of the product.  Right out of the box when a web app is created, files, lists, and list items are retained for 30 days (configurable) before deletion.  The second stage, the site collection admin stage allows you to recover documents deleted by users (even the site admin him or herself)

·         Migration API (Prime)  – This API is a powerful way to copy (import/export) data with SharePoint Products & Technologies. (see the SDK for more info)

·         DPM 2.0 (beta) – Data Protection Manager 2.0 is being designed to backup SharePoint Farms and is early in the beta cycle.  This disk based backup solution is designed for consolidated backup solutions and differencing.  Look for more on this in the future.  (This is a separate Microsoft product, but plans to have a SharePoint solution)

High Availability

·         Web Front End: Web front ends can be easily made redundant via load balancing solutions, software, hardware… either work.  Be sure to use sticky sessions and be sure to test forms deployments with your load balancing solutions.  NLB is the most commonly used intranet solution, Cisco Local Director, F5 Big IP, etc… are common hardware load balancers

·         Query: Query aka Search Servers are easily redundant through the services built into the product.  All web servers need to communicate to all query servers even if there is a query service local to the box. 

·         Index: This is a single point of failure as a service, but if the index box goes down, the users won’t be impacted.  If the index box goes down and you bring it back up service will not be interrupted.  Exceptions to be aware of are if the query box was in the middle of propagation and then the service were to go down.  The query box needs to think it is up to date and has the correct catalogs.  If the query ever thinks it is out of date and is unable to contact the Index box, you will get errors on search requests.  The most creative but ugly scenario for making your index redundant is to create an additional farm that is a single box that has indexing services on it.  That single box farm would have a copy of the index or have it’s own indexing schedule.  Thus if a failure were to happen you could update the connection of the SSP to the other farm.  Not sure who’d do this, but it’s possible.

·         SQL – SQL Clustering provides for a high available SQL platform.  Bascially it’s shared disks with failover between systems ensuring fault tolerance to hardware issues.  You can do clusters in either SQL 2000 or SQL 2005.  You can do SQL Log shipping on top of clustering or without it for further fault tolerance and high availability.  SQL Mirroring is another option is you are using SQL 2005 more detail above.

Gotchas: (Snipped from Using Backup and Restore (Office SharePoint Server 2007) TechNet article)

STSADM Backup and restore will not backup and *restore* the following:

·         The configuration database

o    Any Custom solutions deployed (be sure to keep the solutions so they can be easily re-deployed)

o    Alternate access mappings

o    The Central Administrator Web Application

o    The Central Administrator content database

o    The Internet Information Services metabase

Additional Data on VSS

·         The VSS Writer service needs to be registered using the “stsadm –o registerwssservice” command.

·         The SQL Server VSS Writer service, which is available with SQL Server 2005, needs to be started for the Windows SharePoint Services Writer to work properly. By default this service does not start automatically. For more information about Volume Shadow Copy service (VSS), see Volume Shadow Copy Service Technical Reference and for more information about the SQL Server VSS Writer, see SQL Writer in SQL Server 2005.

Additional Resources: 

• Post on backing up Project Server 2007
• TechNet: Using Backup and Restore (Office SharePoint Server 2007)
• TechNet: White paper: Using database mirroring with Office SharePoint Server
• TechNet: Use the Central Administration user interface to backup and restore Project Server 2007
• Using Backup and Restore (Windows SharePoint Services 3.0)
• White paper: Using database mirroring with Windows SharePoint Services
• Great article on backup strategies on MOSS 2007 from AvePoint. 
• Quest has a great overview of backup/restore and DR.

Some Third Party Vendors:

AvePoint: Farm and Granular Backup and recovery (see DocAve 4.1 for MOSS 2007 and WSS 3.0)

Neverfail: High Availability and DR solutions

Quest: Compressed database backup with Litespeed and Recovery Manager (SPS 2003 and MOSS 2007)