SharePoint Farm Backups and Recovery plan

This post describes a recent Backups and Recovery plan I executed. It aims to describe my thought process, decisions and challenges faced while designing this plan. It could also be applied to SharePoint Farms with topologies different from the one mentioned below.

SharePoint Farm Topology: 3 server farm consisting of a web, app and database server.

Restore Requirements:

Our Requirements are described in the table below:

þ : Client Requirement                 ý: Not a Client Requirement

Requirements No.	Restore Type	a)In Place	b)Out of Place	c) Workflow History
1	Item	þ	þ	ý
2	List	þ	þ	ý
3	Site Collection	þ	þ	þ
4	Platform	þ	þ	þ

FYI, In-Place means restoration on the same farm and Out of Place restoration on another farm (such as in case of a Disaster).

Other Requirements:

5) All backups listed above needed to be fulfilled using a scheduled job

6) Incremental Backups needed each day during the week and Full Backups during the weekend

7) The solution needed to back up the database transaction logs daily

8) The solution should be complete and eliminate the need for Sql Database backups

What is so special about Workflow History: You might be wondering why I gave so much emphasis to Workflow History in the client requirements. The reason is that very often Workflows are implemented for critical business processes and the most important data in them pertains to who said what? when?, i.e: the audit trail.

Now consider the scenario, you have a custom list with a workflow running on it. For some reason you loose a list item, or for that matter, an entire list and need to restore its data along with its workflow history, can you restore it individually from a backup? The answer is: there is no product I know of that will let you restore a list item or an entire list along with its own Workflow History. Why is that so, because Workflow History is one hidden list with its data referencing both, a task list and the actual sharepoiont list on which the workflow is kicked off. So when we were designing this backup solution, it was important for us to know that the only way to restore a sharepoint list with its workflow history was to restore the ENTIRE site collection. We could not go more granular than that.

Solutions Considered:

a) OOTB: SharePoint would only be able to support 3 a), b), c), 4 a) and 4c) OOTB.

1 a), b) or c) would not be possible. 2 a) or b) would be possible but you would have to find the powershell command and write a scheduled job for it. It would be hard to manage such a job for an ever growing farm with new lists being added daily.

As you can see, SharePoint offers very basic Backup and Restore options OOTB and would not meet the requirements of many Enterprise SharePoint Farms. I decided to evaluate third party products.

b) VM Snapshots: Since the web, app and sql are all running virtual in this case, we could take vmware snapshots but that would only work for 4 a) and for no other Requirement. Another downside is that the resotration process is out of control of SharePoint Administrators and we would have always needed to depend on the Infr. team for restoration. Lets assume the Sql was physical and web and app were virtual ( as in many large SharePoint Installations), VM snapshots would still be able to restore the web and app servers and you would still need to depend on Sql backups as described in c ) below. A combination of VM Snapshots and sql backups could satisfy requirements 3 a), 3 c) along with 4 a), but we would still be left with not implementing requirements mentioned in 1) and 2).


c) Sql Backups : Sql Backups could be used to satisfy Requirements in 3 a), 3 b), 3 c) and 4 a), but we would but we would still be left with not implementing requirements mentioned in 1) and 2).



d) DocAve Backups and Recovery (AvePoint): After evaluating several third-party products, I thought the one that gave us the maximum value was DocAve. This product seems to be the most mature product in this product space. It would work for all of the above mentioned requirements, except any of them related to Workflow History. Event though you can restore a Site Collection containing Workflow History for several Workflows using powershell, DocAve does not support it. We had to write a scheduled job separately for that purpose as explained in my previous blog post located here. Avepoint technical support so far has been pretty good and that is one more reason I would like to stay with DocAve.

Conclusion: Adopting Strategy d) was the only way we could implement all the Requirements mentioned above. A combination of strategy a), b) and c) could be used to satisfy a subset of the above Requirements.