Managed servers – Maintain an offsite DFS

Distributed File System – another great application for a virtual private server

DFS is remarkable technology that distributes files and folders over multiple replica servers.  File servers that are part of a DFS replication group will maintain their own copy of every file including permissions, ownership, and modified times.  Whenever a file is added or edited or deleted, the change is replicated over all DFS members.  Other features include remote differential compression which means it only copies over the portion of the file that has changed and also compresses the data stream.  DFS also has configurable bandwidth throttling with time-of-day controls to keep bandwidth usage under check during peak hours.   The technology is well tested and though it has some limitations it is very useful in maintaining an offsite replica for disaster recovery. 

Two important considerations with disaster recovery are the recovery time objective (RTO) and the recovery point objective (RPO).  RTO is the amount of time that the business can be without the service; in regards to DFS we are considering how long you can go without being able to access your files.  The RPO is the amount of time that could be lost because of the failover.  For example, if your RPO was two days then in a disaster recovery scenario you would be satisfied with restoring data backed up two days prior:  You could afford to lose two days worth of work.

An offsite DFS replica allows for dramatically decreased RPO and RTO.  Let’s run through an example:

  • 16:00:  All files are synchronized between the onsite file server and the offsite replica
  • 16:01:  The company’s main file server crashes and will not start.
  • 16:30:  Onsite IT determines that a major hardware failure is to blame and the server will not be coming back online any time soon.
  • 16:45:  After a quick phone call and a look to confirm the share setup, the onsite IT staff switches the group policy to use mapped drives shared from the offsite DFS replica.
  • 16:50:  Onsite IT test the new policy and find that it works.  Users are notified to log off and back in
  • 17:00:  Users have logged off and back in and are accessing data from the offsite replica.
  • 17:30:  A new customer-premises server has been brought online and files are being restored from the locally maintained backup.  The local backup is almost a day old so the newer files are copied from the offsite DFS using robocopy.  Once this done, the restored server is joined to the DFS replica group and group policy is changed so that drive mappings are directed to it instead of the offsite replica.

In this example an RPO of one minute was achieved with an RTO of one hour.  Performance will be much lower when running from the offsite replica which is why you would still want to bring an onsite server back online as soon as possible.  However, you can still access your data so provisioning a new server would be more an ‘urgent’ class rather than ’emergency’.  Your RPO and RTO should be much shorter with an offsite DFS than without.  Offsite DFS data and the connection to access that data are highly available: It is always there waiting to be used.  Let’s compare this to recovery using a daily backup generated from the built in Windows backup tool:

  • 16:00:  The company’s main file server crashes and will not start.
  • 16:30:  Onsite IT determines that a major hardware failure is to blame and the server will not be coming back online any time soon.
  • 16:45:  Replacement server is booted with recovery CD and windows image is pulled from a network attached storage device. The previous night’s backup was generated at 10:00 pm and looks good
  • 23:30:  The 750 GB server image is restored
  • 0:00:  The server passes testing from onsite IT and is ready for use.

Here we’ve achieved a recovery time of 8 hours and a recovery point of 18 hours.  You’ve lost a full day of productivity mainly because the recovery environment didn’t exist; it had to be created.  No data could be accessed until the entire server image was restored from backup.

At Radish Networks, we charge no additional fees for monitoring and managing DFS as part our managed VPS service.  Contact us for more details.