Forum Discussion

mmorocco's avatar
mmorocco
Level 2
11 years ago

Clearwell Backup Best Practices

I know its probably taboo to post a quetsion like this and ask for information so many people arent willing to share (why I dont know), but I dont see why it hurst to ask.

I am curious about best practices others are adhering to in backing up their cases/servers.

Because of the current CW architecture, backups in my opinion are troublesome.  Aside from random failures (i posted another topic about this) I have always found the process that is needed to fully backup and prepare for BR/DR to be very cumbersome with Clearwell.

Do you agree?

Hopefully v8 and the support of Virtualization will help this.

But does anyone care to share how many days of backups they are keeping?

Are you still doing node backups?  And if so, why?

Do you provide clients with specific terms&conditions about nightly backup windows, not allowing case/system access during certain times, and if not, how are you ensuring you get nightly backups and dont miss them because of clients accessing cases? 

How do you workaround/decide/manage taking backups vs processing or OCR'ing data overnight?

You see where I am going with all this.  There are more questions than this that I have been grappling with Clearwell for years.  These are topics I find to be in constant flux, requiring constant revision, because there just doesnt seem to be a very efficient way of protecting/backing up the CW environment.  Dont you agree?

I thank you in advance for anyone willing to share, as I would be willing to do so in return.

I am happy to conduct individual discussions privately if anyone is interested.  If my contact info is not available in this profile, please respond here and we can coordinate a better way of communicating.

Backing Up
Best Practice
Configuring
eDiscovery Platform
Information Governance
Managing Backup Devices
Monitoring
Troubleshooting
Windows Server (2003-2008)
  • jonathan_zak's avatar
    jonathan_zak
    Level 3
    11 years ago

    Hi Matt,

    I've seen many clients schedule case backups to run each week night. Schedule case backups to run as five separate tasks (M,Tu,W,Th,F) as set each to run "Weekly."

    A backup schedule may run like this:

    Monday Case Backup (run 23:00, Weekly)

    Tuesday Case Backup (run 23:00, Weekly)

    Wednesday Case Backup (run 23:00, Weekly)

    Thursday Case Backup (run 23:00, Weekly)

    Friday Case Backup (run 23:00, Weekly)

    Since a case backup runs on each day of the week, you will always have at least one week's worth of backups. 

    I would set a general workflow to make sure that all of your reviewers are out of a case by a time in late evening - maybe 23:00 - when the backups begin - basically let them know when they are going to begin, and that in order to have full backup of their case work product, they will need to log off of their current case.

    Of course, this is not always doable, depending on case deadlines, and other factors. When users are still logged in while a case is starting backup, the backups will fail - I'm sure you've experienced that before.

    If you have active cases where you're actively reviewing processing/OCR/producing and you can't run a nightly backup I would consider backing up these cases on an alternating basis (so you say backup this active case M-W-F each week) or if you know that the case is going to be active 24x7 then set aside a weekend day (preferably Saturday) to schedule a full case backup on the Active case - it is not optimal but better than not having any case backups at all.

    I would not recommend running case backups to output to the appliance D: drive under any circumstances, even if temporary. If the D: drive were to become inaccessible then all of the case backups would be lost. I highly recommend configuring the case backup directory to write to External storage, off Clearwell appliance.

    As far as node backups are concerned, I highly recommend running one each week - it is a task that can be scheduled in Windows (The Clearwell System Admin Guide, it provides instructions on how to configure the appropriate network directory off appliance to write the backups to).

    In the event of an appliance failure a current node backup provides a faster way to restore the appliance quickly to a working state. I recommend running the node backups on a Sunday, when there is usually no activity on the appliance - so set the Windows scheduled task to start at 03:00 in the morning and let it run. Oftentimes the node backup runs for hours, so at least on a Sunday you have a full day to let it complete.

    I hope this helps!

    Jon

     

     

     

  • moorrees's avatar
    moorrees
    Level 3
    11 years ago

    I recognize a lot of points you mention in your post :)

    The 'issue' with the backup window and doing OCR jobs, there is a workaround for this now. You can schedule OCR jobs and have stoptime as well.

    If you do casebackups, and only have a short backup window, you probably best set your backup location to the local machine since this can be about 20 times faster than a CIFS network location. After the backup is finished you can then robocopy (schedule) the backup data to a network location.

    For every case you have to ask how often the customer needs a snapshot. We have cases with a lot of reviewers where the client would e seriously hurt if they even lose 4 hours of work. This means you have to make sure that every 4 hours all clients are logged out of this case so you can do an individual case backup.

    Doing easy full BR/DR is hard! i have been looking at ways to create the appliance where the D drive is an iSCSI target on a SAN that does synchronous mirroring to a second SAN so you could have a warm standby.But if you have a second machine on a standby location, you'd also need licensing for the standby machine.I really hope they would participate in duscussions around availabiliy. It will then still take several new versions before they could actually implement it, but right now i am not sure if they are even considering stuff like this :)