I have asked a couple of questions related to this business requirement and realised I may have been approaching things backwards. Rather than asking specifics, I should state the requirement and ask for the best general way to approach the problem using CC ImageServer (we are using the free version of ImageServer [2.0.12655.37768] as our production PACS).

The requirement is this: Our IT department wishes to do a single backup of all existing studies in our PACS, and to minimise the amount of backups they need to perform on an ongoing basis. At the same time they will be installing another drive into the server (actually a storage array but it will appear to the OS as a single drive). One drive will then be used for the existing studies, and will be backed up once (the "Backup" drive) and the other will be used for all new studies (the "Live" drive). It doesn't really matter which drive is which, but it would be easiest if we didn't need to move all our existing studies from one drive to another (i.e. the existing drive becomes the Backup drive).

We have informed them that historical studies, although "almost" static, will occasionally need to be edited for day-to-day data cleanup operations. So we can guarantee that at least some of the backed-up studies WILL need to be edited at some point in the future, but we can't predict which ones. So these studies will need to come off the Backup drive onto the Live drive.

We also have some of our own software tools, written using the CC API, which access the PACS. It would be best for these tools if operations such as MOVE, FIND etc. behave the same for studies on the Backup drive as they do for studies on the Live drive.

So, in a nutshell we would like to:

- Backup existing studies once from the old drive

- Have new studies stored on the new drive

- Continue to use old and new studies as if they are all Online

- Have any old studies which are subsequently edited automatically moved onto the new drive

What would you recommend as the best approach to this issue? Would you use Storage Tiers or Partition Archives? And how would you use them?

The software itself really doesn't have the capability to perform what you're asking. Although the Partition Archive mechanism can store a "backup" on another drive, it was intended to work on an HSM based filesystem that never "fills up". Use of a partition archive on a fixed size drive is not really recommended. We've had users run into problems with archiving to a fixed size disk, and it requires active monitor to ensure the archive disk does not fill up.

Similarly, we don't have the capability to "replicate" a filesystem onto a different disk in the software. Technically, you probably could set something up to synchronize the drives with non-ClearCanvas software, but we've never tested such a solution. I'd assume that there might end up being issues with collisions between the synchronization software and the ImageServer software itself that may cause study corruption.

So, I'm not really sure if we can offer you a tested solution on how to address your situation.

Steve

Steve: I'm in the same position as the O.P, so thanks for the info!

If we do go ahead and use the Partition Archive ... what failure modes have you seen when it fills up? Is there any danger of losing or corrupting image data?

The main issues are around the potential of the archive disk filling up. There is no code to handle this, and the archive requests will simply fail. This would require adding an additional archive, and there's no way through the GUI to move archive requests to another archive, since the first would be filled up. The maintenance of this configuration just doesn't make sense, and in most cases, our users who have installed these archives would just be as well off with keeping all the data online.

We also recently found a bug, #9673, in partition archive code related to restores when a user has multiple filesystems.

Steve