If you want to reduce the size of a project archive on a current job then the methods above are all good suggestions. They seem a little bit manual but a good way to tidy up an existing project and compress it down to its core size. @justinbromley I have been playing with a python script that will clean up and remove all but the last five batch iterations for all of your shots. One button. Wham!
But if you are trying to reduce the size of your archives then I have an alternative approach. I guess the question is what gets backed up? Not just the Flame project. Probably the job folder on the server as well.
How much of what is in your job folder is also in your Flame archive? All of the graded footage probably. The CG renders. If this is sitting in the job folder and also included in your Flame archive you are doubling up.
Our average size of a job folder on a TVC, containing CG, is 2TB. The Flame archive comes in at 90GB.
By not caching the media on your Archive you archive only soft links. By using Write nodes instead of Render nodes all renders go to the server and brought back in by using OpenClips. The file setups also get saved to the server by way of the Write nodes. The conform is published to shot folders on the server so that all source material lives outside of the Flame.
We still use Render nodes for quick test renders but we make sure that the render destination is to the Batch Renders Library which gets cleaned out before archiving.
This approach removes the Flame archive as the central storage for all of the comp. It keeps it small and manageable. It still contains all of your latest batch groups and significantly your timelines. If you need to tweak the end shot for a future iteration you just restore the folder structure for that one shot and your tiny Flame archive. You can locally cache your media again to help keep thinks moving fast. You can even get into a single shot without the Flame archive if you needed to.