Last updated
Last updated
When a source is too big for a single tape, Canister's Spanning engine comes in handy. This feature makes it quick and easy to Archive datasets that are larger than a single tape.
When Canister for Mac is presented with a source that won't fit on the mounted tape, it enters Spanning mode. Once this happens, you'll be prompted to set a Spanning Label, which makes it easy to find your data later in the Catalog.
In turn, Canister will group the tapes for each Spanning part like this:
From here, rinse and repeat for as many tapes as it takes to complete the Archive.
It's likely that you won't fill up the last tape in a spanning set. Therefore, you can add anything you want to that last tape at a later time - no previously archived data will be affected. If the to-be-archived files fit, Canister will add it to the tape. If not, you'll enter the Spanning flow described above.
Pro Tip: If the data belongs to an existing Spanned Archive, Mac users might find it useful to set the same Spanning Label as before. This way, the newly added data gets grouped in the same Catalog.
Spanning is an advanced feature that tends to lead to a few workflow questions. Be sure to check the FAQs below before reaching out to support.
No, Canister will ask you to format at the relevant points in the Spanning workflow. Insert the next tape, then follow the prompts.
This comes up quite a bit; the best bet is to name each tape sequentially, for example:
Archive_pt1
Archive_pt2
Archive_pt3
Canister doesn't enforce any limits, however Spanning is best used with dataset that requires between 2-4 tapes.
Why? Because Spanning requires Canister keeping the entire source index in memory, which is a heavy lift. The chance of something going wrong on the hardware side of LTO increases exponentially the longer your LTO drive operates. As an upper limit we recommend Spanned Archives containing less than one million items.
As with all things LTO, hiccups happen. While with hard disks it's easy to redo something, due to LTO's linear nature errors tend to be showstoppers - at least temporarily. While Spanning, the chance that a tape runs into an issue or that an archive fails for some reason is, of course, compounded.
Canister gets around this problem by using its Dupe Detection implementation to the fullest: in case an error happens with any tape during a spanning archive, simply start over. It's important to feed Canister the tapes in the order they were finished, then let Duplicate Detection skip items already existing on each part.
While Canister supports Dupe Detection, due to LTO's linear nature, we don't recommend incremental updates to a Spanned Archive. File changes are likely to end up distributed over multiple tapes, which will get complicated when retrieving.
Splitting files between multiple tapes is not a standard LTFS feature. As a result, each tape is likely to have a small pocket of space at the end.
Canister for Mac does not support Spanning while creating simultaneous archives. Canister for Windows follows a different flow, which allows the next tape to be inserted for each drive.
