Tiered storage device #151
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…o page-blob-device
… tests to also run for AzurePageBlobDevice.
# Conflicts: # cs/src/core/Device/AzurePageBlobDevice.cs
…k in commit point for write request.
# Conflicts: # cs/src/core/Device/LocalStorageDevice.cs
…into tiered-storage-device
…red-storage-device # Conflicts: # cs/src/devices/AzureStorageDevice/AzureStorageDevice.cs # cs/src/devices/AzureStorageDevice/BlobEntry.cs # cs/src/devices/AzureStorageDevice/FASTER.devices.AzureStorageDevice.nuspec # cs/test/BasicDiskFASTERTests.cs
Merged
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Rough draft of tiered storage support in FASTER. So far it is assumed that all storage are inclusive (higher tiered storage are considered caches of lower tiers), and the bottom layer is assumed to always be unbounded in capacity. There are no guarantees of synchronizations between layers at any point, although eventually all writes will be issued once to layers that store the address range. (I don't believe there are ordering guarantees)
Two things that need to be elaborated and worked on:
What level of concurrency is expected to happen at this level? TheIDevicelayer certainly sits under the highly concurrent frontend, but tiered storage brings more complexity and we may need the epoch protection framework in this layer.What happens on recovery? Recovering ranges of each storage device is not supported right now. Depending on the commit point, there might also be some recovery work to do for the content on storage (they might not be always consistent)Fix #138
Dependent on #147