batch updater for counter, timer, and dist summary #1003
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
Adds a helper that can be used to batch updates to the delegate meter. The main use-case for this is timers or distributions summaries that need to be updated from a single thread with a high throughput. For counters, this can easily be achieved by the user with a local variable to track the count and then calling increment with the amount when complete or at the desired batch interval. However, for timers and distribution summaries that doesn't work because the implementation needs the raw samples. In Netflix#941 a method was added so the samples could be accumulated in an array and posted in a batch. This helps a bit, but it is cumbersome to use, incurs memory overhead, and the full set of samples have to be traversed and processed when recorded. This change introduces batch updaters that can be customized for a given implementation. So for registries like Atlas where it is possible to accumulate the stats without the memory overhead, it can provide an optimized implementation to do so.
manolama
approved these changes
Nov 11, 2022
| /** Create a new instance. */ | ||
| AtlasMeter(Id id, Clock clock, long ttl) { | ||
| this.id = id; | ||
| this.clock = clock; | ||
| this.ttl = ttl; | ||
| lastUpdated = clock.wallTime(); | ||
| batchUpdaterRefCount = new AtomicInteger(0); |
There was a problem hiding this comment.
Is it worth looking at lazy initialization to avoid the overhead for every meter? Just worried about the high cardinality use case.
There was a problem hiding this comment.
I refactored it to avoid the ref count. It is a bit less self-contained as the swap types need to inspect updater and pass in a supplier so the meter can be resolved on updates in case the registry has expired them.
Swap implementations will check if updater type can take a supplier to resolve the meter to avoid issues with holding on to one that has expired.
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.
Adds a helper that can be used to batch updates to the delegate meter. The main use-case for this is timers or distributions summaries that need to be updated from a single thread with a high throughput.
For counters, this can easily be achieved by the user with a local variable to track the count and then calling increment with the amount when complete or at the desired batch interval. However, for timers and distribution summaries that doesn't work because the implementation needs the raw samples. In #941 a method was added so the samples could be accumulated in an array and posted in a batch. This helps a bit, but it is cumbersome to use, incurs memory overhead, and the full set of samples have to be traversed and processed when recorded.
This change introduces batch updaters that can be customized for a given implementation. So for registries like Atlas where it is possible to accumulate the stats without the memory overhead, it can provide an optimized implementation to do so.