(#6031) - faster IDB changes() with batched cursor #6033
Merged
Conversation
This file contains hidden or 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
nolanlawson
commented
Dec 20, 2016
| }); | ||
| }); | ||
| }); | ||
|
|
Member
Author
There was a problem hiding this comment.
Added a new test to play it safe. This particular test would have failed in my previous implementation in #6031.
nolanlawson
commented
Dec 20, 2016
| runBatchedCursor(bySeqStore, opts.since, opts.descending, limit, onBatch); | ||
| } | ||
|
|
||
| export default changes; No newline at end of file |
Member
Author
There was a problem hiding this comment.
I was uncomfortable with having so much changes logic in index.js so I moved it to its own file.
Contributor
|
That is a huge speedup, awesome job 👍 |
daleharvey
approved these changes
Dec 21, 2016
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
This is a rewrite of #6031. Instead of having two separate implementations of
changes(), I've implemented a newrunBatchedCursor()routine that abstracts away two implementations:getAll()/getAllKeys()-based, for browsers that support itIDBCursor-based, for browsers that don't supportgetAll()/getAllKeys()or for queries where we need cursors (e.g. descending)Mode 1 essentially works by fetching
nchanges for every batch, wherencorresponds tolimitin thechanges()API. In the case ofauto_compacted databases with no conflicts, and when we're not usingopts.filteroropts.doc_ids, this means that we can accomplishchanges()with a single batch (notably this applies to secondary indexes, giving them quite a perf boost).In other cases, we may slightly overfetch, but I'd argue this is still better than fetching one-at-a-time, and that we're still honoring the user's preferences for memory consumption (communicated via
limit). In cases wherelimitis -1/undefined we do not use Mode 1 since it would mean reading the whole DB into memory.Mode 2 works like a classic IDBCursor, as we've done in the past. It's abstracted into Mode 1 by acting like a batched cursor where the batch size is 1.
I ran a benchmark using Firefox 50 and Chrome 55 on a MacBook Air, and got the following results for
temp-viewswith 10 iterations:So roughly, this makes secondary index creation in IDB about 2x faster in Chrome and Firefox. It also has the potential to make replication faster (when pulling from IDB databases) but I need to write a new benchmark to confirm it.