snapshot: delegate reading to arrow/parquet readers #590
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
|
This doesn't seem to quite work because we close the file right after loading the snapshot and parquet reads lazily. Maybe we should keep the file open? The issue is that we might have a bunch of hidden memory that we don't know about until we go to read these parquet row groups. |
asubiotto
force-pushed
the
alfonso-alloc
branch
from
ef32416 to
885a545
Compare
|
|
asubiotto
force-pushed
the
alfonso-alloc
branch
from
885a545 to
7ab6a04
Compare
We were previously copying all snapshot part bytes into a slice that was then
passed to arrow readers. This was an unnecessary indirection and resulted in
extra allocations on startup. Note that parquet bytes are still fully copied
since we cannot read those lazily (underlying file is closed after reading from
snapshot).
Benchmark results on recovery from snapshot only:
```
goos: darwin
goarch: arm64
pkg: github.com/polarsignals/frostdb
│ benchmain │ benchnew │
│ sec/op │ sec/op vs base │
Replay-12 1.681 ± 5% 2.545 ± 4% +51.41% (p=0.002 n=6)
│ benchmain │ benchnew │
│ B/op │ B/op vs base │
Replay-12 2.327Gi ± 0% 2.278Gi ± 0% -2.11% (p=0.002 n=6)
│ benchmain │ benchnew │
│ allocs/op │ allocs/op vs base │
Replay-12 20.81M ± 0% 20.79M ± 0% -0.10% (p=0.002 n=6)
```
asubiotto
force-pushed
the
alfonso-alloc
branch
from
7ab6a04 to
d720639
Compare
|
Closing since after fixing the bug the improvement is not really noticeable/worth it: |
|
Did you forget to close? |
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Outdated suggestions cannot be applied.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.
We were previously copying all snapshot part bytes into a slice that was then passed to arrow readers (parquet copies are left untouched, see #590 (comment)). This was an unnecessary indirection and resulted in extra allocations on startup. Benchmark results on recovery from snapshot only: