store: use loser trees #7304
Merged
store: use loser trees #7304
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
yeya24
reviewed
Apr 26, 2024
yeya24
reviewed
Apr 26, 2024
Because the main version works on: In other words, any type that supports |
GiedriusS
force-pushed
the
switch_to_losertree
branch
4 times, most recently
from
bd65217
to
62faefb
Compare
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the heap is used in a lot of hot paths. Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot. Bench: ``` goos: linux goarch: amd64 pkg: github.com/thanos-io/thanos/pkg/store cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz │ oldkway │ newkway │ │ sec/op │ sec/op vs base │ KWayMerge-16 2.292m ± 3% 2.075m ± 15% -9.47% (p=0.023 n=10) │ oldkway │ newkway │ │ B/op │ B/op vs base │ KWayMerge-16 1.553Mi ± 0% 1.585Mi ± 0% +2.04% (p=0.000 n=10) │ oldkway │ newkway │ │ allocs/op │ allocs/op vs base │ KWayMerge-16 27.26k ± 0% 26.27k ± 0% -3.66% (p=0.000 n=10) ``` Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@vinted.com>
GiedriusS
force-pushed
the
switch_to_losertree
branch
from
62faefb
to
9790323
Compare
yeya24
reviewed
Apr 26, 2024
| @@ -814,15 +754,15 @@ func (l *eagerRespSet) At() *storepb.SeriesResponse { | |||
| return nil | |||
| } | |||
|
|
|||
| return l.bufferedResponses[l.i] | |||
| return l.bufferedResponses[l.i-1] | |||
There was a problem hiding this comment.
Is it because we call Next before At so we need to change to i-1?
There was a problem hiding this comment.
Yeah, our dedup heap tree was a bit buggy - we were calling At() before Next() was ever called. The loser tree properly calls Next first hence this change
yeya24
approved these changes
Apr 26, 2024
Nashluffy
pushed a commit
to Nashluffy/thanos
that referenced
this pull request
May 14, 2024
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the heap is used in a lot of hot paths. Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot. Bench: ``` goos: linux goarch: amd64 pkg: github.com/thanos-io/thanos/pkg/store cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz │ oldkway │ newkway │ │ sec/op │ sec/op vs base │ KWayMerge-16 2.292m ± 3% 2.075m ± 15% -9.47% (p=0.023 n=10) │ oldkway │ newkway │ │ B/op │ B/op vs base │ KWayMerge-16 1.553Mi ± 0% 1.585Mi ± 0% +2.04% (p=0.000 n=10) │ oldkway │ newkway │ │ allocs/op │ allocs/op vs base │ KWayMerge-16 27.26k ± 0% 26.27k ± 0% -3.66% (p=0.000 n=10) ``` Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@vinted.com> Signed-off-by: mluffman <nashluffman@gmail.com>
GiedriusS
added a commit
to vinted/thanos
that referenced
this pull request
May 31, 2024
jnyi
pushed a commit
to jnyi/thanos
that referenced
this pull request
Jun 1, 2024
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the heap is used in a lot of hot paths. Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot. Bench: ``` goos: linux goarch: amd64 pkg: github.com/thanos-io/thanos/pkg/store cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz │ oldkway │ newkway │ │ sec/op │ sec/op vs base │ KWayMerge-16 2.292m ± 3% 2.075m ± 15% -9.47% (p=0.023 n=10) │ oldkway │ newkway │ │ B/op │ B/op vs base │ KWayMerge-16 1.553Mi ± 0% 1.585Mi ± 0% +2.04% (p=0.000 n=10) │ oldkway │ newkway │ │ allocs/op │ allocs/op vs base │ KWayMerge-16 27.26k ± 0% 26.27k ± 0% -3.66% (p=0.000 n=10) ``` Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@vinted.com>
jnyi
pushed a commit
to jnyi/thanos
that referenced
this pull request
Jun 4, 2024
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the heap is used in a lot of hot paths. Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot. Bench: ``` goos: linux goarch: amd64 pkg: github.com/thanos-io/thanos/pkg/store cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz │ oldkway │ newkway │ │ sec/op │ sec/op vs base │ KWayMerge-16 2.292m ± 3% 2.075m ± 15% -9.47% (p=0.023 n=10) │ oldkway │ newkway │ │ B/op │ B/op vs base │ KWayMerge-16 1.553Mi ± 0% 1.585Mi ± 0% +2.04% (p=0.000 n=10) │ oldkway │ newkway │ │ allocs/op │ allocs/op vs base │ KWayMerge-16 27.26k ± 0% 26.27k ± 0% -3.66% (p=0.000 n=10) ``` Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@vinted.com>
jnyi
pushed a commit
to jnyi/thanos
that referenced
this pull request
Oct 17, 2024
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the heap is used in a lot of hot paths. Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot. Bench: ``` goos: linux goarch: amd64 pkg: github.com/thanos-io/thanos/pkg/store cpu: Intel(R) Core(TM) i9-10885H CPU @ 2.40GHz │ oldkway │ newkway │ │ sec/op │ sec/op vs base │ KWayMerge-16 2.292m ± 3% 2.075m ± 15% -9.47% (p=0.023 n=10) │ oldkway │ newkway │ │ B/op │ B/op vs base │ KWayMerge-16 1.553Mi ± 0% 1.585Mi ± 0% +2.04% (p=0.000 n=10) │ oldkway │ newkway │ │ allocs/op │ allocs/op vs base │ KWayMerge-16 27.26k ± 0% 26.27k ± 0% -3.66% (p=0.000 n=10) ``` Signed-off-by: Giedrius Statkevičius <giedrius.statkevicius@vinted.com>
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.
Remove a long-standing TODO item in the code - let's use the great loser tree implementation by Bryan. It is faster than the heap because less comparisons are needed. Should be a nice improvement given that the merging is used in a lot of hot paths.
Since Prometheus also uses this library, it's tricky to import the "any" version. I tried doing bboreham/go-loser#3 but it's still impossible to do that. Let's just copy/paste the code, it's not a lot.
Bench: