perf(countindex): Optimize count index (#4666) #5971
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
Fixes: DGRAPH-890 Currently count index is calculated in the following way: Calculate length of posting list for which mutation is getting performed. Perform the actual mutation. Calculate the length of posting list after mutation is performed. Add count index for new count and delete count index for old count. Finding length of posting list: Length of posting list is calculated by merging mutable and immutable layer. Immutable layer is already sorted. We need to sort mutable layer for merging. This sorting turns out to be a costly function and shows up in CPU profile. Since we are calculating length twice, sorting is happening twice. Optimization: This PR avoids calculating length twice. It uses length calculated before and edge(for mutation) information, to calculate length after mutation. Live loader performance comparison: Master: Number of TXs run : 21240 Number of N-Quads processed : 21239870 Time spent : 14m13.252187311s N-Quads processed per second : 24900 This PR: Number of TXs run : 21240 Number of N-Quads processed : 21239870 Time spent : 12m52.835002981s N-Quads processed per second : 27512 (cherry picked from commit 1167250)
ashish-goswami
requested review from
manishrjain,
martinmr and
a team
as code owners
parasssh
approved these changes
Jul 13, 2020
There was a problem hiding this comment.
Reviewable status: 0 of 3 files reviewed, all discussions resolved (waiting on @manishrjain and @martinmr)
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Outdated suggestions cannot be applied.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.
Fixes: DGRAPH-890
Currently count index is calculated in the following way:
Calculate length of posting list for which mutation is getting performed.
Perform the actual mutation.
Calculate the length of posting list after mutation is performed.
Add count index for new count and delete count index for old count.
Finding length of posting list:
Length of posting list is calculated by merging mutable and immutable layer. Immutable layer is already sorted. We need to sort mutable layer for merging. This sorting turns out to be a costly function and shows up in CPU profile. Since we are calculating length twice, sorting is happening twice.
Optimization:
This PR avoids calculating length twice. It uses length calculated before and edge(for mutation) information, to calculate length after mutation.
Live loader performance comparison:
Master:
Number of TXs run : 21240
Number of N-Quads processed : 21239870
Time spent : 14m13.252187311s
N-Quads processed per second : 24900
This PR:
Number of TXs run : 21240
Number of N-Quads processed : 21239870
Time spent : 12m52.835002981s
N-Quads processed per second : 27512
(cherry picked from commit 1167250)
This change is
Docs Preview: