[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict Elliott Smith updated CASSANDRA-15213: --- Since Version: 4.0-alpha (was: 4.0) > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict Elliott Smith updated CASSANDRA-15213: --- Since Version: 4.0 Source Control Link: [adc3cdde2ae2a78f7d2bb66da47a07545d3e06cf|https://github.com/apache/cassandra/commit/adc3cdde2ae2a78f7d2bb66da47a07545d3e06cf] Resolution: Fixed Status: Resolved (was: Ready to Commit) > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict Elliott Smith updated CASSANDRA-15213: --- Fix Version/s: 4.0 > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0, 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan West updated CASSANDRA-15213: Status: Ready to Commit (was: Review In Progress) > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan West updated CASSANDRA-15213: Test and Documentation Plan: Existing microbenchmarks, new cluster benchmarks, quick theories tests Status: Patch Available (was: Open) > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan West updated CASSANDRA-15213: Reviewers: Benedict Elliott Smith, Jordan West (was: Chris Lohfink) Status: Review In Progress (was: Patch Available) > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jordan West updated CASSANDRA-15213: Attachment: 15213-perf-trunk 15213-perf-branch > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict Elliott Smith >Assignee: Jordan West >Priority: Normal > Fix For: 4.0-beta > > Attachments: 15213-perf-branch, 15213-perf-trunk > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-15213: - Fix Version/s: (was: 4.0) 4.0-beta > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict >Priority: Normal > Fix For: 4.0-beta > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian Jira (v8.3.2#803003) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Chris Lohfink updated CASSANDRA-15213: -- Reviewers: Chris Lohfink > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict >Priority: Normal > Fix For: 4.0 > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian JIRA (v7.6.14#76016) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-15213: - Fix Version/s: 4.0 > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict >Priority: Normal > Fix For: 4.0 > > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian JIRA (v7.6.14#76016) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-15213: - Description: * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user schemas, and this will grow significantly under contention and user schemas with many tables. This is because {{LongAdder}} is a very heavy class designed for single contended values. ** This can likely be improved significantly, without significant loss of performance in the contended case, by simply increasing the size of our primitive backing array and providing multiple buckets, with each thread picking a bucket to increment, or simply multiple backing arrays. Probably a better way still to do this would be to introduce some competition detection to the update, much like {{LongAdder}} utilises, that increases the number of backing arrays under competition. ** To save memory this approach could partition the space into chunks that are likely to be updated together, so that we do not need to duplicate the entire array under competition. * Similarly, binary search is costly and a measurable cost as a share of the new networking work (without filtering it was > 10% of the CPU used overall). We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, to save the random memory access costs. > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict >Priority: Normal > > * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user > schemas, and this will grow significantly under contention and user schemas > with many tables. This is because {{LongAdder}} is a very heavy class > designed for single contended values. > ** This can likely be improved significantly, without significant loss of > performance in the contended case, by simply increasing the size of our > primitive backing array and providing multiple buckets, with each thread > picking a bucket to increment, or simply multiple backing arrays. Probably a > better way still to do this would be to introduce some competition detection > to the update, much like {{LongAdder}} utilises, that increases the number of > backing arrays under competition. > ** To save memory this approach could partition the space into chunks that > are likely to be updated together, so that we do not need to duplicate the > entire array under competition. > * Similarly, binary search is costly and a measurable cost as a share of the > new networking work (without filtering it was > 10% of the CPU used overall). > We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, > to save the random memory access costs. -- This message was sent by Atlassian JIRA (v7.6.14#76016) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org
[jira] [Updated] (CASSANDRA-15213) DecayingEstimatedHistogramReservoir Inefficiencies
[ https://issues.apache.org/jira/browse/CASSANDRA-15213?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-15213: - Severity: Low Complexity: Normal Discovered By: Adhoc Test Bug Category: Parent values: Degradation(12984)Level 1 values: Performance Bug/Regression(12997) Component/s: Observability/Metrics Status: Open (was: Triage Needed) * {{LongAdder}} introduced to trunk consumes 9MiB of heap without user schemas, and this will grow significantly under contention and user schemas with many tables. This is because {{LongAdder}} is a very heavy class designed for single contended values. ** This can likely be improved significantly, without significant loss of performance in the contended case, by simply increasing the size of our primitive backing array and providing multiple buckets, with each thread picking a bucket to increment, or simply multiple backing arrays. Probably a better way still to do this would be to introduce some competition detection to the update, much like {{LongAdder}} utilises, that increases the number of backing arrays under competition. ** To save memory this approach could partition the space into chunks that are likely to be updated together, so that we do not need to duplicate the entire array under competition. * Similarly, binary search is costly and a measurable cost as a share of the new networking work (without filtering it was > 10% of the CPU used overall). We can compute an approximation floor(log2 n / log2 1.2) extremely cheaply, to save the random memory access costs. > DecayingEstimatedHistogramReservoir Inefficiencies > -- > > Key: CASSANDRA-15213 > URL: https://issues.apache.org/jira/browse/CASSANDRA-15213 > Project: Cassandra > Issue Type: Bug > Components: Observability/Metrics >Reporter: Benedict >Priority: Normal > -- This message was sent by Atlassian JIRA (v7.6.14#76016) - To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org