[ https://issues.apache.org/jira/browse/TINKERPOP-1254?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15369889#comment-15369889 ]
ASF GitHub Bot commented on TINKERPOP-1254: ------------------------------------------- Github user okram commented on the issue: https://github.com/apache/tinkerpop/pull/358 Verified the following simply traversal on Grateful Dead graph. For OLTP (`MatchStep.standardAlgorithm()`), I added a lazy barrier internal to `MatchStep` which is similar in behavior to the lazy barrier OLTP optimization added in `PrunePathStrategy`. ``` source.V().match( as("a").out().as("b"), as("b").out().as("c")). select("c").count().profile() ``` *With and then without PrunePathStrategy OLAP* ``` Traversal Metrics Step Count Traversers Time (ms) % Dur ============================================================================================================= GraphStep(vertex,[]) 808 808 40.418 43.47 MatchStep(AND,[[MatchStartStep(a), ProfileStep,... 327370 452 28.015 30.13 MatchStartStep(a) 808 808 61.751 VertexStep(OUT,vertex) 8049 7957 37.971 MatchEndStep(b) (profiling ignored) 0.000 MatchStartStep(b) 8049 562 14.453 VertexStep(OUT,vertex) 327370 7510 16.124 MatchEndStep(c) (profiling ignored) 0.000 SelectOneStep(c) 327370 452 24.531 26.39 CountGlobalStep 1 1 0.004 0.00 >TOTAL - - 92.969 - Traversal Metrics Step Count Traversers Time (ms) % Dur ============================================================================================================= GraphStep(vertex,[]) 808 808 6.844 0.35 MatchStep(AND,[[MatchStartStep(a), ProfileStep,... 327370 326983 1257.050 64.22 MatchStartStep(a) 808 808 11.806 VertexStep(OUT,vertex) 8049 7957 30.286 MatchEndStep(b) (profiling ignored) 0.000 MatchStartStep(b) 8049 7957 14.627 VertexStep(OUT,vertex) 327370 326983 774.934 MatchEndStep(c) (profiling ignored) 0.000 SelectOneStep(c) 327370 326983 693.564 35.43 CountGlobalStep 1 1 0.001 0.00 >TOTAL - - 1957.460 - ``` *With and then without PrunePathStrategy OLTP* ``` Traversal Metrics Step Count Traversers Time (ms) % Dur ============================================================================================================= TinkerGraphStep(vertex,[]) 808 808 2.050 0.07 MatchStep(AND,[[MatchStartStep(a), ProfileStep,... 327370 56814 2794.288 91.47 MatchStartStep(a) 808 808 103.524 VertexStep(OUT,vertex) 8049 8049 107.345 MatchEndStep(b) 8049 8049 159.505 MatchStartStep(b) 8049 7957 105.127 VertexStep(OUT,vertex) 327370 327370 351.935 MatchEndStep(c) 327370 327370 1331.478 NoOpBarrierStep(1000) 327370 452 254.309 8.32 SelectOneStep(c) 327370 452 3.731 0.12 CountGlobalStep 1 1 0.465 0.02 >TOTAL - - 3054.845 - Traversal Metrics Step Count Traversers Time (ms) % Dur ============================================================================================================= TinkerGraphStep(vertex,[]) 808 808 2023.710 36.87 MatchStep(AND,[[MatchStartStep(a), ProfileStep,... 327370 326983 2995.219 54.57 MatchStartStep(a) 808 808 221.105 VertexStep(OUT,vertex) 8049 8049 185.946 MatchEndStep(b) 8049 8049 186.814 MatchStartStep(b) 8049 7957 161.012 VertexStep(OUT,vertex) 327370 327370 405.787 MatchEndStep(c) 327370 327370 423.213 SelectOneStep(c) 327370 326983 324.356 5.91 CountGlobalStep 1 1 145.952 2.66 >TOTAL - - 5489.238 - ``` > Support dropping traverser path information when it is no longer needed. > ------------------------------------------------------------------------ > > Key: TINKERPOP-1254 > URL: https://issues.apache.org/jira/browse/TINKERPOP-1254 > Project: TinkerPop > Issue Type: Improvement > Components: process > Affects Versions: 3.1.1-incubating > Reporter: Marko A. Rodriguez > Assignee: Ted Wilmes > > The most expensive traversals (especially in OLAP) are those that can not be > "bulked." There are various reasons why two traversers at the same object can > not be bulked, but the primary reason is {{PATH}} or {{LABELED_PATH}}. That > is, when the history of the traverser is required, the probability of two > traversers having the same history is low. > A key to making traversals more efficient is to do as a much as possible to > remove historic information from a traverser so it can get bulked. How does > one do this? > {code} > g.V.as('a').out().as('b').out().where(neq('a').and().neq('b')).both().name > {code} > The {{LABELED_PATH}} of "a" and "b" are required up to the {{where()}} and at > which point, at {{both()}}, they are no longer required. It would be smart to > support: > {code} > traverser.dropLabels(Set<String>) > traverser.dropPath() > {code} > We would then, via a {{TraversalOptimizationStrategy}} insert a step between > {{where()}} and {{both()}} called {{PathPruneStep}} which would be a > {{SideEffectStep}}. The strategy would know which labels were no longer > needed (via forward lookahead) and then do: > {code} > public class PathPruneStep { > final Set<String> dropLabels = ... > final boolean dropPath = ... > public void sideEffect(final Traverser<S> traverser) { > final Traverser<S> start = this.starts.next(); > if(this.dropPath) start.dropPath(); > else start.dropLabels(labels); > } > } > {code} > Again, the more we can prune historic path data no longer needed, the higher > the probability of bulking. Think about this in terms of {{match()}}. > {code} > g.V().match( > a.out.b, > b.out.c, > c.neq.a, > c.out.b, > ).select("a") > {code} > All we need is "a" at the end. Thus, once a pattern has been passed and no > future patterns require that label, drop it! > This idea is related to TINKERPOP-331, but I don't think we should deal with > manipulating the species. Thus, I think 331 is too "low level." -- This message was sent by Atlassian JIRA (v6.3.4#6332)