Uploaded image for project: 'Java Couchbase JVM Core'
  1. Java Couchbase JVM Core
  2. JVMCBC-674

need feedback on durability performance




      Hey Michael,

      Can you take a look at the performance matrix I compiled for durability. Here is the sheet: https://docs.google.com/spreadsheets/d/1B8v4OZneOeGxJwUj226zA3YDr0Y0gjRSVLwy0IAP9qw/edit?usp=sharing . SDK2 and SDK3 columns are for old durability params (replicate to and persist to) and SDK3 New is the new durability levels. There are two issues that I am confused by and need some input to make sure I did the testing correctly. First: For SDK3 New, all durability levels except durabilityLevel=None have the same performance. To me, it does not make sense why majority and persistMajority would perform the same. Also, the performance impact is severe, dropping from 387k to 1k going from None to majority, >99% drop. Second: SDK3 with replicateTo=1 persistTo=0 performs significantly slower than replicateTo=1 persistTo=1 and replicateTo=1 persistTo=2, which implies that adding persist to increases performance and this doesn't really make sense. 

      Here is my YCSB code I am using for the tests, I create a branch called couchbase3-new-durability based on couchbase3 branch: https://github.com/couchbaselabs/YCSB/blob/couchbase3-new-durability/couchbase3/src/main/java/com/yahoo/ycsb/db/couchbase3/Couchbase3Client.java


      Here is the set of test files I am using: https://github.com/couchbase/perfrunner/tree/master/tests/durability



        Issue Links

          No reviews matched the request. Check your Options in the drop-down menu of this sections header.


            korrigan.clark Korrigan Clark created issue -
            daschl Michael Nitschinger made changes -
            Field Original Value New Value
            Assignee Michael Nitschinger [ daschl ] Korrigan Clark [ korrigan.clark ]
            korrigan.clark Korrigan Clark made changes -
            Link This issue relates to MB-34261 [ MB-34261 ]
            ritam.sharma Ritam Sharma made changes -
            Labels 6.5mustpass
            daschl Michael Nitschinger made changes -
            Status New [ 10003 ] Open [ 1 ]
            daschl Michael Nitschinger made changes -
            Fix Version/s 2.0.0-alpha.5 [ 16203 ]
            korrigan.clark Korrigan Clark made changes -
            Comment [ [~daschl], I ran some tests... Looks like kvendpoints doesnt have the effect we thought it might. The results show that increasing kv endpoints does not increase throughput by a similar amount:

            d=0 [OVERALL], Throughput(ops/sec), 107.38831615120274
            d=1 [OVERALL], Throughput(ops/sec), 87.04659604286175
            d=2 [OVERALL], Throughput(ops/sec), 91.9303535641398
            d=3 [OVERALL], Throughput(ops/sec), 96.49901571003976

            d=0 [OVERALL], Throughput(ops/sec), 118.2941976696043
            d=1 [OVERALL], Throughput(ops/sec), 100.47423840527289
            d=2 [OVERALL], Throughput(ops/sec), 91.69096477233134
            d=3 [OVERALL], Throughput(ops/sec), 87.89199831247363

            d=0 [OVERALL], Throughput(ops/sec), 119.98752129778504
            d=1 [OVERALL], Throughput(ops/sec), 96.25380202518
            d=2 [OVERALL], Throughput(ops/sec), 81.82971236856102
            d=3 [OVERALL], Throughput(ops/sec), 95.53835865099838 ]
            wayne Wayne Siu made changes -
            Link This issue relates to MB-34631 [ MB-34631 ]
            dfinlay Dave Finlay made changes -
            Labels 6.5mustpass 6.5mustpass durability
            daschl Michael Nitschinger made changes -
            Fix Version/s 2.0.0-alpha.6 [ 16233 ]
            Fix Version/s 2.0.0-alpha.5 [ 16203 ]
            korrigan.clark Korrigan Clark made changes -
            Resolution Fixed [ 1 ]
            Status Open [ 1 ] Resolved [ 5 ]
            korrigan.clark Korrigan Clark made changes -
            Actual End 2019-07-08 11:24 (issue has been resolved)
            korrigan.clark Korrigan Clark made changes -
            Status Resolved [ 5 ] Closed [ 6 ]


              korrigan.clark Korrigan Clark
              korrigan.clark Korrigan Clark
              0 Vote for this issue
              7 Start watching this issue



                Gerrit Reviews

                  There are no open Gerrit changes