Details

    • Type: Bug
    • Status: Closed
    • Priority: Blocker
    • Resolution: Fixed
    • Affects Version/s: 2.0-beta-2
    • Fix Version/s: 2.0
    • Component/s: ns_server
    • Security Level: Public
    • Labels:
      None

      Description

      ____________________
      From: Ronnie Sun
      Sent: Friday, November 09, 2012 6:34 PM
      To: Steve Yen
      Subject: Re: swap rebalance perf?

      Hi Steve,

      Got some preliminary results showing 2.0 is 10x slower, which falls into the same area we've seen for reb-litmus tests.

      1.8.1-938: 203 seconds
      2.0.0-1939: 2983 seconds

      Spec: (reb-litmus-swap-2) 3M mixed, start with 2 nodes, swap 1 node with 9 client firing 1k ops each.

      Current jenkins automation for this has some problems, so I don't have comparison graphs.

      Thanks,
      Ronnie

        Attachments

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

          Activity

          Hide
          ronnie Ronnie Sun (Inactive) added a comment -

          4-3 swap rebalance:

          without foreground load (1x identical):

          1.8.1-938: 988.58 seconds.
          2.0.0- 1954: 974.67 seconds.

          with ~10k foreground (7x slower):

          1.8.1-938: 222 sec
          2.0.0- 1954: 1723 sec

          6-1 swap rebalance:

          with ~10k foreground (1x identical):
          1.8.1-938: 639.52 sec
          2.0.0- 1954: 654.22 sec

          Show
          ronnie Ronnie Sun (Inactive) added a comment - 4-3 swap rebalance: without foreground load (1x identical): 1.8.1-938: 988.58 seconds. 2.0.0- 1954: 974.67 seconds. with ~10k foreground (7x slower): 1.8.1-938: 222 sec 2.0.0- 1954: 1723 sec 6-1 swap rebalance: with ~10k foreground (1x identical): 1.8.1-938: 639.52 sec 2.0.0- 1954: 654.22 sec
          Hide
          ronnie Ronnie Sun (Inactive) added a comment -

          Based on 6-1 results (started with 6 nodes, swap 1 node):

          2.0 has similar reb time as 1.8.1, while latencies are ~30% worse.

          Show
          ronnie Ronnie Sun (Inactive) added a comment - Based on 6-1 results (started with 6 nodes, swap 1 node): 2.0 has similar reb time as 1.8.1, while latencies are ~30% worse.
          Hide
          steve Steve Yen added a comment -

          https://www.yammer.com/couchbase.com/#/Threads/show?threadId=236527848

          4-3 swap rebalance:

          without foreground load (1x identical):

          1.8.1-938: 988.58 seconds.
          2.0.0- 1954: 974.67 seconds.

          with ~10k foreground (7x slower):

          1.8.1-938: 222 sec
          2.0.0- 1954: 1723 sec

          6-1 swap rebalance:

          with ~10k foreground (1x identical):
          1.8.1-938: 639.52 sec
          2.0.0- 1954: 654.22 sec
          Like · Reply · Share · More · 13 minutes ago

          Ronnie Sun: Based on 6-1 results (started with 6 nodes, swap 1 node):

          2.0 has similar reb time as 1.8.1, while latencies are ~30% worse.

          reb-swap-6-1.loop_1.8.1-938-rel-enterprise_2.0.0-1954-rel-enterprise_orange_Nov-21-2012_10-07-08

          Steve Yen: @Aliaksey Kandratsenka, @Chiyoung Seo

          "4-3 swap rebalance" == 4 nodes initially and swap 3 of them.

          "6-1 swap rebalance" == 6 nodes initially and swap 1 of them.

          @Ronnie Sun also reports latencies were worse in 2.0 during swap rebalance.

          This seems to indicate that we don't need @Chiyoung Seo's patch that prioritizes the vbucket takeover even more.

          This was on system test cluster (xen VM's) with SSD, key-value mixed workload (no views), with consistent views enabled (the @Chiyoung Seo .

          Show
          steve Steve Yen added a comment - https://www.yammer.com/couchbase.com/#/Threads/show?threadId=236527848 4-3 swap rebalance: without foreground load (1x identical): 1.8.1-938: 988.58 seconds. 2.0.0- 1954: 974.67 seconds. with ~10k foreground (7x slower): 1.8.1-938: 222 sec 2.0.0- 1954: 1723 sec 6-1 swap rebalance: with ~10k foreground (1x identical): 1.8.1-938: 639.52 sec 2.0.0- 1954: 654.22 sec Like · Reply · Share · More · 13 minutes ago Ronnie Sun: Based on 6-1 results (started with 6 nodes, swap 1 node): 2.0 has similar reb time as 1.8.1, while latencies are ~30% worse. reb-swap-6-1.loop_1.8.1-938-rel-enterprise_2.0.0-1954-rel-enterprise_orange_Nov-21-2012_10-07-08 Steve Yen: @Aliaksey Kandratsenka, @Chiyoung Seo "4-3 swap rebalance" == 4 nodes initially and swap 3 of them. "6-1 swap rebalance" == 6 nodes initially and swap 1 of them. @Ronnie Sun also reports latencies were worse in 2.0 during swap rebalance. This seems to indicate that we don't need @Chiyoung Seo's patch that prioritizes the vbucket takeover even more. This was on system test cluster (xen VM's) with SSD, key-value mixed workload (no views), with consistent views enabled (the @Chiyoung Seo .
          Hide
          FilipeManana Filipe Manana (Inactive) added a comment -

          No views defined, so updating component list.

          Show
          FilipeManana Filipe Manana (Inactive) added a comment - No views defined, so updating component list.
          Hide
          FilipeManana Filipe Manana (Inactive) added a comment -

          No views defined, so updating component list.

          Show
          FilipeManana Filipe Manana (Inactive) added a comment - No views defined, so updating component list.

            People

            • Assignee:
              ronnie Ronnie Sun (Inactive)
              Reporter:
              ronnie Ronnie Sun (Inactive)
            • Votes:
              0 Vote for this issue
              Watchers:
              0 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved:

                Gerrit Reviews

                There are no open Gerrit changes