Uploaded image for project: 'Ignite'
  1. Ignite
  2. IGNITE-8117

Slow Ignite Cache with multi-nodes

    XMLWordPrintableJSON

Details

    • Improvement
    • Status: Open
    • Major
    • Resolution: Unresolved
    • None
    • 2.4
    • cache, jdbc
    • None
    • OS: Ubuntn 14.04.3 LTS

      JAVA: JDK 1.7

      Ignite: 2.4

       

       

       

    Description

      Dear sir,

      I am running Ignite in my project about  with three nodes:

      Three nodes: "ubuntu7","ubuntu8","ubuntu9"

      cacheConf.setIndexedTypes(String.class, String.class)

      cacheConf.setCacheMode(CacheMode.PARTITIONED);

      cacheConf.setAtomicityMode(CacheAtomicityMode.ATOMIC);

      When using above setting for

      • Put,
      • putAll,
      • SqlFieldsQuery,
      • IgniteJdbcThinDriver,
      • IgniteDataStreamer add
      • IgniteDataStreamer addAll

      , there are a situation about data latency, and the experiment results are in the following.

      One node(ubuntu7)

      milliseconds 1,000 10,000 100,000 500,000 1,000,000
      Put 401 1,108 3,874 12,636 23,561
      putAll 146 428 2,092 8,940 17,854
      SqlFieldsQuery 282 943 4,522 20,821 41,223
      IgniteJdbcThinDriver 491 1,581 8,938 41,429 80,599
      IgniteDataStreamer
      Add
      152 439 2,161 7,374 12,296
      IgniteDataStreamer
      addAll
      101 234 2,028 5,037 9,181

       

      Two nodes(ubuntu7 write and ubuntu8)

      milliseconds 1,000 10,000 100,000 500,000 1,000,000
      Put 1,185 5,481 26,142 - -
      putAll 107 292 1,785 - -
      SqlFieldsQuery 786 3,195 30,528 - -
      IgniteJdbcThinDriver 859 3,569 34,080 - -
      IgniteDataStreamer
      add
      169 410 1389 - -
      IgniteDataStreamer
      addAll
      99 206 988 - -

       

      Three nodes(ubuntu7 write, ubuntu8, and ubuntu9)

      milliseconds 1,000 10,000 100,000 500,000 1,000,000
      Put 1,664 7,794 41,116 - -
      putAll 101 294 1,911 - -
      SqlFieldsQuery 1,086 5,718 41,997 - -
      IgniteJdbcThinDriver 1,247 4,899 48,566 - -
      IgniteDataStreamer
      add
      168 385 1,364 - -
      IgniteDataStreamer
      addAll
      86 249 905 - -

      From above results, these tables show that the three ways: put, SqlFieldsQuery, IgniteJdbcThinDriver have gradually bad performance on data latency  with increasing ignite server nodes, and the other ways: putAll, IgniteDataStreameradd, and IgniteDataStreameraddAll have better performances.

      I don't know if the testing via my java codes are suitable and correct to show these results or not.

      Here, i provide my codes to reproduce my experimental situation.

      Attached is my codes for your reference.

      if any further information is needed, I am glad to be informed and will provide to you as soon as possible.

      If you have any idea about this issue, I am looking forward to hearing from you.

      Yours sincerely,

      Rick

       

      Attachments

        1. WriteKV.java
          9 kB
          Rick Lin

        Activity

          People

            Unassigned Unassigned
            Ricklin Rick Lin
            Votes:
            0 Vote for this issue
            Watchers:
            1 Start watching this issue

            Dates

              Created:
              Updated: