Successfully reported this slideshow.

Cassandra Performance: Past, present & future

5,466 views

Published on

Talk by Tom Wilkie at Cassandra London meetup

  • Be the first to comment

Cassandra Performance: Past, present & future

  1. 1. CassandraPerformance Past, Present & Future Tom Wilkie, Founder Acunu Ltd
  2. 2. A Busy Year11 Jan : Cassandra 0.7.03 Jun: Cassandra 0.8.018 Oct: Cassandra 1.0.0
  3. 3. 11 Jan : Cassandra 0.7.0• Dynamic endpoint snitch• Row cache improvements• Secondary indexes3 Jun: Cassandra 0.8.018 Oct: Cassandra 1.0.0
  4. 4. 11 Jan : Cassandra 0.7.03 Jun: Cassandra 0.8.0• Counters• CQL18 Oct: Cassandra 1.0.0
  5. 5. 11 Jan : Cassandra 0.7.03 Jun: Cassandra 0.8.018 Oct: Cassandra 1.0.0• Compression• Off heap caches, arena allocations etc• LevelDB-style compactions
  6. 6. Benchmarking
  7. 7. Read the fine print 24GB of memory ... a 300GB disk ...key size of 13 bytes and data size of 1108 bytes.... 96 nodes storing 2.1 billion records ~350k inserts / sec http://www.oracle.com/technetwork/database/nosqldb/learnmore/nosql-database-498041.pdf
  8. 8. “Read/write performance on a single four-core machine. One million inserts followed by one million updates” http://www.datastax.com/dev/blog/whats-new-in-cassandra-1-0-performance
  9. 9. 1 x Quad core Xeon E5620Machine 24GB RAM, 8x1TB Disks 1 commit log, 7 in RAID 10 Keys: 8 byte,Values: 10 bytes Random key distributionWorkload 10 million Inserts, Random Point Gets, Range Queries
  10. 10. Cassandra Performance Evolution 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 0 Inserts Point Gets Range Queries0.7.90.8.71.0.2
  11. 11. 1 x Quad core Xeon E5620Machine 24GB RAM, 8x1TB Disks 1 commit log, 7 in RAID 10 Keys: 8 byte,Values: 100 bytes Random key distributionWorkload 300 million Inserts, Random Point Gets, Range Queries
  12. 12. Inserts/s Time / hours
  13. 13. Effect of Compression and Leveled Compaction 60,000 200 50,000 160 40,000 120 30,000 80 20,000 40 10,000 0 0 Inserts Point Gets1.0.2w/ Compressionw/ Leveled Compaction
  14. 14. Effect of Compression and Leveled Compaction 60,000 500 50,000 400 40,000 300 30,000 200 20,000 100 10,000 0 01.0.2 Inserts Point Gets Range Queriesw/ Compressionw/ Leveled CompactionAcunu
  15. 15. Future
  16. 16. Questions? Tom Wilkie @tom_wilkie tom@acunu.com http://bitbucket.org/acunu http://github.com/acunuhttp://www.acunu.com/download http://www.acunu.com/insights

×