Key Notes Day 1 – Percona Live 2015

The morning session of key notes started off with some infomation on the conference.  Some of the interesting information included:

  • Over 1200 attendees, 22 from Facebook and 20 from Oracle
  • Percona Live Amsterdam Sept 21-22nd 2015 call for papers open now
  • 205 Speakers
  • 146 Breakout Sessions

Then Peter Zaitsev CEO of Percona came out and gave a great session on the history of MySQL.  With this year being the 20th anniversary of MySQL it was interesting to look back on how it has changed.  Then there was some information about the role that MySQL is playing today.  MySQL is leverage by the top 5 websites today and there is a lot of growth in the future with the move to the cloud and platforms like OpenStack.

After Peter was done then Harrison Fisk from Facebook came up on stage and gave a great overview on how they leverage multiple data platforms to handle different workloads.  Some noted from that presentation are:

  • Different data storage technology to handle varying data storage needs
  • MySQL used for small data also messaging and other systems
  • small data is 10’s of PB
  • get <5ms response times
  • flash storage and flash cache (hybrid flash and disk system built by Facebook)
  • 100’s of millions of QPS
  • HBase is used for small data as well (the messages themselves are in HBase for longer term storage)
  • HBase is used on disk and provides a great platform for archived data
  • HBASE 10’s of PB
  • HBASE Millions of QPS
  • HBASE < 50ms response time
  • RocksDB is an imbedded storage engine developed by Facebook (SQLLite replacement)
  • Used by a lot of internal systems 9B QPS at peak and is open source
  • Used by News feed
  • BigData
    • Hive
      • Minutes/Hours Queries
      • >300 PB logical size
      • >10PB Daily Processing
    • Presto
      • developed by Facebook
      • opensource distributed query engine in memory streaming
      • presto doesn’t handle failures well
      • Seconds/minutes Queries
      • >10K Queries Per Day (big analytic queries)
      • >30 Trilion rows per day
    • Scuba
      • Not open sourced
      • Real time analysis tool used for debugging at Facebook
      • Leaf aggregator in memory columnar store
      • 100s of TB
      • Many M rows ingest
    • New systems
      • RocksDB and MySQL – MySQL with RocksDB storage engine (Rocks is write optimized while InnoDB is read optimized)
      • Presto and MySQL – Sharded MySQL with Presto running on top
        • 100s of Millions of rows at once
        • Updatable all the benefits of MySQL

After Harrison was done then they had a panel discussion on the next disruptive technology after the cloud and big data.  The first question asked was “What’s the next big thing”?  Think about a datacenter as a single element of compute and remove the human element out of the management.  Application needs over technology will lead the way instead of technology leading the applications.  Choosing technology based on needs rather than trying to cram what we have to fulfill a need.  Internet of things is inevitable based on human needs to connect.  Technology is just an abstraction created by humans and it will follow.  Data security is going to be a big growth area.  IoT will drive this need when combined with the cheap costs of storage.  Singularity of information is going to come where the data silos will be broken down.

The question was asked how we would know the net thing is here.  Thomas Hazel from Deep Information Sciences had the answer that I identified with the most when he said “It will change our perspective, then it will change the business, then it will change our lives”.  Other ideas given were:

  • propriety technology will become niche
  • The next big thing will involve data
  • We are in the beginning of a data revolution
  • We can no longer be associated with a platform. Instead we have to be associated with services we provide.
  • The next big opportunity is turning all this data into useful information.
  • The core problem in data is analytics.
  • Data can be used to backup just about anything if we ask the wrong questions.  The big issue is how to ask the correct questions and know that we are doing so.
  • Data is becoming more personal where it used to be corporate.

Last they had a question and answer session with Steve Wozniak one of the founders of Apple.  He shared some great ideas about the direction of technology as well as the meaning of life.  He shared with us some of his concerns surrounding education and what it will take for computers to play a lead role in the education of our youth.  Computers today serve as a powerful tool, but we continue to educate our children very similarly to the way we did in the past.  Steve suggested this is because of a number of factors.  Steve said that the way that computers will take a lead role is to become more cognitive and human like.  People gravitate to humans to learn and until computers are more computer like they will remain as tools.

Steve also shared with us his two formulas for happiness.  His first formula was H=S-F or Happiness = Smiles – Frowns.  Later on he decide that H=F^3.  Happiness is equal to food times fun times friends.  Then Steve wrapped up by sharing information on his relationship with Steve Jobs.

Then just when you thought we were done there was a surprise announcement from Peter that Percona has acquired Tokutek a storage engine for MySQL and MongoDB.

All in all a great morning of information and a great kickoff to the main conference.

One thought on “Key Notes Day 1 – Percona Live 2015

Leave a comment