SLQ vs NOSQL - friends or foes
- 1. SQL vs NOSQL friends or foes Pedro Gomes & Miguel Araújo 8 February - Braga Geek Nights
- 2. SQL “select maturity, experience, consistency from real_world where age > 40;” • Relational Model born in 1970, defined by Codd • Data model with simple interface for the common user • Efficient storage of data • Dozens of mature solutions (most popular commercial and open source databases currently in use are based on the relational model) • Standard/Unified model
- 3. NOSQL “select fun, profit from real_world where relational=false;” • Movement constituted by several types of database engines that break a 40 year monopoly • With origins that can be tracked into the 60s, this movement gained visibility after 2000. • Defined by a controversial term this movement gained traction thanks to the web giants Google, Amazon, Yahoo and others • Between closed and open source solutions, we can such names as Google BigTable, Facebook Cassandra or LinkedIn Voldemort
- 4. SQL Pros • ACID transactions for OLTP • Maturity and stability • Years of expertise and developing • SQL language • Ad-hoc queries. If you need to answer real-time questions about your data that you can’t predict in advance • Data integrity. NOSQL systems rely on applications to enforce data integrity where SQL uses a declarative approach
- 5. SQL Cons • Expensive and/or unnatural scalability • Used as a solution to all problems it cans lead to low performance (10x slower) • Static schemes not natural in a more and more dynamical world • The strict consistency can be unnatural under failure, on high latency communications (e.g. between Data centers) and on disconnected devices.
- 6. NOSQL Pros • Dynamic schemas • Adaptable consistency models • Architectures built for scale and availability • Friendly web APIs • Easier maintainability and administration
- 7. NOSQL Cons • Beta, alpha, 0.X version... • No SQL • User coded consistence, and simple operations can sometimes be verbose • They sometimes lack monitoring and integration tools • Difficult to master
- 8. SQL Used where • Stack Overflow • Google • Facebook • Twitter • Amazon • eBay
- 9. NOSQL Used where • Facebook • Twitter • Google • Amazon • Mozilla • Netflix
- 10. Forget SQL vs NOSQL!
- 11. What should I use? For your personal website or small project ‣ Doesn’t matter! Use MongoDB or other document store if you don’t like SQL Use MySQL or PostgreSQL if you are more familiar with such RDBMS Try something new
- 12. What should I use? For electronic commerce or another OLTP scenarios ‣ Use a classic RDBMS of your choice: When you need to scale use sharding, replication or memcache Other valid options offer ACID transactions In memory VoltDB Non relational options: Berkeley DB, Infinispan Object oriented: ZODB, Db40
- 13. What should I use? For data mining and data warehousing ‣ Use a column oriented database It can lead to a performance gain of 10X Zynga currently uses Vertica Other options include MonetDB and LucidDB
- 14. What should I use? For large scale data with a strong consistency model ‣ Use HBase The main contributors to the project include Yahoo, Facebook, Stubleupon and Cloudera Great for HDFS/Hadoop integration
- 15. What should I use? For large scale real time analytics ‣ Use a Dynamo style database With a Key/Value model: Riak With a BigTable model: Cassandra
- 16. Different problems... Different solutions - Wide your horizons Graph databases FlockBD FlockDB Hacker’s database: Redis
- 17. Questions? •
- 18. References • http://highscalability.com/blog/2010/12/6/what-the-heck-are-you-actually-using-nosql-for.html • http://databasecolumn.vertica.com/database-innovation/mapreduce-a-major-step-backwards/ • http://blog.mozilla.com/data/2010/05/18/riak-and-cassandra-and-hbase-oh-my/ • http://shouldibeworriedaboutscaling.info/ • https://docs.google.com/present/view?id=0Acbfb57m7svVZGRicDd2ZjJfMTFocG10andmeA&hl=en • http://lwn.net/Articles/376626/ • http://techblog.netflix.com/2011/01/nosql-at-netflix.html • http://nosql.mypopescu.com/ • http://nosql-databases.org/ • One Size Fits All? – Part 2: Benchmarking Results (Stonebraker, Çetintemel, et all