NoSQL – What’s that.pdf

Transcript

1. Sergejus Barinovas | Microsoft MVP @sergejusb, sergejus.blogas.lt
2. NoSQL
4. • Limited SQL scalability • Horizontal partitioning (sharding) • Vertical partitioning NoSQL – Why?
5. • Limited SQL availability • Master / slave configuration NoSQL – Why?
6. • SQL limitations for storing huge amount of data • Key / value / type columns NoSQL – Why?
7. • Limited SQL speed of read/write operations • Multiple read replicas NoSQL – Why?
8. • 2009, Eric Evans • NoSQL – open source distributed databases, not relational SQL databases • NoSQL – not only SQL • NoSQL → Big Data NoSQL History
9. • The ability to horizontally scale simple-operation throughput over many servers NoSQL Characteristics (scalability)
10. • A “weaker” concurrency model than the ACID transactions in most SQL systems NoSQL Characteristics (BASE)
11. • Efficient use of distributed indexes and RAM for data storage NoSQL Characteristics (distributed)
12. • The ability to dynamically define new attributes or data schema NoSQL Characteristics (schema-less)
13. • Atomicity – all or nothing • Consistency – state integrity • Isolation – no reads of uncommitted data • Durability – recover committed trans ACID (transactions)
14. • 2000, Eric Brewer • It is impossible for a distributed computer system to simultaneously provide all three of the following guarantees: • Consistency • Availability • Partition tolerance CAP Theorem
15. • Basically – partial system failures are OK Available • Soft state – inconsistency is OK • Eventual consistency – stale data is OK BASE (eventual consistency)
18. • Key / value store • Document database • Graph database • Columnar database NoSQL Categories
19. • <key, value> or Tuple<key, v1,. ., vn> • Simple operations • Get • Put • Delete Key / value store Byte[] Byte[] Key Value
20. Key / value store Key Value “current_date” 2013.02.01 “sergejusb” Binary Object “sergejusb” JSON Object
21. • Dynamo* • Membase • Voldermort • Redis • Azure Table Storage • Riak Key / value store
22. Name: Dynamo Created: 2007, Amazon (proprietary) Implementation: ? Distributed: Yes Replication: Multiple Servers CAP: AP API: ? Key / value store
23. Name: Membase Created: 2010, sponsored by Zinga Implementation: C / C++ / Erlang Distributed: Yes Replication: Multiple Servers CAP: CP API: Memcached API, JSON Key / value store
24. Name: Voldemort Created: 2008, LinkedIn Implementation: Java Distributed: Yes Replication: Multiple Servers CAP: AP API: Java Key / value store
25. Name: Redis Created: 2009, sponsored by VMWare Implementation: C Distributed: No Replication: Master / Slave CAP: CP API: Various Languages Key / value store
26. Name: Azure Table Storage Created: 2008, Microsoft Implementation: ? Distributed: Yes Replication: Multiple Servers (DFS) CAP: CP API: .NET API, JSON Key / value store
27. Name: Riak Created: 2008, Basho (from Akamai) Implementation: Erlang Distributed: Yes Replication: Multiple Servers CAP: AP API: JSON Key / value store
28. • Document == complex object • XML • YAML • JSON / BSON • Support for secondary indexes • Schema can be defined at runtime • Optional support for simple querying using Map / Reduce Document database
29. • MongoDB • CouchDB • RavenDB Document database
30. Name: MongoDB Created: 2008, 10gen Implementation: C++ Distributed: Yes via Shards Replication: Master / Slave CAP: CP API: BSON Document database
31. Name: CouchDB Created: 2005 Implementation: Erlang Distributed: Sort of Replication: Master / Master CAP: AP API: JSON Document database
32. Name: RavenDB Created: 2010, Ayende Rahien Implementation: C# Distributed: Yes via Shards Replication: Master / Master CAP: AP API: .NET API, JSON Document database
33. • Graph == network • Basic constructs • Node • Edge • Properties Graph database sergejus sergejus.blogas.lt tdagys knows knows
34. • FlockDB • Neo4J Graph database
35. Name: FlockDB Created: 2010, Twitter Implementation: Scala Distributed: Yes Replication: Multiple Servers CAP: AP API: Thrift, Ruby Graph database
36. Name: Neo4J Created: 2003, Neo Technologies Implementation: Java Distributed: No Replication: Master / Slave CAP: CP API: JSON, Various Languages Graph database
37. • For HUGE amount of data • Columns are added at a runtime • Great scalability • Horizontal • Vertical Columnar database
38. • Unusual data model • Key Space == Database • Column Family == Table • Columns and Super Columns • Super Column == array of Columns • Column == Tuple<Key, Value, Timestamp, TTL> Columnar database
39. Columnar database • Simple Column
40. Columnar database • Super Column
41. • BigTable* • Cassandra • HBase • Hypertable Columnar database
42. Name: BigTable Created: 2006, Google Implementation: C++ Distributed: Yes Replication: Multiple Servers (GFS) CAP: CP API: C++ Columnar database
43. Name: Cassandra Created: 2008, Facebook Implementation: Java Distributed: Yes Replication: Multiple Servers CAP: AP API: Thrift, Avro Columnar database
44. Name: HBase Created: 2007, Powerset Implementation: Java Distributed: Yes Replication: Multiple Servers (HDFS) CAP: CP API: Thrift, Java, JSON Columnar database
45. Name: Hypertable Created: 2007, Zvents Implementation: C Distributed: Yes Replication: Multiple Servers CAP: CP API: Thrift Columnar database
46. • ORDER BY ? • “Natural Key Order” NoSQL Limitations
47. • GROUP BY ? • Map / Reduce NoSQL Limitations
48. • JOIN ? • Multiple Map / Reduce NoSQL Limitations
49. • SELECT * ? • Multi-Machine Map / Reduce NoSQL Limitations
50. • Maturity • Tooling • Specificity NoSQL Limitations
51. • Choose the right tool for the task • You can use BOTH SQL vs. NoSQL
52. Q & A