Cassandra in Action ApacheCon NA 2013

Size: px

Start display at page:

Download "Cassandra in Action ApacheCon NA 2013"

Clifton Summers
8 years ago
Views:

1 Cassandra in Action ApacheCon NA 2013 Yuki Morishita Software / Apache Cassandra Committer 1

2 2

3 ebay Application/Use Case Social Signals: like/want/own features for ebay product and item pages Hunch taste graph for ebay users and items Many time series use cases Why Cassandra? Multi-datacenter Scalable Write performance Distributed counters Hadoop support ACE 3

4 Time series data 4

5 Multi-Datacenter Support 5

6 Distributed counters 6

7 Hadoop support 7

8 Disney Application/Use Case Meet the data management needs of user facing applications across The Walt Disney Company with a single platform Why Cassandra? DataStax Enterprise can tackle real-time and search functions in the same cluster Scalability 24x7 uptime NDI 8

9 Multitenancy 9

10 Multi-tenancy 10

11 Enterprise search 11

12 Netflix Application/Use Case General purpose backend for large scale highly available cloud based web services supporting Netflix Streaming Why Cassandra? Highly available, highly robust and no schema change downtime Highly scalable, optimized for SSD Much lower cost than previous Oracle and SimpleDB implementations Flexible data model Ability to directly influence/implement OSS feature set Supports local and wide area distributed operations, spanning US and Europe RCE 12

13 Optimized for SSD 13

14 Open source 14

15 Use case patterns High performance Massively Scalable Reliable/Available 15

16 Cassandra Scalability In terms of scalability, there is a clear winner throughout our experiments. Cassandra achieves the highest throughput for the maximum number of nodes in all experiments with a linear increasing throughput from 1 to 12 nodes. 16

17 Classic partitioning with SPOF partition 1 partition 2 partition 3 partition 4 slave slave master router client 17

18 Fully distributed, no SPOF client p3 p1 p6 p1 p1 18

19 Partitioning Primary Key as Partition Key jim carol age: 36 car: camaro gender: M age: 37 car: subaru gender: F johnny suzy age:12 age:10 gender: M gender: F 19

20 PK Hashed Value jim carol johnny suzy 5e a9a f4eb27cea b421309e... MD5* hash operation yields a 128-bit number for keys of any size. 20

21 The Token Ring Node A Node B Node D Node C 21

22 Start End A 0xc x B 0x x C 0x x D 0x xc jim carol johnny suzy 5e a9a f4eb27cea b421309e... 22

23 Start End A 0xc x B 0x x C 0x x D 0x xc jim carol johnny suzy 5e a9a f4eb27cea b421309e... 23

24 Start End A 0xc x B 0x x C 0x x D 0x xc jim carol johnny suzy 5e a9a f4eb27cea b421309e... 24

25 Start End A 0xc x B 0x x C 0x x D 0x xc jim carol johnny suzy 5e a9a f4eb27cea b421309e... 25

26 Start End A 0xc x B 0x x C 0x x D 0x xc jim carol johnny suzy 5e a9a f4eb27cea b421309e... 26

27 Replication Node A Node B Node D Node C carol a9a

28 Replication Factor = 2 Node A Node B Node D Node C carol a9a

29 Replication Factor = 3 Node A Node B Node D Node C carol a9a

30 Tunable Consistency Consistency Level READ ONE, QUORUM, LOCAL_QUORUM, EACH_QUORUM, ALL WRITE ANY, ONE, QUORUM, LOCAL_QUORUM, EACH_QUORUM, ALL 30

31 Virtual Nodes without Vnodes with Vnodes 31

32 Virtual Nodes Cluster of heterogeneous machines 32

33 Virtual Nodes 33

34 CQL - Cassandra Query Language CREATE TABLE users ( id uuid PRIMARY KEY, name text, state text, birth_date int ); INSERT INTO users (id, name, state, birth_date) VALUES ( f-11e2-9e c9a66, john, Texas, 1990); SELECT * FROM users WHERE state= Texas AND birth_date > 1950; 2012 DataStax

35 Strictly realtime focused No joins No subqueries No aggregation functions* or GROUP BY ORDER BY? 35

36 Collections CREATE TABLE users ( id uuid PRIMARY KEY, name text, state text, birth_date int ); CREATE TABLE users_addresses ( user_id uuid REFERENCES users, text ); SELECT * FROM users NATURAL JOIN users_addresses; 2012 DataStax

37 Collections CREATE TABLE users ( id uuid PRIMARY KEY, name text, state text, birth_date int ); X CREATE TABLE users_addresses ( user_id uuid REFERENCES users, text ); SELECT * FROM users NATURAL JOIN users_addresses; 2012 DataStax

38 Collections CREATE TABLE users ( id uuid PRIMARY KEY, name text, state text, birth_date int, _addresses set<text> ); UPDATE users SET _addresses = _addresses + { jbellis@gmail.com, jbellis@datastax.com }; 2012 DataStax

39 Question? Feel free to contact me later if you have one yukim (IRC, twitter) 39

HDB++: HIGH AVAILABILITY WITH. l TANGO Meeting l 20 May 2015 l Reynald Bourtembourg

HDB++: HIGH AVAILABILITY WITH. l TANGO Meeting l 20 May 2015 l Reynald Bourtembourg HDB++: HIGH AVAILABILITY WITH Page 1 OVERVIEW What is Cassandra (C*)? Who is using C*? CQL C* architecture Request Coordination Consistency Monitoring tool HDB++ Page 2 OVERVIEW What is Cassandra (C*)?