This document summarizes a presentation about Cassandra 2.0. It discusses Cassandra's ring architecture and how data is replicated across multiple nodes. It explains how Cassandra provides linear scalability as nodes are added. It also covers tunable consistency levels, client/server communication, the data model using CQL, and new features in the Java driver like asynchronous operations and lightweight transactions. Overall, the presentation provides an overview of Cassandra's architecture and data model.

1. Cassandra 2.0
Michaël Figuière
@mfiguiere

2. Speaker
Michaël Figuière
@mfiguiere
©2012 DataStax
2

3. Ring Architecture
Node
Node
Cassandra
Node
©2012 DataStax
Node
Node
Node
3

4. Ring Architecture
Node
Replica
Node
Replica
Node
©2012 DataStax
Replica
4

5. Linear Scalability
Client Writes/s by Node Count - Replication Factor = 3
©2012 DataStax
5

6. Client / Server Communication
Client
?
Node
Replica
Client
Node
Replica
Client
Client
©2012 DataStax
Node
Replica
6

7. Client / Server Communication
Client
Node
Replica
Client
Node
Replica
Client
Client
Node
Replica
Coordinator node:
Forwards all R/W requests
to corresponding replicas
©2012 DataStax
7

8. Tunable Consistency
Time
A
A
A
3 replicas
©2012 DataStax
8

9. Tunable Consistency
Time
A
A
A
B
A
A
Write ‘B’
Write and wait for
acknowledge from one node
©2012 DataStax
9

10. Tunable Consistency
Time
R +W < N
A
A
A
B
A
A
B
A
A
Read waiting for one node
to answer
©2012 DataStax
Write and wait for
acknowledge from one node
10

11. Tunable Consistency
Time
R +W = N
A
A
A
B
B
A
B
B
A
Read waiting for one node
to answer
©2012 DataStax
Write and wait for
acknowledges from two nodes
11

12. Tunable Consistency
Time
R +W > N
A
A
A
B
B
A
B
B
A
Read waiting for two nodes
to answer
©2012 DataStax
Write and wait for
acknowledges from two nodes
12

13. Tunable Consistency
R = W = QUORUM
A
A
A
B
B
A
B
B
Time
A
QUORUM = (N / 2) + 1
©2012 DataStax
13

14. Request Path
1
Client
Client
Node
2
Replica
3
4
2
Node
3
Client
Replica
2
Client
Node
3
Replica
Coordinator node
©2012 DataStax
14

15. Multi-Datacenter
Datacenter A
Node 1
Node
1
Node 2
Node
2
Datacenter B
Node 3
Node
6
Node
3
Node 4
Datacenter C
Node
5
Node
4
Node 5
Node 6
©2012 DataStax
15

16. Storage Engine
Memtable
Commit
Log
©2012 DataStax
Memtable
SSTable
SSTable
SSTable
SSTable
16

17. CQL Denormalized Model
Cassandra comes with a Data Model abstraction made of denormalized,
statically defined tables
Data duplicated over several tables
©2012 DataStax
17

18. Cassandra Data Model
Timeline Table
user_id
tweet_id
author
gmason
1735
phenry
Give me liberty or give me death
gmason
1742
gwashington
I chopped down the cherry tree
ahamilton
1767
jadams
A government of laws, not men
ahamilton
1794
gwashington
I chopped down the cherry tree
Partition
Key
Clustering
Key
©2012 DataStax
body
18

19. Cassandra Data Model
Timeline Table
user_id
tweet_id
author
body
gmason
1735
phenry
Give me liberty or give me death
gmason
1742
gwashington
I chopped down the cherry tree
ahamilton
1767
jadams
A government of laws, not men
ahamilton
1794
gwashington
I chopped down the cherry tree
CQL
CREATE TABLE timeline (
user_id varchar,
tweet_id timeuuid,
author varchar,
body varchar,
PRIMARY KEY (user_id, tweet_id));
©2012 DataStax
19

20. Cassandra Data Model
Timeline Table
user_id
tweet_id
author
body
gmason
1735
phenry
Give me liberty or give me death
gmason
1742
gwashington
I chopped down the cherry tree
ahamilton
1767
jadams
A government of laws, not men
ahamilton
1794
gwashington
I chopped down the cherry tree
Timeline Physical Layout
gmason
ahamilton
©2012 DataStax
[1735, author]
[1735, body]
[1742, author]
[1742, body]
gwashington
I chopped down the...
phenry
Give me liberty or give...
[1767, author]
[1767, body]
[1794, author]
[1794, body]
gwashington
I chopped down the...
jadams
A government of laws...
20

21. Example: Video DB Application
CREATE TABLE users (
username varchar,
firstname varchar,
lastname varchar,
email list<varchar>,
password varchar,
created_date timestamp,
PRIMARY KEY (username)
);
©2012 DataStax
21

22. Example: Video DB Application
CREATE TABLE videos (
videoid uuid,
videoname varchar,
username varchar,
description varchar,
location map<varchar,varchar>,
tags set<varchar>,
upload_date timestamp,
PRIMARY KEY (videoid)
);
©2012 DataStax
22

23. Example: Video DB Application
CREATE TABLE comments_by_video (
videoid uuid,
username varchar,
comment_ts timeuuid,
comment varchar,
PRIMARY KEY (videoid,comment_ts,username)
) WITH CLUSTERING ORDER BY (comment_ts DESC,
username ASC);
©2012 DataStax
23

24. Example: Video DB Application
CREATE TABLE comments_by_user (
username varchar,
videoid uuid,
comment_ts timeuuid,
comment varchar,
PRIMARY KEY (username,comment_ts,videoid)
) WITH CLUSTERING ORDER BY (comment_ts DESC,
videoid ASC);
©2012 DataStax
24

25. Example: Video DB Application
CREATE TABLE video_rating (
videoid uuid,
rating_counter counter,
rating_total counter,
PRIMARY KEY (videoid)
);
©2012 DataStax
25

26. New Driver Architecture
DB API
CQL API
OO API
CQL Native Protocol
Next Generation Driver
©2012 DataStax
26

27. DataStax Java Driver
• Reference Implementation
• Asynchronous architecture based on Netty
• Prepared Statements Support
• Automatic Fail-over
• Node Discovery
• Cassandra Tracing Support
• Tunable policies
•
•
•
©2012 DataStax
LoadBalancingPolicy
ReconnectionPolicy
RetryPolicy
27

28. Connect and Write
Cluster cluster = new Cluster.builder()
.addContactPoints("10.0.0.1", "10.0.0.2")
.build();
Session session = cluster.connect("myKeyspace");
session.execute(
"INSERT INTO user (user_id, name, email)
VALUES (12345, 'johndoe', 'john@doe.com')"
);
©2012 DataStax
28

29. Read
ResultSet rs = session.execute("SELECT * FROM test");
List<Row> rows = rs.all();
for (Row row : rows) {
String userId = row.getString("user_id");
String name = row.getString("name");
String email = row.getString("email");
}
©2012 DataStax
29

30. Asynchronous Read
ResultSetFuture future =
session.executeAsync("SELECT * FROM test");
for (Row row : future.get()) {
String userId = row.getString("user_id");
String name = row.getString("name");
String email = row.getString("email");
}
©2012 DataStax
30

31. Read with Callback
final ResultSetFuture future =
session.executeAsync("SELECT * FROM test");
future.addListener(new Runnable() {
public run() {
for (Row row : future.get()) {
String userId = row.getString("user_id");
String name = row.getString("name");
String email = row.getString("email");
}
}
}, executor);
©2012 DataStax
31

32. Lightweight Transactions
Session 1
SELECT * FROM users
WHERE username = ’jbellis’
[empty resultset]
INSERT INTO users (...)
VALUES (’jbellis’, ...)
©2012 DataStax
Session 2
SELECT * FROM users
WHERE username = ’jbellis’
[empty resultset]
INSERT INTO users (...)
VALUES (’jbellis’, ...)
32

33. Lightweight Transactions
• Based on Paxos
• Fault tolerant
• Quorum based
• 4 round trips vs. 1 for normal updates
• State is durable
©2012 DataStax
33

34. Lightweight Transactions
INSERT INTO USERS (username, email, ...)
VALUES (‘jbellis’, ‘jbellis@datastax.com’, ... )
IF NOT EXISTS;
UPDATE USERS
SET email = ’jonathan@datastax.com’, ...
WHERE username = ’jbellis’
IF email = ’jbellis@datastax.com’;
©2012 DataStax
34

35. Paging with Cassandra 1.2
CREATE TABLE timeline (
user_id uuid,
tweet_id timeuuid,
tweet_author uuid,
tweet_body text,
PRIMARY KEY (user_id, tweet_id)
);
©2012 DataStax
SELECT *
FROM timeline
WHERE (user_id = :last_key
AND tweet_id > :last_tweet)
OR token(user_id) >
token(:last_key)
LIMIT 100
35

36. Automatic Paging
1st Request
Node
Query
Client
Page 1
+ Paging State 1
Node
Replica
Node
©2012 DataStax
Replica
Replica
36

37. Automatic Paging
2nd Request
Node
Replica
Query + Paging State 1
Node
Client
Replica
Page 2 + Paging State 2
Node
©2012 DataStax
Replica
37

38. Stay Tuned!
blog.datastax.com
@mfiguiere