This document provides an overview of data consistency in Apache Cassandra. It discusses how Cassandra writes data to commit logs and memtables before flushing to SSTables. It also reviews the CAP theorem and how Cassandra offers tunable consistency levels for both reads and writes. Strategies for choosing consistency levels for writes, such as ANY, ONE, QUORUM, and ALL are presented. The document also covers read repair and hinted handoffs in Cassandra. Examples of CQL queries with different consistency levels are given and information on where to download Cassandra is provided at the end.

1. Cassandra Essentials
Tutorial Series
Understanding
Data Consistency
in Apache
Cassandra

2. Agenda
› Overview of reading/writing data in Cassandra
› Details on how Cassandra writes data
› Review of the CAP theorem
› Tunable data consistency
› Choosing a data consistency strategy for writes
› Choosing a data consistency strategy for reads
› CQL examples of data consistency
› Where to get Cassandra
www.datastax.com

3. Reading and Writing in Cassandra
Cassandra is a peer-to-peer, read/write anywhere
architecture, so any user can connect to any node in
any data center and read/write the data they need,
with all writes being partitioned and replicated for them
automatically throughout the cluster.
www.datastax.com

4. Writes in Cassandra
› Data is first written to a commit log for durability
› Then written to a memtable in memory
› Once the memtable becomes full, it is flushed to an SSTable
(sorted strings table)
› Writes are atomic at the row level; all columns are written or
updated, or none are. RDBMS-styled transactions are not
supported
INSERT INTO…
Commit log memtable
SSTable
Cassandra is known for being the fastest database in the industry
where write operations are concerned.
www.datastax.com

5. Writes in Cassandra vs. Other Databases
Cassandra is up to:
4x better in writes!
2x better in reads!
12x better in reads/updates!
Sept, 2011: http://blog.cubrid.org/dev-platform/nosql-benchmarking/
www.datastax.com

6. Review of the CAP Theorem
www.datastax.com

7. Tunable Data Consistency
› Choose between strong and eventual
consistency (All to any node responding)
depending on the need
› Can be done on a per-operation basis, and for
both reads and writes
› Handles Multi-data center operations
1
6 2
Writes Reads
› Any › One 5 3
› One › Quorum
› Quorum › Local_Quorum
4
› Local_Quorum › Each_Quorum
› Each_Quorum › All
› All
www.datastax.com

8. Selecting a Strategy for Writes
› Any – a write must succeed on any available node
› One – a write must succeed on any node responsible for
that row (either primary or replica)
› Quorum – a write must succeed on a quorum of replica
nodes (determined by (replication_factor /2 )+ 1
› Local_Quorum - a write must succeed on a quorum of
replica nodes in the same data center as the coordinator
node
› Each_Quorum - a write must succeed on a quorum of
replica nodes in all data centers
› All – a write must succeed on all replica nodes for a row key
www.datastax.com

9. Hinted Handoffs
› Cassandra attempts to write a row to all replicas for that
row
› If all replica nodes are not available, a hint is stored on one
node to update any downed nodes with the row once they
are available again
› If no replica nodes are available for a row, the use of the
ANY consistency level will instruct the coordinator node to
store a hint and the row data, which it passes to the replica
nodes when they are available Replica 1
Replica3 Replica2
Hint for
Node5
www.datastax.com

10. Selecting a Strategy for Reads
› One – reads from the closest node holding the data
› Quorum – returns a result from a quorum of servers with the
most recent timestamp for the data
› Local_Quorum - returns a result from a quorum of servers
with the most recent timestamp for the data in the same
data center as the coordinator node
› Each_Quorum - returns a result from a quorum of servers with
the most recent timestamp in all data centers
› All – returns a result from all replica nodes for a row key
www.datastax.com

11. Read Repair
› Cassandra ensures that frequently-read data remains
consistent
› When a read is done, the coordinator node compares the
data from all the remaining replicas that own the row in the
background, and if they are inconsistent, issues writes to the
out-of-date replicas to update the row to reflect the most
recently written values.
› Read repair can be configured per column family and is
enabled by default. Replica 1
st
repair
reque
Replica3 Replica2
www.datastax.com

12. CQL Examples
SELECT total_purchases FROM SALES
USING CONSISTENCY QUORUM
WHERE customer_id = 5
UPDATE SALES
USING CONSISTENCY ONE
SET total_purchases = 500000
WHERE customer_id = 4
www.datastax.com

13. Where to get Cassandra?
› Go to www.datastax.com
› DataStax makes free smart start installers
available for Cassandra that include:
› The most up-to-date Cassandra version that is
production quality
› A version of DataStax OpsCenter, which is a visual,
browser-based management tool for managing
and monitoring Cassandra
› Drivers and connectors for popular development
languages
› Same database and application
› Automatic configuration assistance for ensuring
optimal performance and setup for either stand-
alone or cluster implementations
› Getting Started Guide
www.datastax.com

14. Where Can I Learn More?
www.datastax.com
› Free Online Documentation
› Technical White Papers
› Technical Articles
› Tutorials
› User Forums
› User/Customer Case Studies
› FAQ’s
› Videos
› Blogs
› Software downloads
www.datastax.com

15. Cassandra Essentials
Tutorial Series
Understanding
Data Partitioning and
Replication in Apache
Cassandra
Thanks!