4. Ops Friendly
â˘Simple design
â˘no special role, no single point of failure
â˘Lots of exposed metrics via JMX
â˘Nodes and entire datacenters can go down with no
loss of service
â˘DataStax OpsCenter
â˘Visual monitoring tool
â˘REST interface to metric data
â˘Free version
â˘Hands-off services
6. CQL3 examples
CREATE USER bombadil WITH PASSWORD 'goldberry4ever' SUPERUSER;
CREATE KEYSPACE shire WITH
REPLICATION = {'class': 'NetworkTopologyStrategy', 'eu' : 3, 'us-east' : 2};
GRANT ALTER ON KEYSPACE shire TO gandalf;
SELECT * FROM emp WHERE empID IN (130,104) ORDER BY deptID DESC;
INSERT INTO excelsior.clicks (userid, url, date, name)â¨
VALUES (â¨
3715e600-2eb0-11e2-81c1-0800200c9a66,â¨
âhttp://cassandra.apache.org',â¨
â2013-10-09',
âMary')â¨
USING TTL 86400;
UPDATE users
SET email = âcharlie@wonka.comâ
WHERE login = âcbucket64'
IF email = âcbucket@wonka.comâ
8. NetďŹix
â˘50 clusters, 750 nodes
â˘Nearly all data in Cassandra
â˘film metadata
â˘user ratings
â˘recommendations
â˘Interesting use case because:
â˘Sheer size and how much they depend on it
â˘Multi-region (effectively multi-datacenter) within AWS
â˘Highly available (through various AWS outages)
See also: http://planetcassandra.org/blog/post/case-study-netďŹix
9. La Poste
â˘Use case: parcel distribution metadata
â˘From MySQL to Cassandra
â˘Holiday load doubles
â˘4 million parcels/day
â˘Average day for one of 70,000 postmen
â˘Scan parcels
â˘Print parcel list
â˘Deliver parcels
â˘Scans remaining, held up to 15 days (TTL)
See also: http://www.slideshare.net/planetcassandra/c-summit-eu-2013-delivering-christmas-gifts-in-france-since-2012
10. Rackspace
â˘Use case: multi-tenant cloud monitoring services
â˘Common time series use case
â˘raw metric data at varying intervals
â˘raw data expires using TTLs
â˘Supports
â˘Ingestion through modular sources
â˘Rollups
â˘Servicing queries at various resolutions
â˘Currently ingests 120 million metrics/hour
â˘See Blueflood.io for project details
See also: http://www.slideshare.net/gdusbabek/blueďŹood-open-source-metrics-processing-at-cassandraeu-2013
11. Spotify
â˘Use case began with playlist storage
â˘Grew significantly beyond that
â˘Some playlist details
â˘Essentially version control system
â˘More than 1 billion playlists
â˘>40,000 request/second at peak
â˘Off-line mode (both access and changes)
â˘Concurrent changes
See also:
http://www.slideshare.net/planetcassandra/c-summit-eu-2013-playlists-at-spotify-using-cassandra-to-store-version-controlled-objects
12. Questions?
â˘@jeromatron on twitter and #cassandra irc
â˘More real world cases
â˘http://planetcassandra.org/FiveMinuteInterviews
â˘DataStax
â˘Free online training
â˘Free developer tools