Cassandra and Storm at Health Market Sceince

P. Taylor Goetz
Development Lead, Health Market Science
tgoetz@healthmarketscience.com
@ptgoetz

Agenda
 Cassandra @ HMS
 Storm Overview
 Storm-Cassandra
 Examples / Demo
 Future : Trident

Our Products

 Master Data Management
 Good, bad doctors?
 Prescriber eligibility and remediation.

Cassandra to the Rescue

1000’s of Feeds

C* Masterfile

Big Data for us == Variety of Data

Δt

But…
 Search unstructured data
 Real-time Analytics / Reporting
 Transactional Processing
 Changes reflected immediately.
 Wide-row Indexes

What might that look like?

wide-row index
C* I’m happy

RDBMS

Provide for Polyglot Persistence

What we did wrong…

 Could not react to transactional changes
 Needed extra logic to track what changed
 Took too long

C* at HMS
 Load, Standardize, Match, Consolidate
 Write results to C*

 Track Changes over Time
 Practitioner Data
 Feed Quality

C* at HMS
 Best Practices
 Prefer Write over Read (esp. Read before Write)
 Avoid Queries/Scans
○ Fetch by key whenever possible
○ Put Comparators to work
○ Pre-compute whenever possible

How?
 Treat All Data as Immutable
 Updates are inserts with new version/timestamp

 Data Model
 Heavy use of composites
 Timestamps/Versions in Keys

 Treat Feeds as Real-Time Streams

What Storm is to us…
Crud
Op ETL Dimensional
Enrichment
Counts

Fuzzy
SoR RDBMS
Index

A High Throughput
Data Processing Pipeline

Cassandra and Storm at Health Market Sceince

Storm Overview
 Open-Sourced by Twitter in 2011
 Distributed Realtime Computation System
 Fault Tolerant
 Highly Scalable
 Guaranteed Processing
 Operates on one or more streams of data

Anatomy of a Storm Cluster

 Nimbus
 Master Node
 Zookeeper
 Cluster Coordination
 Supervisors
 Worker Nodes

Storm Primatives
 Streams
 Unbounded sequence of tuples
 Spouts
 Stream Sources
 Bolts
 Unit of Computation
 Topologies
 Combination of n Spouts and n Bolts
 Defines the overall “Computation”

Storm Spouts
 Represents a source (stream) of data
 Queues (JMS, Kafka, Kestrel, etc.)
 Twitter Firehose
 Sensor Data
 Emits “Tuples” (Events) based on source
 Primary Storm data structure
 Set of Key-Value pairs

Storm Bolts
 Receive Tuples from Spouts or other Bolts
 Operate on, or React to Data
 Functions/Filters/Joins/Aggregations
 Database writes/lookups
 Optionally emit additional Tuples

Storm Topologies
 Data flow between spouts and bolts
 Routing of Tuples between spouts/bolts
 Stream “Groupings”
 Parallelism of Components
 Long-Lived

Storm and Cassandra
 Use Cases:
 Write Storm Tuple data to C*
○ Computation Results
○ Pre-computed indices

 Read data from C* and emit Storm Tuples
○ Dynamic Lookups

http://github.com/hmsonline/storm-cassandra

Storm Cassandra Bolt
Types
CassandraBolt

Cassandra
LookupBolt
C*
 CassandraBolt
 Writes data to Cassandra
 Available in Batching and Non-Batching
 CassandraLookupBolt
 Reads data from Cassandra

Storm-Cassandra Project
 Provides generic Bolts for writing/reading
Storm Tuples to/from C*

Tuple
Tuple Mapper Rows

Tuples
Columns
Mapper Columns C*

 TupleMapper Interface
 Tells the CassandraBolt how to write a tuple to an
arbitrary data model

 Given a Storm Tuple:
 Map to Column Family
 Map to Row Key
 Map to Columns


 ColumnsMapper Interface
 Tells the CassandraLookupBolt how to transform a
C* row into a Storm Tuple

 Given a C* Row Key and list of Columns:
 Return a list of Storm Tuples


 Current State:
 Version 0.4.0
 Uses Astyanax Client
 Several out-of-the-box *Mapper Implementations:
○ Basic Key-Value Columns
○ Value-less Columns
○ Counter Columns
○ Lookup by row key
○ Lookup by range query
 Composite Key/Column Support
 Trident support


 Future Plans:
 Switch to CQL
 Enhanced Trident Support


Word Count Demo


Trident
 Provides a higher-level abstraction for stream
processing
 Constructs for state management and Batching
 Adds additional primitives that abstract away
common topological patterns
 Deprecates transactional topologies
 Distributes with Storm

Sample Trident Operations
 Partition Local
 Functions ( execute(x)  x + y )
 Filters ( isKeep(x)  0,x )
 PartitionAggregate
○ Combiner ( pairwise combining )
○ Reducer ( iterative accumulation )
○ Aggregator ( byoa)

A sample topology
TridentTopology topology = new TridentTopology();
TridentStatewordCounts =
topology.newStream("spout1", spout)
.each(new Fields("sentence"),
new Split(),
new Fields("word"))
.groupBy(new Fields("word"))
.persistentAggregate(
MemcachedState.opaque(serverLocations),
new Count(),
new Fields("count"))
.parallelismHint(6);

https://github.com/nathanmarz/storm/wiki/Trident-state

Trident State
Sequenced writes by batch/transaction id.
 Spouts
 Transactional
○ Batch contents never change
 Opaque
○ Batch contents can change
 State
 Transactional
○ Store tx_id with counts to maintain sequencing of writes.
 Opaque
○ Store previous value in order to overwrite the current value
when contents of a batch change.

Shameless Shoutouts
 HMS (https://github.com/hmsonline/)
 storm-cassandra
 storm-elastic-search
 storm-jdbi (coming soon)

 ptgoetz (https://github.com/ptgoetz)
 storm-jms
 storm-signals

Cassandra and Storm at Health Market Sceince

More Related Content

What's hot (19)

Viewers also liked (12)

Similar to Cassandra and Storm at Health Market Sceince (20)

Recently uploaded (20)

Cassandra and Storm at Health Market Sceince