Developing streaming applications with apache apex (strata + hadoop world)

Developing Streaming Applications with Apache Apex

David Yan <[email protected]>PMC Member, Apache ApexSoftware Engineer, Google

Strata+Hadoop World, San Jose, CAMarch 16, 2017

Agenda● Technical Overview of Apache Apex as a streaming

platform● Developing an Application Pipeline● Developing an Operator, and the Operator Library● Event Time Windowing● Debugging, Scaling and Tuning● Q & A

What is Apache Apex? Why?● A platform for streaming applications● Written in Java● Uses YARN for resource management and HDFS for storage● Has a comprehensive operator library.● Scalable, fault tolerant, high throughput and low latency● Provides low-level control for performance tuning● Users include GE (Predix), Capital One, Royal Bank of Canada,

Pubmatic, SilverSpring Network, etc. ○ (more at https://apex.apache.org/powered-by-apex.html)

Technical Overview

Application Development Model

● Directed Acyclic Graph (DAG)● A stream is a sequence of data tuples● An operator takes one or more input streams, performs computations, emits

one or more output streams

Filtered

Stream

Output StreamTuple Tuple

Filtered Stream

Enriched Stream

Enriched

Stream

er

Operator

er

Operator

er

Operator

er

Operator

er

Operator

er

Operator

From Logical DAG to Actual Work ● CLI issues a launch command with

App Package● YARN starts up Application Master● Logical DAG is translated to

Physical DAG

Apex CLI YARN RM

NM NM NM NM

AM

1 23

45 6Input Filter Transform Output

From Logical DAG to Actual Work(cont'd)● AM requests worker containers

from YARN to run physical operators

● Worker Containers send data using a pub-sub mechanism

Apex CLI YARN RM

AM Worker WorkerWorker

Worker

Worker6

4

1

3

25

1 23

45 6

HDFSPubSub overflow & Checkpoints

Streaming Windows & Checkpointing● Stream is divided into fixed time slices

called streaming windows● Checkpoint is performed by Worker

Containers at streaming window boundaries

● Worker Containers send heartbeats to AM

● Recovery is incremental without resetting full DAG

● Checkpoints are purged after the corresponding window is committed

● AM is also checkpointed

BeginWindow nEndWindow n

BeginWindow n+1

EndWindow n+1

Time

......

Bookkeeping & Checkpointing done here

Writing an Apex Application

Pipeline

Writing an Apex Application Pipeline ● Creating a project: Maven archetype● Two APIs to specify an Apex pipeline

○ Compositional○ Declarative

Compositional APIInput Parser Counter Output

CountsWordsLines

Kafka Database

FilterFiltered

Declarative API

StreamFactory.fromKafka09(brokers, topic)

.flatMap(input -> Arrays.asList(input.split("\\s+")))

.filter(input -> filterWords.contains(input))

.window(new WindowOption.GlobalWindow(),

new TriggerOption().accumulatingFiredPanes().withEarlyFiringsAtEvery(1))

.sumByKey(input -> new Tuple.PlainTuple<>(new KeyValPair<>(input, 1L)))

.map(input -> input.getValue())

.endWith(new JdbcOutput())

.populateDag(dag);

Input Parser Counter OutputCountsWordsLines

Kafka Database

FilterFiltered

Operators

Operator API

Next streamingwindow

Next streaming window

Input Adapters - Starting of the pipeline. Interacts with external system to generate streamGeneric Operators - Processing part of pipelineOutput Adapters - Last operator in pipeline. Interacts with external system to finalize the processed stream

OutputPort::emit()

Operator Example (Stateless)

Operator Example (Stateful)

RDBMS• JDBC• MySQL• Oracle• MemSQL

NoSQL

• Cassandra, HBase• Aerospike, Accumulo• Couchbase/ CouchDB• Redis, MongoDB• Geode

Messaging• Kafka• JMS (ActiveMQ etc.)• Kinesis, SQS • Flume, NiFi• MQTT

File Systems• HDFS/ Hive• Local File• S3

Parsers• XML • JSON• CSV• Avro• Parquet

Transformations• Filters, Expression, Enrich• Windowing, Aggregation• Join• Dedup

Analytics• Dimensional Aggregations

(with state management for historical data + query)

Protocols• HTTP• FTP• WebSocket• SMTP

Other• Elastic Search• Script (JavaScript, Python, R)• Solr• Twitter

Apex "Malhar" Operator Library

Queryable StateA set of operators in the library that supports real-time queries and updates of operator state.

HashTag Extractor

Top Counter

Twitter Feed Input Operator

Unique Counter

Query

Result

PubSub Broker

Query Input

Queryable State (cont'd)

Event Time Windowing

What, Why, and How?● Event Time Windows not to be confused with "streaming windows"● Time of the event vs processing time and ingression time● Data often arrives late and out of order● Concepts outlined by Google's Millwheel and Apache Beam: Watermarks,

Allowed Lateness, Accumulation Modes, and Triggers

ExampleStreamFactory.fromFolder("/tmp")

.flatMap(input -> Arrays.asList(input.split( "\\s+")))

.map(input -> new TimestampedTuple<>(System.currentTimeMillis(), input))

.window(new TimeWindows(Duration.standardMinutes(5)), TriggerOption.AtWatermark() .accumulatingFiredPanes() .withEarlyFiringsAtEvery(Duration.standardSeconds(1)), Duration.standardSeconds(15)) .sumByKey(input -> new TimestampedTuple<>(input.getTimestamp(), new KeyValPair<>(input.getValue(),

1L ))))

.map(new FormatAsTableRowFn()) // format for printing

.print()

.populateDag(dag);

Debugging, Scaling & Tuning Your

Application

Debugging● Logs● Local Mode● Remote JVM Attachment● Tuple Recording

Scaling: PartitioningPartitioning with Unifiers: NxM Partitioning:

0 1 2

0

1a

1b

1c

U 2

Logical DAG

Physical DAG with operator 1 with 3 partitions

0 21 3

0

2a

1b 3

1a

1c

2b

U U

0

2a

1b 3

1a

1c

2b

U1

U

U2

Scaling: Partitioning (cont'd)Parallel Partition: Cascading Unifiers:

0

1a

1b

U 2 3 4

0

1a

1b

U 4

2a

3a

2b

3b

1 2

1

1

1

1

U 2

1

1

1

1

U1

2U2

U3

Dynamic Partitioning● Partitioning change while

application is running○ Change number of partitions at

runtime based on stats○ Supports re-distribution of state when

number of partitions change○ API for custom scaler or partitioner

StatelessThroughputBasedPartitioner<MyOperator> partitioner = new StatelessThroughputBasedPartitioner<>(); partitioner.setCooldownMillis(45000);partitioner.setMaximumEvents(30000);partitioner.setMinimumEvents(10000);

dag.setAttribute(op, OperatorContext.STATS_LISTENERS, Arrays.asList(new StatsListener[]{partitioner}));

dag.setAttribute(op, OperatorContext.PARTITIONER, partitioner);

Locality● Node-Local: The two operators should be placed in the same node● Container-Local: Same worker container (saves serialization)● Thread-Local: Same thread (simple function call)

// setting locality between op1 and op2 to be NODE_LOCALdag.addStream("streamName", op1.output, op2.input) .setLocality(DAG.Locality.NODE_LOCAL);

Recent Additions and Roadmap● Apex runner in Apache Beam● Iterative processing● Integrated with Apache Samoa, opens up ML● Integrated with Apache Calcite, allows SQL

---------------------------------------------

● Enhanced support for Batch Processing● Support for Mesos and Kubernetes● Encrypted Streams● Python API

Q & A

Resources● http://apex.apache.org/● Learn more: http://apex.apache.org/docs.html ● GitHub Repositories

○ http://github.com/apache/apex-core○ http://github.com/apache/apex-malhar

● Subscribe - http://apex.apache.org/community.html● Download - http://apex.apache.org/downloads.html● Follow @ApacheApex - https://twitter.com/apacheapex● Meetups – http://www.meetup.com/topics/apache-apex/● Slideshare: http://www.slideshare.net/ApacheApex/presentations● https://www.youtube.com/results?search_query=apache+apex