This is an overview of architecture with use cases for Apache Apex, a big data analytics platform. It comes with a powerful stream processing engine, rich set of functional building blocks and an easy to use API for the developer to build real-time and batch applications. Apex runs natively on YARN and HDFS and is used in production in various industries. You will learn more about two use cases: A leading Ad Tech company serves billions of advertising impressions and collects terabytes of data from several data centers across the world every day. Apex was used to implement rapid actionable insights, for real-time reporting and allocation, utilizing Kafka and files as source, dimensional computation and low latency visualization. A customer in the IoT space uses Apex for Time Series service, including efficient storage of time series data, data indexing for quick retrieval and queries at high scale and precision. The platform leverages the high availability, horizontal scalability and operability of Apex.
TrustArc Webinar - How to Build Consumer Trust Through Data Privacy
Hadoop Summit SJ 2016: Next Gen Big Data Analytics with Apache Apex
1. Next Gen Big Data Analytics with Apache Apex
Thomas Weise, Pramod Immaneni
Hadoop Summit, San Jose, June 30th 2016
2. Stream Data Processing
• Data from variety of sources (IoT, Kafka, files, social media etc.)
• Unbounded, continuous data streams
ᵒ Batch can be processed as stream (but a stream is not a batch)
• (In-memory) Processing with temporal boundaries (windows)
• Stateful operations: Aggregation, Rules, … -> Analytics
• Results stored to variety of sinks or destinations
ᵒ Streaming application can also serve data with very low latency
2
Browser
Web Server
Kafka Input
(logs)
Decompress,
Parse, Filter
Dimensions
Aggregate Kafka
Logs
Kafka
3. Apache Apex
3
• In-memory, distributed stream processing
• Application logic broken into components called operators that run in a distributed fashion
across your cluster
• Natural programming model
• Unobtrusive Java API to express (custom) logic
• Maintain state and metrics in your member variables
• Scalable, high throughput, low latency
• Operators can be scaled up or down at runtime according to the load and SLA
• Dynamic scaling (elasticity), compute locality
• Fault tolerance & correctness
• Automatically recover from node outages without having to reprocess from beginning
• State is preserved, checkpointing, incremental recovery
• End-to-end exactly-once
• Operability
• System and application metrics, record/visualize data
• Dynamic changes
6. Application Development Model
6
A Stream is a sequence of data tuples
A typical Operator takes one or more input streams, performs computations & emits one or more output streams
• Each Operator is YOUR custom business logic in java, or built-in operator from our open source library
• Operator has many instances that run in parallel and each instance is single-threaded
Directed Acyclic Graph (DAG) is made up of operators and streams
Directed Acyclic Graph (DAG)
Output
Stream
Tupl
e
Tupl
e
er
Operator
er
Operator
er
Operator
er
Operator
er
Operator
er
Operator
9. Scalability
9
NxM PartitionsUnifier
0 1 2 3
Logical DAG
0 1 2
1
1 Unifier
1
20
Logical Diagram
Physical Diagram with operator 1 with 3 partitions
0
Unifier
1a
1b
1c
2a
2b
Unifier 3
Physical DAG with (1a, 1b, 1c) and (2a, 2b): No bottleneck
Unifier
Unifier0
1a
1b
1c
2a
2b
Unifier 3
Physical DAG with (1a, 1b, 1c) and (2a, 2b): Bottleneck on intermediate Unifier
10. Advanced Partitioning
10
0
1a
1b
2 3 4Unifier
Physical DAG
0 4
3a2a1a
1b 2b 3b
Unifier
Physical DAG with Parallel Partition
Parallel Partition
Container
uopr
uopr1
uopr2
uopr3
uopr4
uopr1
uopr2
uopr3
uopr4
dopr
dopr
doprunifier
unifier
unifier
unifier
Container
Container
NICNIC
NICNIC
NIC
Container
NIC
Logical Plan
Execution Plan, for N = 4; M = 1
Execution Plan, for N = 4; M = 1, K = 2 with cascading unifiers
Cascading Unifiers
0 1 2 3 4
Logical DAG
11. Dynamic Partitioning
11
• Partitioning change while application is running
ᵒ Change number of partitions at runtime based on stats
ᵒ Determine initial number of partitions dynamically
• Kafka operators scale according to number of kafka partitions
ᵒ Supports re-distribution of state when number of partitions change
ᵒ API for custom scaler or partitioner
2b
2c
3
2a
2d
1b
1a1a 2a
1b 2b
3
1a 2b
1b 2c 3b
2a
2d
3a
Unifiers not shown
12. Fault Tolerance
12
• Operator state is checkpointed to persistent store
ᵒ Automatically performed by engine, no additional coding needed
ᵒ Asynchronous and distributed
ᵒ In case of failure operators are restarted from checkpoint state
• Automatic detection and recovery of failed containers
ᵒ Heartbeat mechanism
ᵒ YARN process status notification
• Buffering to enable replay of data from recovered point
ᵒ Fast, incremental recovery, spike handling
• Application master state checkpointed
ᵒ Snapshot of physical (and logical) plan
ᵒ Execution layer change log
13. • In-memory PubSub
• Stores results emitted by operator until committed
• Handles backpressure / spillover to local disk
• Ordering, idempotency
Operator
1
Container 1
Buffer
Server
Node 1
Operator
2
Container 2
Node 2
Buffer Server
13
14. End-to-End Exactly Once
14
• Important when writing to external systems
• Data should not be duplicated or lost in the external system in case of
application failures
• Common external systems
ᵒ Databases
ᵒ Files
ᵒ Message queues
• Exactly-once = at-least-once + idempotency + consistent state
• Data duplication must be avoided when data is replayed from checkpoint
ᵒ Operators implement the logic dependent on the external system
ᵒ Platform provides checkpointing and repeatable windowing
22. Maximize Revenue w/ real-time insights
22
PubMatic is the leading marketing automation software company for publishers. Through real-time analytics,
yield management, and workflow automation, PubMatic enables publishers to make smarter inventory
decisions and improve revenue performance
Business Need Apex based Solution Client Outcome
• Ingest and analyze high volume clicks &
views in real-time to help customers
improve revenue
- 200K events/second data
flow
• Report critical metrics for campaign
monetization from auction and client
logs
- 22 TB/day data generated
• Handle ever increasing traffic with
efficient resource utilization
• Always-on ad network
• DataTorrent Enterprise platform,
powered by Apache Apex
• In-memory stream processing
• Comprehensive library of pre-built
operators including connectors
• Built-in fault tolerance
• Dynamically scalable
• Management UI & Data Visualization
console
• Helps PubMatic deliver ad performance
insights to publishers and advertisers in
real-time instead of 5+ hours
• Helps Publishers visualize campaign
performance and adjust ad inventory in
real-time to maximize their revenue
• Enables PubMatic reduce OPEX with
efficient compute resource utilization
• Built-in fault tolerance ensures
customers can always access ad
network
23. Industrial IoT applications
23
GE is dedicated to providing advanced IoT analytics solutions to thousands of customers who are using their
devices and sensors across different verticals. GE has built a sophisticated analytics platform, Predix, to help its
customers develop and execute Industrial IoT applications and gain real-time insights as well as actions.
Business Need Apex based Solution Client Outcome
• Ingest and analyze high-volume, high speed
data from thousands of devices, sensors
per customer in real-time without data loss
• Predictive analytics to reduce costly
maintenance and improve customer
service
• Unified monitoring of all connected sensors
and devices to minimize disruptions
• Fast application development cycle
• High scalability to meet changing business
and application workloads
• Ingestion application using DataTorrent
Enterprise platform
• Powered by Apache Apex
• In-memory stream processing
• Built-in fault tolerance
• Dynamic scalability
• Comprehensive library of pre-built
operators
• Management UI console
• Helps GE improve performance and lower
cost by enabling real-time Big Data
analytics
• Helps GE detect possible failures and
minimize unplanned downtimes with
centralized management & monitoring of
devices
• Enables faster innovation with short
application development cycle
• No data loss and 24x7 availability of
applications
• Helps GE adjust to scalability needs with
auto-scaling
24. Smart energy applications
24
Silver Spring Networks helps global utilities and cities connect, optimize, and manage smart energy and smart city
infrastructure. Silver Spring Networks receives data from over 22 million connected devices, conducts 2 million
remote operations per year
Business Need Apex based Solution Client Outcome
• Ingest high-volume, high speed data from
millions of devices & sensors in real-time
without data loss
• Make data accessible to applications
without delay to improve customer service
• Capture & analyze historical data to
understand & improve grid operations
• Reduce the cost, time, and pain of
integrating with 3rd party apps
• Centralized management of software &
operations
• DataTorrent Enterprise platform, powered
by Apache Apex
• In-memory stream processing
• Pre-built operator
• Built-in fault tolerance
• Dynamically scalable
• Management UI console
• Helps Silver Spring Networks ingest &
analyze data in real-time for effective load
management & customer service
• Helps Silver Spring Networks detect
possible failures and reduce outages with
centralized management & monitoring of
devices
• Enables fast application development for
faster time to market
• Helps Silver Spring Networks scale with
easy to partition operators
• Automatic recovery from failures
25. Resources for the use cases
25
• Pubmatic
• https://www.youtube.com/watch?v=JSXpgfQFcU8
• GE
• https://www.youtube.com/watch?v=hmaSkXhHNu0
• http://www.slideshare.net/ApacheApex/ge-iot-predix-time-series-data-ingestion-service-using-
apache-apex-hadoop
• SilverSpring Networks
• https://www.youtube.com/watch?v=8VORISKeSjI
• http://www.slideshare.net/ApacheApex/iot-big-data-ingestion-and-processing-in-hadoop-by-
silver-spring-networks
In-memory stream processing platform
Developed since 2012, ASF TLP since 04/2016
Unobtrusive Java API to express (custom) logic
Scale out, distributed, parallel
High throughput & low latency processing
Windowing (temporal boundary)
Reliability, fault tolerance, operability
Hadoop native
Compute locality, affinity
Dynamic updates, elasticity
Large amount of data to process, arrival at high velocity
Pipelining and partitioning
Backpressure
Partitioning
Run same logic in multiple processes or threads
Each partition processes a subset of the data
Apex supports partitioning out of the box
Different partitioning schemes
Unification
Static & Dynamic Partitioning
Separation of processing logic from scaling decisions
Partitioning decision (yes/no) by trigger (StatsListener)
Pluggable component, can use any system or custom metric
Externally driven partitioning example: KafkaInputOperator
Stateful!
Uses checkpointed state
Ability to transfer state from old to new partitions (partitioner, customizable)
Steps:
Call partitioner
Modify physical plan, rewrite checkpoints as needed
Undeploy old partitions from execution layer
Release/request container resources
Deploy new partitions (from rewritten checkpoint)
No loss of data (buffered)
Incremental operation, partitions that don’t change continue processing
API: Partitioner interface
Thank Thomas, Transition to application development
Quick note Talk about apex logo presence, if you see apex logo on top it’s a functionality in Apache Apex otherwise it is a addon functionality in the DataTorrent RTS enterprise offering on top of Apache Apex.
Different ways of building applications, first through our application builder UI
App Builder UI – Thomas talked about operators as basic building blocks of the application containing the buisness logic, UI can be used to stitch these operators together by drag and drop and configuring properties
Creating an application in Java provides a much greater flexibility as you can create and use the operators in the same place, two ways
Compositional API is low level API where you specify the individual operator implementations and connect them up in a DAG or a graph. Talk about example above.
Declarative API where you specify operations on data at a high level and the ApexStream API takes care of converting it into a DAG underneath. Talk about example above.
For next version we are working on support for Apache Beam and the different even time windowing options in the specification
This is the previous example modified with windowing support.
There is one global window as we are going to continue to accumulate the counts
We want the results every second and we want to keep the counts in state
With Apex you have the flexibility to create your own custom operators and be able to use it with both low level and high level API.
Here are two examples
A snaphot of our apex malhar library listing the commonly needed operators.
Talk about robust support for kafka, dynamic partitioning, 0.8 and 0.9 API support with offset management and idempotent recovery.
Operators for streaming and batch
When your application is running you would want to know what is going on with your application. That is where monitoring console comes in. Touch upon a few salient features
Gives you a detailed look into your application including all operators and their partitions.
Gives you performance statistics, resource usage and container information for each of the partitions
Helps development and operations by providing access to the individual logs right in the console and allowing users to search the logs or change log levels on the fly.
Also lets you record data in a stream live at any stage.
Talk about locality here quickly
By default operators are deployed in containers (processes) on different nodes across the Hadoop cluster
Locality options for streams
RACK_LOCAL: Data does not traverse network switches
NODE_LOCAL: Data transfer via loopback interface, frees up network bandwidth
CONTAINER_LOCAL: Data transfer via in memory queues between operators, does not require serialization
THREAD_LOCAL: Data passed through call stack, operators share thread
Host Locality
Operators can be deployed on specific hosts
New in 3.4.0: (Anti-)Affinity (APEXCORE-10)
Ability to express relative deployment without specifying a host
Use the built-in real-time dashboards and widgets in your application or connect to your own. This picture shows the variety of widgets we have.
Wanted to talk about some instances where Apache Apex and DataTorrent are being used in production today. These cases mentioned here, customers have talked about using Apache Apex openly and have presented them in meetups. If you want to know more in depth we have provided links to the those meetup resources at the end.
Pubmatic is in advertising space and provides real time analytics around ad impressions and clicks for publishers. They are using dimensional computation operators provided by datatorrent to slice and dice the data in different ways and provide the results of those analytics through real-time dashboards.
GE is the leader in Industrial IoT and has built a cloud platform called Predix to store and analyze machine data from all over the world. There are using datatorrent and apache apex for high speed ingestion and processing in a fault tolerant way.
Silver spring networks also in the IoT space provides technology to collect and analyze data from smart energy meters for utilities. They perform ingestion and analytics with datatorrent to detect failures in the systems in advance and reduce outages.