Mumbai MuleSoft Meetup:Batch Processing, Anypoint Messaging Queue and Custom Connectors

31st August,2019: Batch Processing, Anypoint Messaging Queue and
Custom Connectors
Mumbai - India

All contents © MuleSoft Inc.
Agenda
2
2:00 PM Introduction
2:30 PM Technical Session – Anypoint MQ
3:30 PM Technical Session – Batch Processing
4:30 PM Networking & Break: Tea, Coffee and Snack
5:00 PM Technical Sessions - Custom Connectors
5:30 PM Wrap-up and Plan For Next Meetup

Introductions
3
• About the organizer:
– Sudeshna Mitra
– Manish Yadav
– Akshata Sawant
– Ranveer Meel
• About the sponsor:
– 91 SPRINGBOARD & APISERO
A SHOW OF HANDS:
Who is new to this MeetUp?

Speakers
4

Surprises!
5
• Thanks for your positive
enthusiasm …You earned
it ..
• Certification Vouchers
for all!
• Be geared up for the
quiz sessions at the end
of each modules! 

Introduction
7
• Anypoint MQ features:-
- multi-tenant
- Cloud messaging service
- performs async messaging scenarios
- Fully integrated with Anypoint platform
- Role based access control
- Only available in enterprise addition

Type of Queue’s
8
Types :-
• Queue
• FIFO Queue
• Exchange

Queue
9
Configuring Queue:
- ID: a-z, A-Z, 0-9, period (.), and hyphen (-)
- Message TTL - (Time to live)
- Message Lock Default TTL
- Encryption (Optional)- PBE with MD5 and triple DES
- Assign a Dead Letter Queue (Optional)
- Dead Letter Queue Name
- Delivery attempts before reroute
- In-Queue
- In-Flight

All contents © MuleSoft Inc. 10
• An in-flight message is:
- A message received by a queue, but not deleted.
- A message awaiting ACK or NACK.
- A message with an expired Message Lock Default TTL (time-to-live) setting.
• Anypoint MQ supports up to 120,000 in-flight messages per each standard (non-
FIFO) queue.
• For FIFO queues permit is up to 10 in-flight messages per queue due to the extra
processing required for FIFO queues
Note :- A single queue can contain an unlimited number of messages
The maximum duration for any message, either in-flight or not, is 2 weeks, after
which Anypoint MQ deletes the message
In Flight

Queue
11
Constraints :
- maximum 127 characters in name
- maximum 10 mb message size (There is a way to overcome this limit)
- maximum time to live (TTL) for a message is 2 weeks
- maximum time to lock (TTL) is 12 hrs.
- maximum 120,000 in-flight messages for non FIFO queues
- maximum 10 in flight messages for FIFO queues

Queue
12
Create Queue :

FIFO Queue
13
WHY?
For applications where the order of messages needs to be strictly preserved and
enforced

FIFO Queue keypoints :
14
- Works on First In First Out principal.
- Strict message ordering.
- Cannot be bound to a message exchange.
- Encryption doesn’t effect ordering.
- Up to 10 inflight messages per queue.
- Supports 300 TPS. Can increase to 3,000 TPS using 10 messages batch.
- Guaranteed exactly once delivery with same message ID
- Deduplication interval is 5 mins.
- Comparatively slow.

DLQ Queue
15
WHY?
To ensure that messages that are not successfully delivered are sent for backup to
a queue

DLQ Queue key points :-
16
- Works as backup to a queue.
- Same as any other queue.
- At least 2 queues required.
- Sideline and isolate the unsuccessfully processed messages.
-- Must be either non-FIFO or FIFO.
- Only FIFO to FIFO or non-FIFO to non-FIFO allowed.
- Same region as other queues.
- Messages not charged if sent from other queue.
- Messages charged if sent directly.
- TTL and Encryption depends on configuration.

DLQ Flow Chart :-
17

MQ connector – Basic config
18

MQ connector – Advanced config
19
• Anypoint MQ connector 3.x
released on 29th July’19 supports
below functionalities –
- Circuit breaker is simplified
- Subscriber can be configured in
3 ways:
• Polling
• Fixed frequency
• Cron Job

Circuit Breaker

• Publish
• Subscribe
• Consume
• Ack
• Nack
Note :- Above operations can also be performed directly from
cloudhub with message send, message browse, return and
delete/purge functionality
MQ Operations :-

• Horizontal scalability
• Reliability
• Decoupling
• Aggregation
Queue Benefits :-

- Provides a way to distribute a single message to many consumers
- Queue can be bounded to multiple exchanges
- Any message published to exchange will be multicast to all binded
queue’s
- There is no provision to apply Fan-Out logic at Exchange incase an
message needs to be routed to specific queues only
- Similar to Queue, an Exchange can also be encrypted
- With Exchange, we can implement pub/sub integration pattern
Exchange

• Currently, Anypoint MQ cannot be deployed on-premises
• However, we can still access this from on-prem with below approach-
- Publishing a message from an on-premises system to Anypoint MQ(Cloudhub)
- Another on-premises subscriber pulling messages from Anypoint MQ(Cloudhub)
Use Anypoint MQ on-premises

- The Anypoint MQ APIs let you use REST to send and receive
messages, administer queues and message exchanges, and analyze
results
• Anypoint MQ provides these APIs:
- Anypoint MQ Admin API
- Anypoint MQ Broker API
- Anypoint MQ Stats API
Anypoint MQ REST API

Anypoint MQ Admin API
• Browse queues and message exchanges
• Create queues and message exchanges
• Manage bindings for message exchanges
Anypoint MQ Broker API
This API lets you send and receive messages from queues and
message exchanges.

Anypoint MQ Stats API
• Anypoint MQ metrics:
- Real time statistics for queues
- Historic statistics for queues
- Historic statistics for message exchanges
• Usage metrics - View your entire organization’s usage of Anypoint MQ for billing
purposes:
- Usage per environment
- Usage per organization
• Browse statistics for queues and message exchanges
• View the number of messages currently in a queue

- As per documentation, max payload size is 10 MB while publishing
- If we have message >10MB then we can use below approach to
overcome the limit –
a. Encrypt the message using Gzip module
b. Encode the above output using base64 function in DW
c. Publish this payload to Queue
Note :- While subscribing repeat above process in reverse
Advantage :- Compression ratio achieved vary from 0.8 to 0.95,
thereby we can publish payload of size 80+ MB also safely using above
method
Compression/Decompression

• ANYPOINT-MQ:UNKNOWN
• ANYPOINT-MQ:RETRY_EXHAUSTED
• ANYPOINT-MQ:CONNECTIVITY
• ANYPOINT-MQ:RESOURCE_NOT_FOUND
• ANYPOINT-MQ:TIMEOUT
Error Types :-

Anypoint MQ Stats :-

Different Message brokers
Anypoint MQ Rabbit MQ Active MQ
Availability Licensed open source open source
Infrastructure Available only on cloud Can be on cloud or
on-prem
Can be on cloud or on-
prem
Protocols Only REST Multi-protocol
AMQP,STOMP,MQTT,HT
TP etc.
Multi protocol
AMQP, MQTT, REST,
Stomp etc.
Message size 10 MB limit Theoretical size limit
is 2GB (recommended
not more than 128
MB)
Equivalent to heap
memory size
Filtering support Filtering cannot be
applied directly
Filtering cannot be
applied
Filtering can be applied
Message TTL Message TTL – 2 Weeks Message TTL –>
infinite (By default)
Message TTL –> infinite
(By default)

• https://docs.mulesoft.com/release-notes/connector/anypoint-mq-connector-release-
notes-mule-4#connector/mq-connector-release-notes
• https://docs.microsoft.com/en-us/azure/architecture/patterns/circuit-breaker
• https://docs.mulesoft.com/connectors/anypoint-mq/3.x/anypoint-mq-connector
• https://docs.mulesoft.com/mq/mq-understanding
References :-

• With Batch, Mule aims to simplify integration use cases where basic
ETL functionality is needed
• For example, you can use batch processing when:
- Synchronizing data sets between business applications, such as syncing contacts between
NetSuite and Salesforce.
- Extracting, transforming and loading (ETL) information into a target system, such as
uploading data from a flat file (CSV) to Hadoop.
- Handling large quantities of incoming data from an API into a legacy system.
Introduction

Introduction
Each batch job contains three different phases:
1.Load and Dispatch.
2.Process.
3.On Complete.

• Load and Dispatch
- Implicit
- No action required by developer
- Mule splits the message into collection of records for processing
- It also creates batchJobInstanceId variable
- It also creates persistent queue and associate it with batch job
instance
- Each generated record is saved into queue. This activity is – “all or
nothing”
Phase 1

• Process
- Asynchronous – meaning Flow will not wait for the Batch job to finish
Phase 2
- Batch job instance does not wait for all its queued
records to finish processing in one batch step before
pushing any of them to the next batch step. Queues
are persistent.
- Mule persists a list of all records as they succeed or fail to
process through each batch step, thereby skipping failed
one in subsequent steps
- Filters can be applied by adding acceptExpressions
over batch step
- Batch aggregator processor can be used to aggregate
records and upsert the same in bulk

• On Complete
- This phase create a report or summary of the records that got
processed in Batch Job Instance
- As Mule execute batch job as an asynchronous, one-way flow. The
result of batch processing(BatchJobResult obj) does not feed back
into the flow which may have triggered it
Phase 3

• Batch was a top-level element just like flows in Mule 3. In Mule 4, it
exist as a scope which lives inside a flow
• There was input phase in Mule 3 which is not present in mule 4
• The batch:execute operation doesn’t exist any more
• No more record variable. In mule 4, regular variable can be used in
Batch Steps, with each record having its own copy
• In Mule 3, input phase was required to o/p Java structure in case
input was XML or JSON using DW. In mule 4, this all is done by new
automatic streaming framework and everything will just work out of
the box
• Batch commit is now called as batch aggregator
Batch component Mule 3 vs Mule 4

• Every record has its own set of variables which can be new variables
or modification of already existing variables
• These variables are not visible to other records getting processed
parallelly
• For each record, these variables are propagated across batch steps
• In the on Complete phase, none of these variables (not even original
ones) are visible
• Since Batch Job executes in Asynchronously from the rest of flow, no
variable set in within batch step or on complete phase are visible
outside the batch scope
Variable Propagation

• Handling error inside the batch job can be complicated considering
the huge amount of data this scope is designed to take
• For any failures in any of batch step, Mule logs stack trace for the
first occurrence of that exception, for e.g. :-
com.mulesoft.mule.runtime.module.batch.internal.DefaultBatchStep: Found exception processing record on step 'batchStep1' for job
instance 'Batch Job Example' of job 'CreateLeadsBatch’.
This is the first record to show this exception on this step for this job instance. Subsequent records with the same failures will not be
logged for performance and log readability reasons:
• Mule logs on a "by step" basis. If another step also raises the same
error, the runtime logs it again for that step.
• when the batch job reaches the On Complete phase, Mule displays an
error summary with every error type, and how many times it
happened in each batch step.
Error Handling

• Mule has three options for handling a record-level error :
1. Finish Processing
- stop execution of current job instance
- finish execution of current in-flight records
- do no pull any more records from queue and set job instance to FALIURE state
- Invoke onComplete phase
2. Continue Processing
- continue processing the batch regardless of any failures
- use acceptExpression and acceptPolicy attributes to instruct subsequent batch
steps on how to handle batch failures
- use maxFailedRecords attribute to define threshold value for No of failures
before stopping the batch job instance
Batch Processing Strategies for Error Handling

• https://docs.mulesoft.com/mule-runtime/4.2/batch-processing-
concept
• https://docs.mulesoft.com/mule-runtime/4.2/batch-error-handling-
faq
• https://docs.mulesoft.com/mule-runtime/4.2/batch-filters-and-
batch-aggregator
• https://blogs.mulesoft.com/dev/mule-dev/batch-module-mule-4-
improvements/
References :-

Using Mule SDK
Custom Connectors

Why Custom Connector (PART 1)
45
– Act as a reusable component in the integration system.
– Act as an abstract layer between the API’s complexity.
• Don’t want to expose actual connectivity logic
– Developers can focus more on the business logic.
– Your Own Connector.

Why Custom Connector (PART 2)
46

Custom Connectors
47
– It can be built in 2 Ways:
– Mule SDK
– XML SDK

Mule 4:
Mule SDK Vs DevKit
48
• DevKit is not available for Mule 4.
• The Mule SDK is not tightly bound to Anypoint Studio.
– instead of having to install the custom module in Anypoint Studio, you just have to add it
as a Maven dependency.
• Built on the same idea of annotated Java classes.
• The SDK is not a code generator.
– It is just a plain extension mechanism.
<parent>
<groupId>org.mule.extensions</groupId>
<artifactId>mule-modules-parent</artifactId>
<version>1.1.3</version>
</parent>

Prerequisites :
Custom Connector Mule SDK
49
• You must have these components installed.
– Java Development Kit 8 (JDK 8)
– Apache Maven 3.3.9 or higher
– Mule Target Versions:
• Mule ESB 4
• Anypoint Studio 7

Let’ Start : 1 )
50
• mvn org.mule.extensions:mule-extensions-archetype-maven-plugin:1.2.0:generate

2 ) Formatting the generated project
51

3.1 ) Import the generated project
52

3.2 ) Import the generated project
53

4 ) Import the generated project
54

5 ) Connector is ready to use. ( Part 1 )
55

6 ) Connector is ready to use. ( Part 2 )
56
• Add as a dependency in pom.xml

Connector is ready with default configuration (Part 1)
57

Connector is ready with default configuration (Part 2)
58

59
Primary Contents Mule SDK Project
– pom.xml
– FirstExtension.java
– FirstConfiguration.java
– FirstConnectionProvider.java
– FirstConnection.java
– FirstOperations.java

Mule SDK Project
60
• 1. SampleWebDavExtension.java
– Main entry point class.
– Marked with @Extension
annotation.
– @Xml (prefix=“samplewebdav”)

Mule SDK Project
61
• 2. FirstConfiguration.java
• Configuration class defines the operations
available with it.
• @Parameter

Mule SDK Project
62
• 3. FirstConnectionProvider.java
– @Parameter @DisplayName("Friendly
Name") @Optional(defaultValue =
"100")

Mule SDK Project
63
• 3. FirstConnectionProvider.java
– The actual pooling is not implemented on the provider itself but performed by the runtime.
Implements:
– PoolingConnectionProvider:
• connections resolved by this provider will be pooled and reused.
– connect() method will be called only once during start of the application.
– disconnect() method will be called only during shutdown/restart of the application.
– CachedConnectionProvider:
• lazily creates and caches connections.
– ConnectionProvider:
• new connection each time.
– connect()/disconnect() method will be called during start of the application.
– Later on for each process connect() -> respective operation() -> disconnect() method will
be invoked.

Mule SDK Project
64
• 4. FirstConnection.java
• FirstConnection Connection Object
Instance managed by Connection
Provider.
• This class should have any methods
such as invalidate(), and attributes
that makes sense for an instance of a
Connection, once established.

Mule SDK Project
65
5. FirstOperations.java (PART 1)
• Every public method in this class
becomes the operation supported by
this module.
• void: As the name indicates, this
operation will not return anything.
This operation will not modify the
message and next component will
receive the same message.

Mule SDK Project
66
• Only String will be accepted.
• return type will be String

Mule SDK Project
67
• Parameters will be available as a part of
connection.

Mule SDK Project
68
• Payload can be accessed Via
@Content(primary = true) String payload

Sample Program for Open Drive(Part 1)
69
• List down important parameters for
connection
• Check for dependency
– <dependency>
<groupId>com.github.look
first</groupId>
<artifactId>sardine</artifactId>
<version>5.8</version>
</dependency>

Sample Program for Open Drive(Part 2)
70
pom.xml No changes in entry point
class
No Changes in Configuration
class

Convert Java Program to Mule SDK Connector
(Part 3)
71
• Supply username and password to
create a connection instance.
• Points to remember:
– connect() and disconnect() method will be
called based upon implementation.
– Responsibilities :
• Connect()
– Is to create connection instance.
• Disconnect()
– Is to invalidate connection.

(Part 4)
72
• Responsibilities:
– Getter method is required
to get the connection
object.
– Make sure you write a code
for invalidating the
connection.

(Part 5)
73
• Responsibilities:
– Connection instance should be
used within operation.
– Rest logic remains the same.
– Run maven below maven
command to install the connector:
– Mvn clean install –DskipTests
– Add the target jar to your project.

Reconnection on Connectors (PART 1)
74
Connector’s Responsibility
– responsible for communicating when a reconnection should occur, when a connection is
down and when a new connection is required.
Examples of Connection Errors
– Connection Timeout
– Invalid Credentials
– Token Expired
Examples of Non-connection Errors
– Missing required HTTP Query Param when doing a HTTP request
– Directory does not exist when trying to copy a file
– Syntax error when executing a Database Query

75
Reconnection can occur during below stages:
Creating Connection Validating Connection

76
Operations
public void
operationTriggersReconnection(@Connection
Connection connection,boolean
connectionException)
{
if(connectionException){
throw new ConnectionException("Connection
Error"); (1)
} else {
throw new RuntimeException(); (2)
}
}
Sources
• Sources have two stages
• Starting
• Running

77

Module Structure
78

More SDK Features (PART 1)
79
Operations Sources Functions Routers Scopes
Operations process
messages.
Sources create messages
and push them to a flow.
Mule SDK contribute
functions to DataWeave
Example:
route can be used to
decide whether or not
the contained processors
should be executed.
Scopes are similar
to Operations, but their
execution includes the
execution of other
child Operations.
A connection is
obtained each time
the operation is
executed.
A connection is obtained
each time the message
source is started or when
reconnection happens.
Config-less and
Connection-less
Routers are required to
extend the Route class.
Config-less and
Connection-less
Sources are required to
extend the Source class.
A Message source must
have start() and stop()
phases
Void Routers: At the
end of its execution
leaves the event and
message the same as it
was before the execution
of the router
Examples of message
sources can be:
• An HTTP listener
• A File watcher
• …
Config-less and
Connection-less

More SDK Features (PART 2)
80
• Transactions
– Connection Provider class should implementing TransactionalConnection interface requires implementation of
the following methods:
• begin() : This method is called, when the transaction must begin.
• commit() : This is called when the transaction finished correctly and must be committed.
• rollback() : This is called then the transaction finished with an error and this one must be be rollbacked.
• Security
– Support for
• TLS Configuration
• OAuth
• Dynamic configurations
– #[attributes.queryParams.encoding]
• Non-blocking operations
• Classloading isolation

Take a stand !
81
• Nominate yourself for
the next meetup speaker
and suggest a topic as
well.

What’s next
82
• Share:
– Tweet your pictures with the hashtag #MuleMeetup #MuleSoftMeetup
– Invite your network to join: https://meetups.mulesoft.com/mumbai/
• Feedback:
– Contact your organizers Manish Yadav, Sudeshna Mitra, Akshata Sawant and Ranveer Meel to
suggest topics
– Tweet your organizers at @SudeshnaMitra14, @Manish_Kyadav, @sawantakshata02 and
@meel_Ranveer
– Contact MuleSoft at meetup@mulesoft.com for ways to improve the program
– Follow us on Instagram (mumbai_mulesoftofficialpage)
– Your Feedback is Food for us
• Our next meetup:
– Date: TBD
– Location: Mumbai
– Topic: TBD

Networking time
Introduce yourself to your neighbors!

See you next time
Please send topic suggestions to the organizer

Mumbai MuleSoft Meetup:Batch Processing, Anypoint Messaging Queue and Custom Connectors

Mumbai MuleSoft Meetup:Batch Processing, Anypoint Messaging Queue and Custom Connectors

Recommended

More Related Content

What's hot (20)

Similar to Mumbai MuleSoft Meetup:Batch Processing, Anypoint Messaging Queue and Custom Connectors (20)

More from Manish Kumar Yadav (10)

Recently uploaded (20)

Mumbai MuleSoft Meetup:Batch Processing, Anypoint Messaging Queue and Custom Connectors