Concurrent Programming in Java

Java & Concurrency
Lakshmi Narasimhan

2
Agenda
 United at birth
 Why more relevant now
 Concurrency over the years
 Basics & Myths
 Concurrency Patterns
 Deep-dive
 There's more than one way to solve it!!
 Common gotchas
 Q&A

3
Java & Concurrency : United at birth
●
In-built into the language
●
Not an after thought
●
Support for threads from Day 1
●
Everything is a monitor
●
Simplicity over complexity
●
Multiple inheritance
●
Operator overloading
●
Choice of languages
●
Scala, Clojure

4
Why is it more relevant now
• H/W & S/W trends
– No longer privilege of few
– Scaling accentuates Vertical and horizontal scaling mean more
complexity
– Distributed caches, faster I/O
– Weak Cache coherence
• Application requirements
– Impunity of slow system lifted
– Elasticity is the mantra
– The weakest link draws the whole system down
– Bottlenecks can be really really costly

5
Concurrency over the years
Java VersionJava Version FeaturesFeatures
JDK 1.0 ●Threads
●“synchronized” keyword
●HashTable(thread-safe but synchronized classes)
JDK 1.2 ●Collections framework(updated)
●un-synchronized classes bound by
synchronized accessors
●eg: Collections.synchronizedMap()
●Use when you need sync
JDK 1.5 ● concurrency package JSR166 java.util.concurrent
JDK 7 ● Updated java.util.concurrent package(JSR166y)
Fork-Join and Executor Framework
JDK 8 ● Modularity
● Lamda

6
Concurrency : Basics
Three elements of a concurrent Application

Atomicity

Certain pieces of an application must all be executed as one unit
●
Visibility

Changes that you make to a value to be visible precisely when you intend them to
be
●
Synchronization

Concurrent access to shared variables should be guarded
Terms you will come across
●
JMM – The specification that guarantees how the JVM should work
●
Monitor – Intrinsic lock (Every object has a monitor associated with it)
●
Race condition – An operation on shared resources whose outcome may be in-
deterministic
●


7
Concurrency : Myths
●
The more the threads, faster my application is
• A bigger system means faster execution
• Processors/compilers can optimize my code for
parallelism
• Locks make the program go slow
• Re-factoring is easy
• Writing concurrent application is a black-art

8
Common concurrency patterns
Dynamic Exclusion Through Implicit Locks
• a.k.a “synchronized”
• Lock obtained if method is declared synchronized
• Watch out for fine/coarse grained locks.
• Dos/Dont's:
– Avoid CPU/IO intensive operations inside “synchronized” block.
– “synchronize” on a code block and not method (if appropriate)
• Remember that JIT can re-order execution

9
Structural Exclusion Through Confinement
• Options:
– Thread confinement : Access object from a single thread using
thread-per-session approach
– Instance confinement : Use encapsulation techniques to restrict
access to object state from multiple threads
– Method confinement : Do not allow an object to escape from
method by referencing it through local variables
• Do's/Dont's:
– Confinement cannot be relied on unless a design is leak proof
– Combine confinement with appropriate locking discipline
– Use ThreadLocal variables to maintain Thread Confinement

10
Immutability
• Object state cannot change after construction
• Are automatically thread-safe
• Immutability is guranteed only when :
– All fields are declared final
– Object reference fields must not allow modifications anywhere in the object
graph reachable from the fields after construction
– The class should be declared final (to prevent a subclass from subverting
these rules)
• Immutable objects eliminate a bunch of issues around concurrency. Use whenever you
can.
• Have “failure atomicity” - upon an exception, state is never undesirable or
indeterministic.
"Classes should be immutable unless there's a very good reason to make them mutable....If a class cannot"Classes should be immutable unless there's a very good reason to make them mutable....If a class cannot
be made immutable, limit its mutability as much as possible." - Joshua Blochbe made immutable, limit its mutability as much as possible." - Joshua Bloch

11
Co-operation
• Wait-notify mechanisms
• One thread creates a condition and another one waits for it
• All threads are treated the same - when multiple threads are waiting,
no guarantee on which one would wake up
• Missed or early notifications
• Do's/Dont's:
– Use notifyAll() instead of notify()
– Wait() in a while loop
– Call wait() and notify() inside synchronized block

12
Deep Dive
• JSR166 - “java.util.concurrent” packages
• Introduced in JDK 5
• Motivation
– Creating APIs for commonly used functionality
– Application/scenario agnostic
– “volatile” guaranteed visibility but not atomicity
– Synchronization is expensive (and can cause dead-lock)
●
Atomic Variables, nanosecond timing, Condition variables, Lock classes,
tryLock, Queues, Barriers, Executors.....
●

13
Deep Dive – Atomic Classes
• java.util.concurrent.atomic.Atomic* classes
●
Provides a worthy substitute for “volatile” and “synchronized”
●
Available on Integer, Long, Boolean, Object Ref, Arrays
●
CompareAndSet(), incrementAndGet() etc guarantee atomicity

14
Deep Dive : Locks
ReentrantLock

“synchronization” on steroids

Supports Lock state, non-blocking tryLock() and interruptible locking

Throughput can be higher by order of magnitudes

Can be extended to set “fairness” and “Condition”







Use ONLY and ONLY if “synchronized” will not solve your problems

If your Application needs timed-locks, interruptible locks, multiple condition locks

15
Deep Dive : Executor Framework
●
Implemented via java.util.concurrent.ExecutorService
●
Provides flexible thread pool implementation
●
Producer-consumer pattern
– Producer : Creates work
– Consumer : Executes the work
●
●
●
●
●
Consider newCachedThreadPool if you don't want to bound no. of threads

16
Deep Dive : Executor Framework
●
Remember to shut it down
●
Asynchronous execution may hide hidden tasks
●

17
Deep Dive: Fork Join Framework
“Framework for supporting a style of programming in which problems are solved by recursively splitting them in
to sub tasks that are solved in parallel, waiting for them to complete and then composing results”
●
Introduced through JSR166y and part of Java 7
●
Introduces Parallel-Array (PA) concept to Java (simlar to Map-Reduce?)
●
Extends the executor framework for Recursive style problems
●
Makes use of work stealing algorithm
●
Transparent Parallelism requiring no tuning (almost!)
●
Suited for MMP Architectures.
●
Number of sub tasks can be altered at run-time

“Fork” Starts a new parallel fork/join subtask

“Join” causes current task not to proceed until the forked task is complete

This can happen recursively till tasks are smaller enough to be solved sequentially
Components of a FJ framework
●
The guy who does the work (business logic)
●
The guy who tells how the work can be split
●
The guy who coordinates it all (FJ)

18
There's more than one way to solve it
●
There's always more than one way to solve the problem
●
Know the pros/cons of each approach
●
Example :
Not Thread Safe

19
Option 1 : Using “volatile” keyword
Pros:
●
Simple, easy to write and understand
●
Would solve most of the requirement
Cons:
●
Thread looping
●
No option to know/set priority of waiting threads

20
Option 2 : Using “AtomicBoolean”
Pros:
●
Fairly straight-forward
●
“tryLock()” can prevent costly thread spins
●
Flexibility to set Condition and “fairness” of waiting threads
Cons:
●
Slight over-head if you need primitive data locking

21
Option 3 : Using “synchronized”
Pros:
●
Locks are intrinsic and so no need to release them explicitly
Cons
●
Code is a lot more verbose
●
Empty thread looping

22
Common concurrency mistakes
●
synchronization
– Fine/coarse grained synchronization
– Synchronizing just data
– Validating data integrity at one level
– Synchronizing on “null”
– Changing synchronizing object instance
– Synchronizing on string literals
– Writing with synchronization but reading w/o it
– Volatile array -
• “valarr” is a volatile reference to an array, but array elements are not

23
Common concurrency mistakes
●
Volatile will work only when
– write to variable does not depend on current value
– Write does not depend on any other non-atomic operation
●
CachedThreadPool is un-bound – You may run out of resources
sooner than you thought
●
●
The list never ends.....
●
Recommend go through “Java Platform Concurrency Gotchas”
- Alex Miller
●

24
Re-cap
 Relevance of concurrency
 Basics & Myths
 Concurrency Patterns
 Deep-dive
 There's more than one way to solve it!!
 Common gotchas
 Q&A

Concurrent Programming in Java

More Related Content

What's hot

Similar to Concurrent Programming in Java

Recently uploaded

Concurrent Programming in Java