The object pool pattern is a software creational design pattern that uses a set of initialized objects kept ready to use – a "pool" – rather than allocating and destroying them on demand. A client of the pool will request an object from the pool and perform operations on the returned object. When the client has finished, it returns the object to the pool rather than destroying it; this can be done manually or automatically.
Object pools are primarily used for performance: in some circumstances, object pools significantly improve performance. Object pools complicate object lifetime, as objects obtained from and returned to a pool are not actually created or destroyed at this time, and thus require care in implementation.
When it is necessary to work with numerous objects that are particularly expensive to instantiate and each object is only needed for a short period of time, the performance of an entire application may be adversely affected. An object pool design pattern may be deemed desirable in cases such as these.
The object pool design pattern creates a set of objects that may be reused. When a new object is needed, it is requested from the pool. If a previously prepared object is available, it is returned immediately, avoiding the instantiation cost. If no objects are present in the pool, a new item is created and returned. When the object has been used and is no longer needed, it is returned to the pool, allowing it to be used again in the future without repeating the computationally expensive instantiation process. It is important to note that once an object has been used and returned, existing references will become invalid.
In some object pools the resources are limited, so a maximum number of objects is specified. If this number is reached and a new item is requested, an exception may be thrown, or the thread will be blocked until an object is released back into the pool.
The object pool design pattern is used in several places in the standard classes of the .NET Framework. One example is the .NET Framework Data Provider for SQL Server. As SQL Server database connections can be slow to create, a pool of connections is maintained. Closing a connection does not actually relinquish the link to SQL Server. Instead, the connection is held in a pool, from which it can be retrieved when requesting a new connection. This substantially increases the speed of making connections.
Object pooling can offer a significant performance boost in situations where the cost of initializing a class instance is high and the rate of instantiation and destruction of a class is high – in this case objects can frequently be reused, and each reuse saves a significant amount of time. Object pooling requires resources – memory and possibly other resources, such as network sockets, and thus it is preferable that the number of instances in use at any one time is low, but this is not required.
The pooled object is obtained in predictable time when creation of the new objects (especially over network) may take variable time. These benefits are mostly true for objects that are expensive with respect to time, such as database connections, socket connections, threads and large graphic objects like fonts or bitmaps.
In other situations, simple object pooling (that hold no external resources, but only occupy memory) may not be efficient and could decrease performance.[1] In case of simple memory pooling, the slab allocation memory management technique is more suited, as the only goal is to minimize the cost of memory allocation and deallocation by reducing fragmentation.
Object pools can be implemented in an automated fashion in languages like C++ via smart pointers. In the constructor of the smart pointer, an object can be requested from the pool, and in the destructor of the smart pointer, the object can be released back to the pool. In garbage-collected languages, where there are no destructors (which are guaranteed to be called as part of a stack unwind), object pools must be implemented manually, by explicitly requesting an object from the factory and returning the object by calling a dispose method (as in the dispose pattern). Using a finalizer to do this is not a good idea, as there are usually no guarantees on when (or if) the finalizer will be run. Instead, "try ... finally" should be used to ensure that getting and releasing the object is exception-neutral.
Manual object pools are simple to implement, but harder to use, as they require manual memory management of pool objects.
Object pools employ one of three strategies to handle a request when there are no spare objects in the pool.
Care must be taken to ensure the state of the objects returned to the pool is reset back to a sensible state for the next use of the object, otherwise the object may be in a state unexpected by the client, which may cause it to fail. The pool is responsible for resetting the objects, not the clients. Object pools full of objects with dangerously stale state are sometimes called object cesspools and regarded as an anti-pattern.
Stale state may not always be an issue; it becomes dangerous when it causes the object to behave unexpectedly. For example, an object representing authentication details may fail if the "successfully authenticated" flag is not reset before it is reused, since it indicates that a user is authenticated (possibly as someone else) when they are not. However, failing to reset a value used only for debugging, such as the identity of the last authentication server used, may pose no issues.
Inadequate resetting of objects can cause information leaks. Objects containing confidential data (e.g. a user's credit card numbers) must be cleared before being passed to new clients, otherwise, the data may be disclosed to an unauthorized party.
If the pool is used by multiple threads, it may need the means to prevent parallel threads from trying to reuse the same object in parallel. This is not necessary if the pooled objects are immutable or otherwise thread-safe.
Some publications do not recommend using object pooling with certain languages, such as Java, especially for objects that only use memory and hold no external resources (such as connections to database). Opponents usually say that object allocation is relatively fast in modern languages with garbage collectors; while the operator new needs only ten instructions, the classic new - delete pair found in pooling designs requires hundreds of them as it does more complex work. Also, most garbage collectors scan "live" object references, and not the memory that these objects use for their content. This means that any number of "dead" objects without references can be discarded with little cost. In contrast, keeping a large number of "live" but unused objects increases the duration of garbage collection.
The following Go code initializes a resource pool of a specified size (concurrent initialization) to avoid resource race issues through channels, and in the case of an empty pool, sets timeout processing to prevent clients from waiting too long.
import ("errors" "log" "math/rand" "sync" "time")
const getResMaxTime = 3 * time.Second
var (ErrPoolNotExist = errors.New("pool not exist") ErrGetResTimeout = errors.New("get resource time out"))
//Resourcetype Resource struct
//NewResource Simulate slow resource initialization creation// (e.g., TCP connection, SSL symmetric key acquisition, auth authentication are time-consuming)func NewResource(id int) *Resource
//Do Simulation resources are time consuming and random consumption is 0~400msfunc (r *Resource) Do(workId int)
//Pool based on Go channel implementation, to avoid resource race state problemtype Pool chan *Resource
//New a resource pool of the specified size// Resources are created concurrently to save resource initialization timefunc New(size int) Pool
//GetResource based on channel, resource race state is avoided and resource acquisition timeout is set for empty poolfunc (p Pool) GetResource (r *Resource, err error)
//GiveBackResource returns resources to the resource poolfunc (p Pool) GiveBackResource(r *Resource) error
// package mainpackage main
import ("github.com/tkstorm/go-design/creational/object-pool/pool" "log" "sync")
func main
In the .NET Base Class Library there are a few objects that implement this pattern. System.Threading.ThreadPool is configured to have a predefined number of threads to allocate. When the threads are returned, they are available for another computation. Thus, one can use threads without paying the cost of creation and disposal of threads.
The following shows the basic code of the object pool design pattern implemented using C#. For brevity the properties of the classes are declared using C# 3.0 automatically implemented property syntax. These could be replaced with full property definitions for earlier versions of the language. Pool is shown as a static class, as it's unusual for multiple pools to be required. However, it's equally acceptable to use instance classes for object pools.
// The PooledObject class is the type that is expensive or slow to instantiate,// or that has limited availability, so is to be held in the object pool.public class PooledObject
// The Pool class controls access to the pooled objects. It maintains a list of available objects and a // collection of objects that have been obtained from the pool and are in use. The pool ensures that released objects // are returned to a suitable state, ready for reuse. public static class PoolIn the code above, the PooledObject has properties for the time it was created, and another, that can be modified by the client, that is reset when the PooledObject is released back to the pool. Shown is the clean-up process, on release of an object, ensuring it is in a valid state before it can be requested from the pool again.
Java supports thread pooling via java.util.concurrent.ExecutorService and other related classes. The executor service has a certain number of "basic" threads that are never discarded. If all threads are busy, the service allocates the allowed number of extra threads that are later discarded if not used for the certain expiration time. If no more threads are allowed, the tasks can be placed in the queue. Finally, if this queue may get too long, it can be configured to suspend the requesting thread.