Thinking > story
How Does Caching Work in AFNetworking? : AFImageCache & NSUrlCache Explained

If you are an iOS developer using Mattt Thompson's 'delightful networking framework' AFNetworking (and if you aren't, what are you waiting for?), perhaps you have been been curious or confused about the caching mechanism employed and how you can tweak it to your advantage.

AFNetworking actually takes advantage of 2 separate caching mechanisms:

  • AFImagecache: a memory-only image cache private to AFNetworking, subclassed off of NSCache

  • NSURLCache: NSURLConnection's default URL caching mechanism, used to store NSURLResponse objects : an in-memory cache by default, configurable as an on-disk persistent cache

In order to understand how each caching system works, let's look at how they are defined:

How AFImageCache Works

AFImageCache is a part of the UIImageView+AFNetworking category. It is a subclass of NSCache, storing UIImage objects with a URL string as its key (obtained from an input NSURLRequest object).

AFImageCache definition:

@interface AFImageCache : NSCache <AFImageCache>

// singleton instantiation :

+ (id <AFImageCache>)sharedImageCache {
    static AFImageCache *_af_defaultImageCache = nil;
    static dispatch_once_t oncePredicate;
    dispatch_once(&oncePredicate, ^{
        _af_defaultImageCache = [[AFImageCache alloc] init];

// clears out cache on memory warning :

    [[NSNotificationCenter defaultCenter] addObserverForName:UIApplicationDidReceiveMemoryWarningNotification object:nil queue:[NSOperationQueue mainQueue] usingBlock:^(NSNotification * __unused notification) {
        [_af_defaultImageCache removeAllObjects];

// key from [[NSURLRequest URL] absoluteString] :

static inline NSString * AFImageCacheKeyFromURLRequest(NSURLRequest *request) {
    return [[request URL] absoluteString];

@implementation AFImageCache

// write to cache if proper policy on NSURLRequest :

- (UIImage *)cachedImageForRequest:(NSURLRequest *)request {
    switch ([request cachePolicy]) {
        case NSURLRequestReloadIgnoringCacheData:
        case NSURLRequestReloadIgnoringLocalAndRemoteCacheData:
            return nil;

    return [self objectForKey:AFImageCacheKeyFromURLRequest(request)];

// read from cache :

- (void)cacheImage:(UIImage *)image
        forRequest:(NSURLRequest *)request {
    if (image && request) {
        [self setObject:image forKey:AFImageCacheKeyFromURLRequest(request)];

AFImageCache is a private implementation of NSCache. There is no customization that you can do outside of editing the implementation in the the UIImageView+AFNetworking category, directly. It stores all accessed UIImage objects into its NSCache. The NSCache controls when the UIImage objects are released. If you wish to observe when images are released, you can implement NSCacheDelegate's cache:willEvictObject method.

Edit (03.14.14) : Mattt Thompson has gratiously informed me that as of AFNetworking 2.1, AFImageCache is configurable. There is now a public setSharedImageCache method. Here's the full AFN 2.2.1 UIImageView+AFNetworking specification.

How NSURLCache Works

Since AFNetworking uses NSURLConnection, it takes advantage of its native caching mechanism, NSURLCache. NSURLCache caches NSURLResponse objects returned by server calls via NSURLConnection.

Enabled by Default, but Needs a Hand

An NSURLCache sharedCache is enabled by default and will be used by any NSURLConnection objects fetching URL contents for you.

Unfortunately, it has a tendency to hog memory and does not write to disk in its default configuration. To tame the beast and potentially add some persistance, you can simply declare a shared NSURLCache in your app delegate like so:

NSURLCache *sharedCache = [[NSURLCache alloc] initWithMemoryCapacity:2 * 1024 * 1024
                                              diskCapacity:100 * 1024 * 1024
[NSURLCache setSharedURLCache:sharedCache];

Here we declare a shared NSURLCache with 2mb of memory and 100mb of disk space

Setting the Cache Policy on NSURLRequest Objects

NSURLCache will respect the caching policy (NSURLRequestCachePolicy) of each NSURLRequest object. The policies are defined as follows :

  • NSURLRequestUseProtocolCachePolicy: specifies that the caching logic defined in the protocol implementation, if any, is used for a particular URL load request. This is the default policy for URL load requests

  • NSURLRequestReloadIgnoringLocalCacheData: ignore the local cache, reload from source

  • NSURLRequestReloadIgnoringLocalAndRemoteCacheData: ignore local & remote caches, reload from source

  • NSURLRequestReturnCacheDataElseLoad: load from cache, else go to source.

  • NSURLRequestReturnCacheDataDontLoad: offline mode, load cache data regardless of expiration, do not go to source

  • NSURLRequestReloadRevalidatingCacheData: existing cache data may be used provided the origin source confirms its validity, otherwise the URL is loaded from the origin source.

Caching to Disk with NSURLCache

Cache-Control HTTP Header

Either the Cache-Control header or the Expires header MUST be in the HTTP response header from the server in order for the client to cache it (with the existence of the Cache-Control header taking precedence over the Expires header). This is a huge gotcha to watch out for. Cache Control can have parameters defined such as max-age (how long to cache before updating response), public / private access, or no-cache (don't cache response). Here is a good introduction to HTTP cache headers.

Subclass NSURLCache for Ultimate Control

If you would like to bypass the requirement for a Cache-Control HTTP header and want to define your own rules for writing and reading the NSURLCache given an NSURLResponse object, you can subclass NSURLCache.

Here is an example that uses a CACHE_EXPIRES value to judge how long to hold on to the cached response before going back to the source:

(Thanks to Mattt Thompson for the feedback and code edits!)

@interface CustomURLCache : NSURLCache
static NSString * const CustomURLCacheExpirationKey = @"CustomURLCacheExpiration";
static NSTimeInterval const CustomURLCacheExpirationInterval = 600;
@implementation CustomURLCache
+ (instancetype)standardURLCache {
    static CustomURLCache *_standardURLCache = nil;
    static dispatch_once_t onceToken;
    dispatch_once(&onceToken, ^{
        _standardURLCache = [[CustomURLCache alloc]
                                 initWithMemoryCapacity:(2 * 1024 * 1024)
                                 diskCapacity:(100 * 1024 * 1024)
    return _standardURLCache;
#pragma mark - NSURLCache
- (NSCachedURLResponse *)cachedResponseForRequest:(NSURLRequest *)request {
    NSCachedURLResponse *cachedResponse = [super cachedResponseForRequest:request];
    if (cachedResponse) {
        NSDate* cacheDate = cachedResponse.userInfo[CustomURLCacheExpirationKey];
        NSDate* cacheExpirationDate = [cacheDate dateByAddingTimeInterval:CustomURLCacheExpirationInterval];
        if ([cacheExpirationDate compare:[NSDate date]] == NSOrderedAscending) {
            [self removeCachedResponseForRequest:request];
            return nil;
    return cachedResponse;
- (void)storeCachedResponse:(NSCachedURLResponse *)cachedResponse 
                 forRequest:(NSURLRequest *)request
    NSMutableDictionary *userInfo = [NSMutableDictionary dictionaryWithDictionary:cachedResponse.userInfo];
    userInfo[CustomURLCacheExpirationKey] = [NSDate date];
    NSCachedURLResponse *modifiedCachedResponse = [[NSCachedURLResponse alloc] initWithResponse:cachedResponse.response userInfo:userInfo storagePolicy:cachedResponse.storagePolicy];
    [super storeCachedResponse:modifiedCachedResponse forRequest:request];

Now that you have your NSURLCache subclass, don't forget to initialize it in your AppDelegate in order to use it :

CustomURLCache *URLCache = [[CustomURLCache alloc] initWithMemoryCapacity:2 * 1024 * 1024
                                                   diskCapacity:100 * 1024 * 1024
[NSURLCache setSharedURLCache:URLCache];

Overriding the NSURLResponse before caching

The -connection:willCacheResponse delegate is a place to intercept and edit the NSURLCachedResponse object created by NSURLConnection before it is cached. In order to edit the NSURLCachedResponse, return an edited mutable copy as follows (code from NSHipster blog):

- (NSCachedURLResponse *)connection:(NSURLConnection *)connection
                  willCacheResponse:(NSCachedURLResponse *)cachedResponse {
    NSMutableDictionary *mutableUserInfo = [[cachedResponse userInfo] mutableCopy];
    NSMutableData *mutableData = [[cachedResponse data] mutableCopy];
    NSURLCacheStoragePolicy storagePolicy = NSURLCacheStorageAllowedInMemoryOnly;

    // ...

    return [[NSCachedURLResponse alloc] initWithResponse:[cachedResponse response]

// If you do not wish to cache the NSURLCachedResponse, just return nil from the delegate function:

- (NSCachedURLResponse *)connection:(NSURLConnection *)connection
                  willCacheResponse:(NSCachedURLResponse *)cachedResponse {
    return nil;

Disabling NSURLCache

Don't want to use the NSURLCache? Not Impressed? That's okay. To disable the NSURLCache, simply zero out memory and disk space in the shared NSURLCache definition in your appDelegate:

NSURLCache *sharedCache = [[NSURLCache alloc] initWithMemoryCapacity:0
[NSURLCache setSharedURLCache:sharedCache];


I wanted to write this blog post for the benefit of the iOS community, to summarize all of the information I found dealing with caching releated to AFNetworking. We had an internal app loading a lot of images that had some memory issues and performance problems. I was tasked with trying to diagnose the caching behavior of the app. During this exercise, I discovered the information on this post through scouring the web and doing plenty of debugging and logging. It is my hope that this post summarizes my findings and provides an opportunity for others with AFNetworking experience to add additional information. I hope that you have found this helpful.

Recent Posts