Understand HTTP cache in 30 minutes

One 、 Reading guide

Whether you're the front end , Or the back end , You can't avoid meeting and in your daily work HTTP Cache related issues . For example, it is found that the file that should be updated is not updated because of the cache , The usual practice is to timestamp the file path of the requested resource , Simple, crude and effective , But as a aspiring programmer , We shouldn't just stop at solving the problem OK At the level of , Further more , Think about whether there is a better solution ？ For different scenes 、 Different types of resources , Is there a more efficient cache design ？ I hope that after reading this article , Be able to HTTP A deeper and more comprehensive understanding of caching , Master them , It will be helpful for us to solve some complex daily problems .

Two 、HTTP What is caching ？

Suppose there is no cache , Just imagine , When a client accesses an original server page multiple times , The server transmits the same document multiple times , One client at a time . Some of the same content will be transmitted over and over the network . These redundant data transmissions will exhaust expensive network bandwidth , Reduce the transmission speed , aggravate Web Server load .
With caching , You can keep a copy of the first server response , Subsequent requests can be handled by cached copies , This can reduce those inflows / Out of the original server 、 Wasted duplicate traffic . By reusing previously acquired resources , Can significantly improve the performance of websites and Applications . besides , Other collateral benefits are also obvious , For example, it can reduce the network request time 、 Reduce bandwidth consumption 、 Lifting performance .

Like what? HTTP The response is cached by the client ？

• By default , The request method is as follows: GET、HEAD The response content of is cacheable , In the case of including freshness information ,POST The response content can also be cached ;
• By default , Response code such as 200、206、300、301、302、404 And so on can be cached ;
• The response header and request header do not indicate that caching is not used , Such as Cache-Control: no-store.

3、 ... and 、 What are private cache and shared cache ？

- Private cache
A cache for one client only , That is, the cache on the client is for your own use only , It usually only exists on clients such as browsers .
The first request initiated by each client is processed by the source server . When the cache is in effect , Subsequent requests from the same client will not even be sent , Instead, the service is provided by the local cache .
- Shared cache
A cache that can be used by multiple clients , Usually rely on proxy servers .
The first request initiated by the client accesses the source server through the proxy server , After the cache takes effect, it will be stored in the proxy server , The same request initiated by subsequent clients , The proxy server provides caching services , Shared caching can reduce the pressure on the source server .

The content related to the concept is finished , We're starting to go deeper HTTP The specific working mechanism of cache .

Four 、HTTP Cache processing flow

Before the official start , Let's understand from a macro perspective through the following picture HTTP Cache processing flow （ Execution order ）.
Please study this picture carefully , It is very important , Many of the contents below are extended explanations for a link in the figure .
For websites , Caching is an important part of high performance , Cache needs to be configured properly , Because not all resources are permanent .Cache-Control The header can control the cache ,Cache-Control Can be used for HTTP Request and response , Support multiple instructions , Separated by commas ：

Cache-Control Several instructions are particularly confusing , You can't take meaning from words . such as no-cache, It doesn't mean you can't use cache, The client will still send the with no-cache Cache your response , Just don't use the cache directly every time , But first go to the server to verify , So actually no-cache The right name is must-revalidate. If you want the client to not cache the response at all , Should use the no-store, with no-store The response will not be cached to any disk or memory , It's the real no-cache

• How to detect whether the cache is fresh ？
  Service life ：Cache-Control: max-age= second
  We can specify the maximum lifetime of a cache , Relative to cache creation time , If the maximum service life is exceeded , It means that the cache is no longer fresh

Cache-Control: max-age= second 

Take a chestnut ：

As shown in the figure , When in the response header Cache-Control Of max-age Set to 10 seconds , It means starting with the first request , The cache validity of this resource is 10 second ,10 Requesting the resource again within seconds will read from the cache ; exceed 10 second , Then the client sends a request to the source server , The validity of the cache starts counting again

• Expiration date ：Expires
  We can also specify an absolute expiration date , If the expiration date has passed , It means that the cache is no longer fresh

Expires:  Expiration date 

Expires yes HTTP/1.0 The first ,Cache-Control yes HTTP/1.1 The first ,Expires The first and Cache-Control:max-age The first thing I did was essentially the same , But because of Cache-Control The first part uses relative time instead of absolute date , So I prefer to use newer Cache-Control The first one . The absolute date depends on the correct setting of the computer clock

5、 ... and 、HTTP Server revalidation

• What is server-side revalidation ？
  Just because a cached document has expired doesn't mean it's actually different from the document on the source server , This just means checking with the server , Note that the cache needs to ask the source server whether the document has changed , This situation is called “ Server revalidation ”
  There are two situations for server-side revalidation ：
• The server documentation has changed ： The cache will get a new copy of the document , And store it in the location of the old document , Then send the document to the client ;
• The server-side document has not changed ： The cache just needs to get the new header , Include a new expiration date , And update the first part of the cache , The document can still be used
  Caching does not have to verify the validity of the document for each request —— It only needs to be revalidated with the server when the document expires
• How to revalidate the server ？
  Modify the date and then verify ：If-Modified-Since： The most common cache revalidation header is If-Modified-Since

If-Modified-Since:  Date last modified 

It means , Only when resources have changed since a certain date , The server will execute the request ：

• If after the specified date , The document has been modified ,If-Modified-Since The condition is true , Usually the request will execute successfully . The document with the new header will be returned to the cache , The first new one, among other things , It also includes a new expiration date
• If after the specified date , The document has not been modified , Conditions are false , The server will return a 304 NotModified response message , The body of the document will not be returned
• If-Modified-Since The value of is the last returned by the server last-modified Value . If the content is modified during this period , The last modification date will be different , The source server will send back the new document . otherwise , The server will think that the last modification date of the cache matches the current last modification date of the server document , Will return a 304 NotModified Respond to
  
  In some cases, it is not enough to use only the last modification date for revalidation , such as ：
• Some documents may be rewritten periodically （ such as , Write... From a background process ）, But the actual data is often the same . Although the content has not changed , But the date of revision will change
• Some documents may have been modified , But the changes don't matter , There's no need to have the world's caches reload data （ Such as changes in spelling or notes ）
• Some servers can't accurately determine the last modification date of their pages
• Some servers provide documents that change in milliseconds （ such as , Real time monitor ）, For these servers , A one second modification date may not be enough

To solve these problems ,HTTP Allow users to label entities as they are called （ETag） Of “ Version identifier ” Compare

• If there is an entity tag in the cache ETag, It will revalidate with the source server , If the entity tag on the server still matches it , Will return a message 304 NotModified Respond to . If the entity tag on the server has changed , The server will be in a 200 OK New content and corresponding new... Are returned in the response Etag

• If-None-Match The value of is the last returned by the server ETag Value . stay ETag Add... Before the value of W/ Prefix representation can adopt relatively loose algorithm

• When and when to use the validation tag again ？

• If the server returns a ETag The first one , The client must use If-None-Match Entity tag revalidation ;

• If the server returns a Last-Modified The first one , The client can use it If-Modified-Since Modify the date and then verify ;

• If ETag and Last-Modified provide , The client should use both revalidation schemes at the same time
  If HTTP/1.1 The request received by the cache or server has both If-Modified-Since, And with If-None-Match, So only when both of these conditions are met , Before returning 304 Not Modified Respond to

• ending

• Thank you for reading this patiently ,HTTP Caching is not very simple ？ Put this to use , If you have mastered all the above , Cache related problems encountered in daily work , I believe you can handle it easily

• For special situations at work , How to make an efficient cache design , I believe there is an answer . Limited to space ,HTTP Some knowledge of caching is not exhaustive , You can find more detailed references on the Internet to learn them —— When you need them