High concurrency of the ethereal

Often see high concurrency problems, but high concurrency is the last thing to consider. Why, he is a disembodied, few sites really need these things, and many of these techniques, in fact, you are already in use. It is enough to have this consciousness, and there is no need to stare at the problem at all times. Only a few sites can really reach high concurrency.

A simple induction, from low cost, high performance and high expansion of the point of view, there are the following treatment solutions:
1. Static HTML
2, Image server separation
3. Database cluster and library table hash
4. Cache
5. Mirror
6, load balancing; A typical use of load balancing strategy is to build a squid cluster on the basis of software or hardware four-layer switching, which is adopted on many large Web sites including search engines, with low cost, high performance and strong extensibility, and it is easy to add or subtract nodes to the architecture at any time.

The following is also a summary of the cattle, the same as the above section.
High concurrency, performance bottlenecks and current common countermeasures


1. Database bottlenecks. MySQL Concurrent link 100

2.apache Concurrent Links 1500

3. Program execution efficiency



1. There is a database bottleneck, the current processing scheme is not subordinate to master, cluster. Add cache (memcached).

such as: Mobile Home new system introduction and architecture sharing (http://www.slideshare.net/Fenng/ss-1218991?from=ss_embed)

is to optimize at the cache layer.

Another Racket network architecture (http://www.bopor.com/?p=652)

is to increase the database, sub-table sub-library method to solve.

Sina adds MQ (Message Queuing) to distribute the data.

And the wind station used the Key-value database. In fact, this can be understood as a persistent cache.



2.apache bottleneck.

Increase the server. Load Balancing. such as Sina's F5

Due to the limit of the number of processes. Will move some basic immutable code out and put it on a separate server. such as css/js/pictures.

The domestic success story is Tom's CDN


And such as Nginx turned out and squid of the reverse proxy are based on this reason.


3.php of execution efficiency. There are multiple reasons.

1). Its efficiency is low.

The solution to the success story is the hiphop of Zend Optimizer and Facebooke

Taobao is to compile the PHP code into modules to solve the efficiency problem.

2). Database query efficiency issues. If there may be an order by, the group by and other SQL data issues.

This should actually be attributed to the database design problem.


The solution is to establish the correct index. Add Memcache.

Search services for like tables with dedicated sphinx. and lucence.

Programmers should use explain to analyze SQL statements.

High concurrency of the ethereal