Tuesday, September 23, 2014

HAproxy Load Balancing Algorithms

HAproxy Load Balancing Algorithms
The algoritm you define determines how HAproxy balances load across your servers. You can set the algorithm to use with the balance parameter.

Round Robin
Requests are rotated among the servers in the backend.

Servers declared in the backend section also accept a weight parameter which specifies their relative weight. When balancing load, the Round Robin algorithm will respect that weight ratio.
Example:

...
option tcplog
balance roundrobin
maxconn 10000
...
Static Round Robin
Each server is used in turn, according to the defined weight for the server. This algorithm is a static version of the round-robin algoritm, which means that changing the weight ratio for a server on the fly will have no effect. However, you can define as many servers as you like with this algorithm. In addition, when a server comes online, this algoritm ensures that the server is immediately reintroduced into the farm after re-computing the full map. This algoritm also consome slightly less CPU cycles (around -1%).

Example:

...
option tcplog
balance static-rr
maxconn 10000
...
Least Connection
Each server is used in turn, according to the defined weight for the server. This algorithm is a static version of the round-robin algoritm, which means that changing the weight ratio for a server on the fly will have no effect. However, you can define as many servers as you like with this algorithm. In addition, when a server comes online, this algoritm ensures that the server is immediately reintroduced into the farm after re-computing the full map. This algoritm also consome slightly less CPU cycles than the Round Robin algorithm (around -1%).

Example:

...
option tcplog
balance leastconn
maxconn 10000
...
Source
A hash of the source IP is divided by the total weight of the running servers to determine which server will receive the request. This ensures that clients from the same IP address always hit the same server, which is a poor man's session persistence solution.

Example:

...
option tcplog
balance source
maxconn 10000
...
URI
This algorithm hashes either the left part of the URI (before the question mark) or the whole URI (if the whole parameter is present) and divides the hash value by the total weight of the running servers. The result designates which server will receive the request. This ensures that the proxy will always direct the same URI to the same server as long as all servers remain online.

This is used with proxy caches and anti-virus proxies in order to maximize the cache hit rate. This algorithm is static by default, which means that changing a server's weight on the fly will have no effect. However, you can change this using a hash-type parameter.

You can only use this algorithm for a configuration with an HTTP backend.
Exampple:

...
option tcplog
balance uri
maxconn 10000
...
URL Parameter
The URL parameter specified in argument will be looked up in the query string of each HTTP GET request.

You can use this algorithm to check specific parts of the URL, such as values sent through POST requests. For example, you can set this algorithm to direct a request that specifies a user_id with a specific value to the same server using the url_param method. Essentially, this is another way of achieving session persistence in some cases (see the official HAproxy documentation for more information).

Example:

...
option tcplog
balance url_param userid
maxconn 10000
...
or

...
option tcplog
balance url_param session_id check_post 64
maxconn 10000
...

No comments:

Post a Comment