Apache HTTP Server Version 2.2
Description: | mod_proxy extension for load balancing |
---|---|
Status: | Extension |
Module Identifier: | proxy_balancer_module |
Source File: | proxy_balancer.c |
Compatibility: | Available in version 2.1 and later |
This module requires the service of mod_proxy
. It provides load balancing support for
HTTP
, FTP
and AJP13
protocols
Thus, in order to get the ability of load balancing,
mod_proxy
and mod_proxy_balancer
have to be present in the server.
Do not enable proxying until you have secured your server. Open proxy servers are dangerous both to your network and to the Internet at large.
This module provides no directives.
At present, there are 2 load balancer scheduler algorithms available
for use: Request Counting and Weighted Traffic Counting. These are controlled
via the lbmethod
value of the Balancer definition. See
the ProxyPass
directive for
more information.
Before we dive into the technical details, here's an example of
how you might use mod_proxy_balancer
to provide
load balancing between two back-end servers:
<Proxy balancer://mycluster>
BalancerMember http://192.168.1.50:80
BalancerMember http://192.168.1.51:80
</Proxy>
ProxyPass /test balancer://mycluster/
Enabled via lbmethod=byrequests
, the idea behind this
scheduler is that we distribute the requests among the
various workers to ensure that each gets their configured share
of the number of requests. It works as follows:
lbfactor is how much we expect this worker to work, or the workers's work quota. This is a normalized value representing their "share" of the amount of work to be done.
lbstatus is how urgent this worker has to work to fulfill its quota of work.
The worker is a member of the load balancer, usually a remote host serving one of the supported protocols.
We distribute each worker's work quota to the worker, and then look which of them needs to work most urgently (biggest lbstatus). This worker is then selected for work, and its lbstatus reduced by the total work quota we distributed to all workers. Thus the sum of all lbstatus does not change(*) and we distribute the requests as desired.
If some workers are disabled, the others will still be scheduled correctly.
for each worker in workers
worker lbstatus += worker lbfactor
total factor += worker lbfactor
if worker lbstatus > candidate lbstatus
candidate = worker
candidate lbstatus -= total factor
If a balancer is configured as follows:
worker | a | b | c | d |
---|---|---|---|---|
lbfactor | 25 | 25 | 25 | 25 |
lbstatus | 0 | 0 | 0 | 0 |
And b gets disabled, the following schedule is produced:
worker | a | b | c | d |
---|---|---|---|---|
lbstatus | -50 | 0 | 25 | 25 |
lbstatus | -25 | 0 | -25 | 50 |
lbstatus | 0 | 0 | 0 | 0 |
(repeat) |
That is it schedules: a c d a c d a c d ... Please note that:
worker | a | b | c | d |
---|---|---|---|---|
lbfactor | 25 | 25 | 25 | 25 |
Has the exact same behavior as:
worker | a | b | c | d |
---|---|---|---|---|
lbfactor | 1 | 1 | 1 | 1 |
This is because all values of lbfactor are normalized with respect to the others. For:
worker | a | b | c |
---|---|---|---|
lbfactor | 1 | 4 | 1 |
worker b will, on average, get 4 times the requests that a and c will.
The following asymmetric configuration works as one would expect:
worker | a | b |
---|---|---|
lbfactor | 70 | 30 |
lbstatus | -30 | 30 |
lbstatus | 40 | -40 |
lbstatus | 10 | -10 |
lbstatus | -20 | 20 |
lbstatus | -50 | 50 |
lbstatus | 20 | -20 |
lbstatus | -10 | 10 |
lbstatus | -40 | 40 |
lbstatus | 30 | -30 |
lbstatus | 0 | 0 |
(repeat) |
That is after 10 schedules, the schedule repeats and 7 a are selected with 3 b interspersed.
Enabled via lbmethod=bytraffic
, the idea behind this
scheduler is very similar to the Request Counting method, with
the following changes:
lbfactor is how much traffic, in bytes, we want this worker to handle. This is also a normalized value representing their "share" of the amount of work to be done, but instead of simply counting the number of requests, we take into account the amount of traffic this worker has seen.
If a balancer is configured as follows:
worker | a | b | c |
---|---|---|---|
lbfactor | 1 | 2 | 1 |
Then we mean that we want b to process twice the amount of bytes than a or c should. It does not necessarily mean that b would handle twice as many requests, but it would process twice the I/O. Thus, the size of the request and response are applied to the weighting and selection algorithm.
At present there are 6 environment variables exported:
This is assigned the stickysession value used in the current request. It is the cookie or parameter name used for sticky sessions
This is assigned the route parsed from the current request.
This is assigned the name of the balancer used for the current
request. The value is something like balancer://foo
.
This is assigned the name of the worker used for the current request.
The value is something like http://hostA:1234
.
This is assigned the route of the worker that will be used for the current request.
This is set to 1 if the session route does not match the worker route (BALANCER_SESSION_ROUTE != BALANCER_WORKER_ROUTE) or the session does not yet have an established route. This can be used to determine when/if the client needs to be sent an updated route when sticky sessions are used.
This module requires the service of
mod_status
.
Balancer manager enables dynamic update of balancer
members. You can use balancer manager to change the balance
factor or a particular member, or put it in the off line
mode.
Thus, in order to get the ability of load balancer management,
mod_status
and mod_proxy_balancer
have to be present in the server.
To enable load balancer management for browsers from the example.com
domain add this code to your httpd.conf
configuration file
<Location /balancer-manager>
SetHandler balancer-manager
Order Deny,Allow
Deny from all
Allow from .example.com
</Location>
You can now access load balancer manager by using a Web browser
to access the page
http://your.server.name/balancer-manager