|
Load balancing is a technique used to spread a network service workload
between two or more devices. Benefits include scalability, reliability,
efficiency, redundancy, and minimized response time.
Load balancing can be achieved either by running an application on a server
(software load balancing) or by using a special purpose device (hardware
load balancing). Software load balancing applications use a server's CPU to
process requests. A hardware load balancer uses a processor made
specifically for this purpose to handle more requests. NewServers™ includes
hardware load balancing free for all account with two or more servers.
The load balancer brokers connections and distributes the load between the
client and the servers. The server that receives the connection is chosen
using a preset algorithm. The packets that make up the client request are
translated by the load balancer (using a process called NAT) before being
sent to the server. The client and the server are equally unaware of the
load balancer.
Click for how to setup a server farm in your NewServers™ account.
Services commonly load balanced include:
HTTP
HTTPS (SSL)
SMTP
IMAP
POP3
NTP
DNS
TFTP
Algorithms used to select which server will receive a request:
Round Robin
A common method. Requests are distributed to each server evenly.
Weighted Round Robin
Each device is assigned a weight. Servers capable of handling more requests
will receive more.
Least Connections
This method hands off a new connection to the device that currently has the
least amount of connections.
Response Time
The load balancer keeps a list of servers and their response times. When a
new connection is made the load balancer directs the request to the device
with the best response time.
Session Persistence
Each of the above algorithms above can be used with session persistence.
After the normal algorithm is used to select a server each subsequent
request from the same client will be sent to the same server.
Next » Set up a web server |
 |
