WhizzBee
system architecture
Each
product in the WhizzBee Family 2.0 works by seamlessly integrating
the computing resources of independent server nodes into a powerful
and scalable server system. The whole unit is accessible to
the outside world by a single cluster Virtual IP (VIP) address
and hostname. The following figure illustrates the system architecture
of a typical WhizzBee cluster (note: this architecture
is applicable to all four products in the WhizzBee Family 2.0).
As
shown in the figure, a WhizzBee cluster consists of the following
four components.
Dispatcher
The
dispatcher accepts incoming HTTP requests, and dispatches the
requests to the web server nodes according to some distribution
policy using an efficient algorithm.
Backup
Dispatcher
The
backup dispatcher keeps monitoring the healthiness of the dispatcher.
Whenever the dispatcher is found offline due to software or
hardware failure, the backup dispatcher will promote itself
to a normal dispatcher so as to maintain the web service. Installation
of a backup dispatcher is optional.
Server
Node(s)
A
server node is responsible for receiving and processing the
HTTP requests dispatched from the dispatcher, and delivering
the resulting web contents to the clients directly through the
network. A cluster may have one or more server nodes. Obviously,
more server nodes mean higher performance and server throughput.
Management
Console
The
management console can be a dedicated machine for monitoring
or it can be one of the cluster nodes other than server nodes.
It is recommended, however, that a dedicated machine be used.
Administrators can connect to the web server running at the
management console to perform cluster administrative tasks remotely.
It
should be noted that the dispatcher, backup dispatcher, and
all server nodes are regarded as cluster nodes, while
the management console is a separate entity that helps
monitor and manage the whole cluster.
