Nomadix Access Gateway, Руководство пользователя

Страница: 43 / 392

A
CCESS
G
ATEWAY
Introduction 31
Load Balancing and Link Failover
The NSE supports individual configuration of multiple WANs on an Access Gateway
(supported on AG2400, AG5600, and AG5800 hardware). Hotels can use this capability in a
number of ways, including load balancing, failure protection, and subscriber allocation.
This section provides use cases and scenarios to help you consider the full advantage of these
capabilities.
Definitions and Concepts
Load Balancing
Load balancing refers to the general process of balancing user traffic across multiple ISP
connections. All load-balancing appliances, as well as the Nomadix NSE, support load
balancing.
Link Aggregation
Link aggregation refers to the process of connecting multiple ISP connections to an appliance
and having the sum of all of the ISP bandwidth available to be shared across all users.
However, one individual connection is limited to the speed of the ISP connection that is
currently being used. For example, a hotel may aggregate 5 x 1.5Mbps DSL connections
together. This means that a total of 7.5Mbps of bandwidth is available to be shared across all
users, but a single user can receive a maximum of 1.5Mbps. All load-balancing appliances, as
well as the Nomadix NSE, support link aggregation. In most cases, link aggregation and load
balancing is effectively the same thing.
Link Failover
Link failover (sometimes referred to ISP redundancy) is the process of providing a second (or
occasionally a third or more) ISP link as a back up to the primary ISP link. In the event that the
primary link fails, all traffic is re-routed to the backup link, until such time as the primary link
becomes available.
Combined Load Balancing and Link Failover
This is the process where both load balancing and link failover are combined together. It
represents the best of both worlds. Where multiple ISP links are used in load balancing mode,
in the event that one or more links fail, all traffic is automatically rerouted to the remaining
surviving links. When the failed links recover, new connections are routed toward these until
the normal balanced configuration is reached.