Link load balancers smooth Web traffic
By Jason Needham

As organizations move to the Internet to deliver their business-critical applications, maintaining only one link to the public network represents a single point of failure. But when an organization relies on one link, it often oversubscribes on bandwidth and might even select a more expensive ISP in an attempt to mitigate its risk or achieve greater bandwidth scalability.

The term "multihoming" describes a network that utilizes multiple connections to the Internet. Provisioning multiple network connections has become a critical part of any organization building a reliable, fault-tolerant data center.

Many administrators have tried to remedy part of this traffic management problem by using Border Gateway Protocol (BGP) to route traffic across links. BGP was designed to let IP routers direct how packets traverse the Internet from Point A to Point B. However, deploying a multihoming solution via BGP is costly and extremely complex, and requires ISP cooperation.

And so, the key to multihoming is not just obtaining diversified links and ISPs, but optimally managing traffic over those connections. To direct traffic down the "best path," organizations need a solution that dynamically manages traffic across many Internet or intranet access points.

Link load balancers automatically monitor line availability and performance to manage bidirectional IP traffic flows to a site without using BGP - providing fault-tolerant Internet access regardless of connection type or provider. By moving intelligent switching functionality to the edge of an enterprise network, these products place a new level of control in the hands of administrators, letting them dynamically switch traffic based on ISP availability, capacity, performance, price or other set distribution policies.

Using link load balancing provides organizations a way to achieve:

• Automatic ISP failover in less than 1 minute.
• Even traffic distribution that doesn't overburden any connection.
• Better user performance by sending traffic over the fastest link.
• Reduced connectivity cost by intelligently managing around high-priced traffic spikes.
• Increased bandwidth scalability and ISP choice.

Link load balancers typically are deployed inline between gateway routers and firewalls. To monitor the status of WAN connectivity, they conduct a series of transparent health and performance checks, evaluating the quality and reliability for each ISP line. Armed with this information, link load balancers intercept traffic flowing in and out of the data center and automatically switch users to the preferred WAN connections.

Where to send traffic is determined via advanced algorithms that take into account things such as bandwidth utilization, client and end-user round-trip times, packet loss and the price an organization pays for its connectivity. Administrators can set these policies and define how the traffic should be directed to providers to best utilize their bandwidth investments.

To handle external clients that access a data center and corporate clients that access the Internet, these products use different methods for directing inbound and outbound traffic flows.

Providing high availability for corporate users going out to the Internet, a link load balancer switches traffic via intelligent Secure Network Address Translation (SNAT). SNAT provides a secure mechanism for translating internal, nonroutable addresses into routable addresses. As traffic flows out of a data center, the gateway and source address of IP packets are translated and switched to the appropriate upstream gateway router. This ensures that traffic is sent and returned through the desired path.

For high availability of Web services and applications hosted at a site, a link load balancer uses intelligent DNS to direct external users to the preferred ISP link. DNS translates common names, such as www.company.com, into IP addresses. When a client accesses a site, a DNS request is made to the link load balancer asking for the IP address to find the desired resources. The device then responds with the IP address that will direct the user to the appropriate ISP for that data center. If a failed or poor-performing link is found, clients are switched over to other providers to ensure the best possible service.

This document is provided for informational purposes only. The information contained represents the current view of the author on issues discussed as of the date of publication. Because TheManageMentor must respond to changes in market conditions, it should not be interpreted to be a commitment on the part of TheManageMentor or the content author. TheManageMentor cannot guarantee the accuracy of any information presented after the date of publication. The user assumes the entire risk as to the accuracy and the use of this document.

Information Provided In This Document Is Provided "As Is" Without Warranty Of Any Kind, Either Express Or Implied, Including But Not Limited To The Implied Warranties Of Merchantability, Fitness For A Particular Purpose And Freedom From Infringement. This document may not be copied or distributed. All trademarks acknowledged.