Windows Server 2003 NLB OverviewWhen a single Web Server machine isn’t enough to handle the traffic on your Web site it’s time to look into building a Web Farm that uses multiple machines on the network acting as a single server. In this article Rick looks at the Windows Load Balancing Service and the new interface it sports in Windows Server 2003, which makes creating a Web Farm quick and easy and – gasp –even an affordable solution.
With the release of Windows Server 2003 Network Load Balancing has become a much more visible tool as a part of the operating system, providing a very usable and relatively easy to configure interface that makes it easy to build a Web Farm. The Network Load Balancing Service has been around in one incarnation or another since Windows NT SP4, but Windows Server 2003 is the first operating system that brings this service into the forefront as a main component of the OS. A new Network Load Balancing Manager application is now directly available from the Adminstrative Tasks menu and it’s powerful enough to allow to configure the entire cluster from a single console. The service is now available for all products in the Windows Server family including the lower end Web Edition which means that you now have a much more affordable solution to create Web Farms at your disposal. Just add servers please.
In this article I’ll review the basics of a Load Balancing service and then show you how to set up configure a basic installation using two machines.
Web Farms for city folk – do you need it?
A Web Farm is a not so fancy term for a collection of servers that act as a single Web server. The process behind the scenes maps a ‘virtual’ IP address to multiple machines. Software such as the Network Load Balancing Service or hardware like a specialized router or Load Balancer then deals with dishing up requests to the appropriate machine in the server pool.
Web Farms are an obvious choice if you’ve hit the limits of your single machine hardware. But before jumping on the Web Farm band wagon (or is that a tractor?) you should look closely at your hardware and application and be sure that you can’t make it all run on a single machine first. Although the process of creating a Web Farm isn’t difficult, administration of two or more servers and keeping them properly synched is actually a lot more work than administering a single server.
Upgrading your hardware is certainly one option available to you. Today’s hardware is incredibly capable and should be sufficient to handle all but the most demanding Web applications on a single box. Multiprocessor machines with up to 16 processors on Windows make a pretty powerful platform to run Web applications with, even if those high end machines are rather pricey. While the Yahoo’s and Amazon’s won’t run on a single box (or on Windows for that matter), a vast majority of applications are likely to be able to comfortably serve millions of transactional hits a day from a single machine even with a single processor.
But Load Balancing can also provide benefits in the overload scenario. For one, it’s generally cheaper to throw mid-level machines at a load problem rather than buying one top of the line high end machine. Even with server licenses involved multiple low end machines might provide a more cost efficient solution.
Load Balancing also provides something else that has nothing to do with scalability: The ability to have failover support if something goes wrong on one of the servers in the pool. Because a Web Farm are made up of essentially identically configured servers, a failure on a single server will not bring down the entire Web site. Other servers in the pool can continue to process requests and pick up the slack. For many companies this feature of load balancing is often important for peace of mind both in the knowledge that a single point of failure on the Web Server is avoided as well as providing an in place mechanism to grow the application should the need arise at a later point.
How does it work?
The concept behind Network Load Balancing is pretty simple: Each server in a Load Balancing Cluster is configured with a 'virtual' IP address. This IP address is configured on all the servers that are participating in the load balancing 'cluster' (a loose term that's unrelated to the Microsoft Cluster Service). Whenever a request is made on this virtual IP a network driver on each of these machines intercepts the request for the IP address and re-routes the request to one of the machines in the Load Balancing Cluster based on rules that you can configure for each of the servers in the cluster. Microsoft this process Network Load Balancing (NLB). Figure 1 shows how the process works graphically. 
Figure 1 – A network load balancing cluster routes requests to a single virtual IP to available servers in the load balancing cluster. Note that each machine is self-sufficient and runs independent of the others duplicating all of the resources on each server. The database sits on a separate box(es) accessible by all servers.
