Web servers, FTP servers, proxy servers and virtual private network connections can easily become overwhelmed during periods of peak demand -- resulting in poor performance and unsatisfactory user experiences. To combat client/server performance issues, network load balancing distributes user request traffic across a virtual cluster of servers running the same application.
While the concept of load balancing is hardly new, native support within Windows Server 2012 to create and manage virtual clusters is. And this feature eliminates the need for third-party load balancing hardware and software. Here are answers to some common questions about network load balancing (NLB) and dynamic load balancing in Windows Server 2012.
What is network load balancing? And where does Windows Server 2012 fit?
Load balancing eases network traffic to one server by distributing traffic across two or more servers. For example, a server that's overtaxed to 140% of its traffic-handling capacity will experience poor performance, such as long waits for requests. By using load balancing and distributing the workload across a second server, the average load per server drops to 70%. As a result, users receive good performance, even during times of high utilization or traffic load.
Windows Server 2012 incorporates NLB features, combining two or more servers running the same application (such as a Web server) into a virtual cluster that can distribute network traffic across cluster nodes.
Load balancing incorporates scalability and resiliency features into the server cluster. IT administrators can stipulate the load level each host should handle so that older servers can be added to the cluster, even though they may not be able to shoulder an equal load compared with newer servers. Traffic can also be routed to a default server until a threshold is reached and then start distributing the excess traffic to other servers. Windows Server 2012 can support NLB clusters of up to 32 nodes.
Windows Server 2012 also allows nodes to be added and removed dynamically from the cluster. If a cluster requires more computing power, for example, additional servers running the same application can easily join the cluster and take on some of the traffic. Conversely, troubled or failed servers can leave the cluster; NLB will rebalance the traffic load and allow remaining servers to continue operating.
How can I manage network load balancing in Windows Server 2012? What tools are available to manage it?
The Network Load Balancing Manager in Windows Server 2012 handles network load balancing, as do load-balancing cmdlets for Windows PowerShell. Both resources allow IT professionals to manage and configure local and remote NLB clusters. PowerShell control supports the use of scripts to automate important or repetitive load-balancing management tasks.
These tools allow you to add or remove servers from the cluster, create or delete clusters, set traffic thresholds and distribution behaviors, and so on. Windows Server 2012 produces and logs events related to network load balancing, so IT admins can use the Windows Event Log to inspect a record of NLB events.
Port management rule sets allow administrators to configure load balancing for a single IP port or a group of IP ports, or even virtual IP addresses. Desired IP ports can be blocked for greater security. When servers are running different applications, user traffic can be routed to a single server running a specific application. Switch port flooding from multicast traffic is prevented via Internet Group Management Protocol, which prevents switch traffic from being echoed to all ports and helps maintain good network performance.
What hardware and software are needed to create and support an NLB cluster in Windows Server 2012?
Each server in the cluster must have the appropriate hardware to support Windows Server 2012 and be configured to reside on the same network subnet. Otherwise, the hardware requirements are primarily focused on the servers' network interface cards (NICs). The actual number of network adapters on each node can vary, which is handy when mixing hardware types and vintages. However, all of the NICs within a cluster must be configured for unicast or multicast operation -- mixing traffic types within a cluster is prohibited, and unicast NICs must support media access control address spoofing.
Software requirements for NLB include Windows Server 2012 and TCP/IP networking with static IP addresses for each server in the cluster. Consequently, dynamic host configuration protocol is disabled on each NLB server.
NLB spreads traffic load across clustered servers running the same application. This adds resiliency to mission-critical applications by ensuring prompt handling of user traffic, even when a server in the cluster fails or goes offline. NLB also adds scalability, with support for up to 32 servers in a cluster. However, proper configuration and optimization may require a level of technical comfort that many IT professionals lack, so always test and benchmark NLB in a lab environment before rolling it out to production.
This was first published in February 2013