Big/IP2

Load balance your TCP/IP traffic with F5 Labs' Server Array Controller

As demand for an Internet presence becomes integral to your company's marketing strategy, the need to keep your Internet servers online becomes paramount. However, because of inherent inadequacies in the TCP/IP protocol and the services that run under it, your Internet servers can become overloaded. One solution is to use a load-balancing program to spread access requests from the Internet across multiple content servers. BIG/ip² from F5 Labs can help you accomplish this goal.

BIG/ip² is a server array controller that distributes access requests from the Internet among the network servers it controls. BIG/ip² monitors each server in its domain to determine service availability and performance, which improves response time for your Internet-originating traffic. If Internet requests overload one Web server, BIG/ip² redirects new requests to another Web server. If you take a mail server offline for maintenance, BIG/ip² detects that the server is offline and routes inbound email to another server. The system is a fully redundant, fail-safe server array controller with automatic failover.

BIG/ip² is both a hardware and software product. It is a complete, embedded Pentium-based computer system in an industrial rack-mount case. The BIG/ip² unit measures 7" high X 19" wide X 17.8" deep and weighs 30 lbs. The unit comes with a 1.6GB hard disk and 64MB of RAM. It includes two network cards, but the factory custom installs the cards because everyone's network has different requirements. The unit supports 10Base-T, 100Base-TX, Fiber Distributed Data Interface (FDDI), and asynchronous transfer mode (ATM) network topologies.

Installing and using BIG/ip² is not a task for the neophyte systems administrator. The product is based on Berkeley Software Distribution (BSD) UNIX and uses shell-based and X-based applications for administration. Understanding UNIX is beneficial to using the product. For example, when you first start the unit, you must supply a root password. Unless you're familiar with UNIX, you probably would not know the significance of the root password.

The installation process runs automatically the first time you turn on the BIG/ip² unit. You must connect a monitor and a keyboard to the unit. The unit's two network cards are labeled internal and external. You need to place BIG/ip² between your LAN and your Internet router. In a 10/100Base-T environment, you need a crossover cable to connect the unit directly to your router or the unit will not function correctly. BIG/ip² provides an additional layer of security to servers located inside or outside of firewalls.

Installing the BIG/ip² is a snap, but its configuration requires network administration knowledge. You must supply the unit with an IP address for the internal and external network interfaces and your router's IP address. You must designate which machines on your network can remotely access the BIG/ip² host to perform maintenance and configuration.

After you configure your network interfaces, you must define a series of virtual Internet protocol (VIP) addresses, nodes, and ports using the X-based xbighost or shell-based bighost program. I recommend the X program. Using the GUI interface is easier than using the shell program to manually configure the system. To use the X interface, you must have an X server (e.g., Hummingbird Communications' Exceed) that can connect to the BIG/ip² server and run xbigpipe. After you configure the nodes and ports, you must define the load-balancing method you want to use. Then you're ready use the product.

The BIG/ip² unit's price includes a license for unlimited content servers, 2-day on-site installation and integration into your environment, and a 1-year, 7 X 12 toll-free support line contract (7 X 24 support optional). BIG/ip² probably won't become the standard for every company with a large Web site. However, the product is a logical choice for larger installations with decentralized Internet services.