F5 Networks (News - Alert) is a major provider of solutions that optimize the security, performance and availability of IP-based applications. It was founded in 1996, went public in 1999 and has an impressive 1,355 employees. Its 2006 revenue was $394 million and its year-to-date revenue as of September 2007 was $380 million.

 

Recently F5 released the newest version of their highly flexible BIG-IP Local Traffic Manager (LTM), already embedded in service provider networks around the world. BIG-IP enables IP services to smoothly scale and solves interoperability issues, as carriers and service providers find themselves moving to deploy rich multimedia applications and continue their long migration toward an IMS (IP Multimedia Subsystem) infrastructure. BIG-IP device also enhances reliability and essentially establishes an intelligent, end-to-end service delivery network.

 

Mike Krasnow, Product Marketing Manager for F5 Networks, says, “To understand where F5 fits in service provider networks, let’s first look at how we evolved in the enterprise data center. F5’s position in the data center began at the Web server tier 10 years ago. F5 allowed our customers to make many servers look like one to the outside world. This became known as load balancing. Companies were able to manage growth and customers gained availability, scale and additional security for their web infrastructure. We then extended that same approach to data centers and links. While there is network intelligence required to provide this functionality, most of it occurs at the packet level.”

 

“About five years ago, our customers asked us to extend those same benefits to the application servers in their data centers,” says Krasnow. “F5 responded by launching TMOS — our Traffic Management Operating System. TMOS allowed us to provide a network that was significantly more intelligent.”

 

“Since then, we’ve grown so quickly because of our penetration in both the enterprise and service provider market,” says Krasnow. “Our new announcement is a service provider offering, but as you can see we’re also in the enterprise space.”

 

“We basically help deliver applications,” says Krasnow. “We call ourselves the leader in application delivery controllers. As service providers or enterprises are rolling out new applications, we help those applications scale and make them actually work. We make sure that users can connect to these applications and that those connections are fast, secure and available. It’s about doing intelligent traffic management; not just load balancing but actually managing the traffic at Layer 7 to really optimize it. We handle compression, caching, SSL (Secure Sockets Layer) off-load, rate shaping — really, the whole point is to ‘let servers serve.’ And we do all of the traffic management to get to those servers. So rather than making each server manage its own SSL certificate and do those set-ups and teardown, we do the SSL offload ourselves.”

 

“By understanding the applications and being fluent in them,” says Krasnow, “we have been able to change the network from a passive, connectivity-centric piece of plumbing to something that can provide intelligence at the application level while operating at network speeds. Being a fast application level proxy allows customers to take work that would normally have to be done in many servers and instead do it once on the network. This frees servers to do what they were designed to do — serve applications. As a result, customers save money and enjoy improved control over how their applications behave.”

 

“Our TMOS architecture runs the Big-IP and that allows us to do intelligent traffic and line rates [wire speeds],” says Krasnow, “So it’s really a fast and intelligent engine to run all of this traffic management on top of. During the dot com bust we realized that enterprises were deploying IP applications, and so we developed partnerships with people such as Oracle (News - Alert) and Microsoft so, as they deploy these applications, we help explain how they’re deployed and really help manage those applications. We just recently did an application-ready network for Microsoft, where we partnered with them and other entities on how to deploy something like Exchange Server and make it work well.”

 

“We just don’t think of the network as a connection between the data center and the client,” says Krasnow. “There’s also a layer where you can be smart in the application and do all of the things I mentioned such as rate shaping, acceleration, SSL encryption, and caching. All of those things can actually be done in the application layer of the network rather than having a server do it. The key to do that is our TMOS, which gives us the ability to plug new pieces of functionality into the OS microkernel while still being able to have the fast, intelligent engine on which to run it. One thing unique to F5 is our iControl and iRules which are two scripting languages. iControl does external monitoring and control, and handles the control of the Big-IP, which has an open API. Similarly, iRules is a network programming language that controls the traffic. So it’s an open scripting language that you can actually really do any kind of manipulation with the traffic.”

 

“We also offer security solutions and global availability solutions that all sit on top of the TMOS architecture and are managed by our centralized management product, the Enterprise Manager,” says Krasnow. “Recently we announced acquired a company called Acopia. What we do the app level they do at the storage level. Rather than manage multiple storage devices, they create a pool of storage reminiscent of the way we create a pool of web and application servers. Their technology decouples access from physical file location and presents a ‘Global Namespace’ view of the data — a federation of the underlying file systems, and masks changes to underlying storage systems from users and applications. It also automates common storage management tasks such as migration, tiering, dynamic load balancing and replication. These tasks now take place without affecting access to the file data or requiring client re-configuration.”

 

“We have a new release of the local traffic manager that takes this concept of intelligent traffic management and extends it out for multimedia services,” says Krasnow. “Service providers that are pulling multimedia applications can use this same intelligent traffic management to ‘scale out’ their multimedia services. With iRules, we can work as an interconnection or bridge to solve some inner-workings challenges that occur between the different pieces of the network. In the process we’ve basically we’ve added support for three protocols: SIP [Session Initiation Protocol (News - Alert)], RTSP [Real-Time Streaming Protocol] and SCTP [Stream Control Transmission Protocol]. [Voice apps use SIP, while streaming video uses RTSP.] BIG-IP can thus do application-layer switching for SIP, RTSP and SCTP to provide scalability and high availability. Indeed, thanks to our TMOS, the BIG-IP can manage any IP application. The addition of SIP, RTSP, and SCTP support in TMOS is a proof point that F5 has the only unified application switch. Now this improved application control can be applied to core IMS control as well as multimedia applications.”

 

“Service providers, whether Cable, Wireless, or traditional wireline, have two primary goals: increase average revenue per user (ARPU), and decrease operating costs,” says Krasnow. “To increase ARPU, service providers are deploying multimedia applications. This is available today for many users — services such as video on your mobile phone, or VoIP service from your cable company. Most service providers are deploying these apps over their existing networks.”

 

“The problem with deploying these multimedia apps over the legacy networks is that these applications are siloed,” says Krasnow. “They can’t interact with each other, and must have separate controls and billing functions. Service providers have to deal with the fact that each application requires its own support layer. This means that to roll out a new application is very expensive, and each application must be maintained separately. Because of this, the costs to deploy and maintain these applications and networks are very high. That’s why many service providers are looking at the IMS standard as a way to create a unified network to reduce the maintenance costs associated with multiple networks and provide a simple way to roll out new applications.”

 

“The IMS standard creates a unified framework to deliver applications,” says Krasnow. “So now a new application can just be plugged into the existing architecture. This can significantly reduce costs for deployment and maintenance, and can also allow end users to take advantage of new cool applications. Indeed, because the applications can now share the same network, IMS enables features never before possible. You could be watching TV, and when the phone rings, the caller ID info would show up on the TV. Or you could be talking on your mobile phone over the 3G network, but when you step into your house, the phone seamlessly switches to your WiFi network”

 

“With this release,” says Krasnow, “the BIG-IP LTM can now virtualize multimedia application servers and core IMS components. The intelligent traffic management that F5 is known for is extended to the service providers’ next generation services. This provides huge benefits for both the application tier and the IMS control tier involving load balancing and high availability, flexibility via our iRules, rate shaping, security, and so forth.”

 

“Even though IMS presents this vision of a scalable, easy to manage framework, there will be some challenges for wide scale deployment,” says Krasnow. “The biggest problem is probably one of scale. How do you make sure that the applications and network can handle millions of users while still providing a highly reliable network? This is a classic load balancing/high availability problem. The other primary problem is complexity, both at the application layer, and in the network. This is, after all, a complex architecture. Many vendors are needed to deliver the applications, and not all implement the standard in the same way, or choose to follow it. Integration issues lead to unforeseen problems that increase the time and cost of deployment.”

 

“As I said, previously, our BIG-IP LTM now supports the multimedia protocols SIP, RTSP, and SCTP,” says Krasnow. “This means that the LTM can now intelligently manage multimedia applications for service providers just like it manages Web application traffic. But it also means that as service providers look to deploy IMS networks, BIG-IP LTM can provide traffic management at the core IMS control layer as well. With these features, BIG-IP LTM is the first IMS-ready application delivery controller, creating an end-to-end, intelligent service delivery network.”

 

Service providers and network operators the world over should find F5’s BIG-IP LTM to be a veritable breath of fresh air for the strictures that confined their activities in the past.

 

-----