WAN optimization is important to global businesses. The pressure to improve the performance of applications running over the WAN keeps increasing as more and more enterprises centralize their systems, find they must provide higher performance to more remote users, and try to cut their bandwidth costs. Because performance optimization devices are relatively inexpensive, return on investment (ROI) is fast. Payback is often less than a...
year, counting only hard costs. The softer advantages, such as greatly improved employee productivity, usually result in even faster ROI.
Large enterprises are therefore planning to use WAN optimization on more than just a couple of links. They need systems that are:
- Scalable (in both bandwidth and in total number of endpoints).
- Easy to integrate into existing large network architectures (working with network rerouting, asynchronous routes, QoS structures and access control lists).
- Compatible with security and authentication designs (including acceleration of encrypted data flows and compatibility with end-to-end authentication mechanisms).
- Easy to configure and manage (including performance measurement and problem diagnosis).
In response, the market is moving to larger companies with sales forces that can sell optimization technology to their existing base and can also make the powerful "integrated solution" argument. Large vendors claim that the technologies are now becoming mature enough for a single-vendor, integrated solution to beat out a marginally better "best-of-breed in optimization but not in integration" solution. The major players are well capitalized or have large installed bases; it's difficult for a new player to find an open, exploitable niche.
Smaller vendors will probably grow only if they can develop strong channel sales to small and medium-sized enterprises, hoping to sell to customers that are tied to their channel suppliers. Otherwise, they probably hope to be acquired quickly by a larger vendor that needs to solve a problem with an existing product.
Over the next couple of years, single, integrated systems will appear that handle all application front end, security, site-to-site, and site-to-user acceleration functions within the data center. That integration will greatly simplify some current configuration issues:
- Compression, protocol modification and load balancing can be performed only on cleartext data. Therefore, SSL encryption, which was formerly performed only in the application front end or the server system, may need to be performed in the acceleration device or between that device and the external network.
- Creating high-reliability architectures is difficult when many different appliances are involved.
- Most of the existing appliances have overlapping functions, such as QoS, compression, caching and some protocol modifications.
- All of these functions must be integrated with security technologies.
- All of these functions must be configured and controlled without creating inconsistencies.
At remote offices, there's also pressure to combine everything into one easily maintained appliance, preferably one that needs no local IT staff and can be easily swapped with an identical box if there's a problem. Indeed, some vendors are beginning to offer appliances that provide all remote office functions, including access routing and a virtual machine for running Unix or Windows applications.
These fully integrated solutions are not completely ready for major production use; so enterprises -- remembering that ROI is often less than a year -- should acquire WAN optimization now but set expectations properly with upper management. They should be ready to replace their current optimization solution in approximately three years, when new integrated products are available and when the enterprise itself may have a new network architecture as a result of changes in security architectures, Microsoft Windows, and application delivery designs.
About the author:
Eric Siegel, a Senior Analyst at the Burton Group, is a nationally-known authority on network performance optimization, measurement, management, and QoS. He is the author of two acclaimed books, Designing Quality of Service Solutions for the Enterprise (John Wiley & Sons) and Practical Service Level Management: Delivering High-Quality Web-Based Services (John McConnell with Eric Siegel; Cisco Press). A member of the Internet community since 1978, Mr. Siegel was the Principal Internet Consultant at Keynote Systems and was a Senior Network Architect with Tandem Computers, where he was the technical leader and coordinator for all of Tandem's data communications specialists worldwide. Mr. Siegel also worked for Network Strategies and for the MITRE Corporation, specializing in computer network design and performance evaluation.