Unravelling the IP VPN Dilemma

The booming WAN services market is proving to be bad news for Frame Relay. According to In-Stats recent report, "High Growth and Lots of Opportunity," Frame Relay service revenue will steadily decline between 2005 and 2008. One leading reason for this decrease is the uptake of IP Virtual Private Networks (IP VPNs), which account for the lion’s share of the predicted $20 billion to $25 billion wide-area network (WAN) services market.

While the popularity of IP VPNs for increasing WAN bandwidth is most certainly tied to the significant cost savings over traditional dedicated lines, there is a growing awareness of the technology’s downside. While IP VPNs provide an order of magnitude more bandwidth for less cost, they frequently have significantly higher latency, packet loss, and jitter than do dedicated lines, which often results in less actual throughput and reliability.

With native IP VPN connections, there is a minimal amount of traffic engineering available, creating a “first come, first served” environment on the WAN. This creates an erratic WAN experience, with little control or ability to accelerate or prioritize traffic. So, while a customer can get 10 Mbit/s of bandwidth using an IP VPN for the same price as a 1.5 Mbit/s connection via a dedicated line, the actual gain in usable throughput can be minimal.

As companies seek cost-effective ways to handle and support enterprise applications, portals, business continuity, disaster recovery, data replication, and voice over IP (VOIP), WAN performance issues represent serious challenges.

Enter WAN optimization and application acceleration technologies.As the IP VPN trend continues, deploying WAN optimization technologies enables companies to make better use of the increased bandwidth. Taking it one step further, technologies are emerging that combine both WAN optimization and application acceleration technologies on one platform, managing traffic at the session layer to provide the highest rate of performance for individual applications and the WAN as a whole.

These so-called enterprise application shaping solutions are ideally suited for both bandwidth-hungry applications such as networked storage, disaster recovery, data replication, and large file transfer, and for transactional applications such as customer relationship management (CRM), enterprise resource planning (ERP), or enterprise portals. They become even more essential when these applications are deployed over IP VPNs.

In order to meet typical enterprise performance thresholds, bandwidth-hungry applications often require high WAN bandwidth (10 Mbit/s or greater), low round-trip latency (50 milliseconds or less), compensation for high packet loss (greater than 1 percent), and compensation for network jitter. While transactional applications typically don’t require high speed in order to perform well, latency, loss, and jitter issues can be devastating to these applications as well. Deploying these applications over a less-controlled IP VPN connection, where these problems are prevalent, is inherently troublesome.

By managing traffic user-session by user-session, enterprise application-shaping solutions mitigate the issues presented by IP VPNs. Using adaptive Transmission Control Protocol (TCP) acceleration technology, these solutions dynamically evaluate the requirements of each application and the capabilities of the WAN. The technology then negotiates TCP window sizes that ensure optimal throughput, then compresses and encrypts data streams before they merge onto the network, to minimize redundancies. This eliminates the constraints of TCP window size and other parameters, thus eliminating the “bandwidth delay product problem” and minimizing retransmission in case of packet loss. The result is 100 percent utilization of IP VPN WAN links, even over extreme distances, for both compressed and uncompressed data.

While allocating IP VPN bandwidth to the problem is a common reaction when application performance is suffering, this method is ineffective if the additional bandwidth cannot be utilized to its full extent. Rather than simply purchasing additional IP VPN bandwidth, deploying a WAN optimization and application acceleration technology that handles traffic at the session layer can ensure that you are getting the most of the IP VPN bandwidth you already have. The result: Your existing IP VPN bandwidth achieves higher quality of service and throughput than even dedicated lines, utilizing 100 percent of your IP VPN’s capabilities and providing an even greater return on investment.— Dan Kolkowitz, Vice President of Engineering, Swan Labs Corp.