Americas

  • United States

Asia

Oceania

by No Analyst or Consultant

Intelligent Routing: Bringing Reliability to the Internet

Feature
Sep 18, 20024 mins
CSO and CISOData and Information Security

By Zeus Kerravala,

Nick Maynard,

and Anil Phull

The low cost of Internet access today, combined with new Web-based and peer-to-peer applications, has created many business opportunities for enterprises and service providers. However, there are still many problems when using the Internet for business applications. Unpredictable performance, slow response times, and the inability to assign meaningful service-level agreements (SLAs) to Internet-based traffic often make the decision to use the Internet for mission-critical applications a gamble. However, there is a group of start-up vendors that deliver intelligent routing (IR) to enterprises, and allow companies to gain control and maintain business-class performance over the public Internet.

Corporations that wish to protect themselves from the problems of using the Internet have done so by purchasing connectivity to two or more Internet service providers (ISPs). This process, also known as multihoming, protects the enterprise from situations where an ISP is unavailable, since, in most situations, one of the two ISPs will always be up and running. One problem is that multihoming does not address is the brownout situation, where both links are available but one ISP is performing poorly. Border Gateway Protocol (BGB) is an exterior routing protocol used by routers on the Internet to determine where to send traffic. As is the case with all routing protocols, BGP was not designed to make routing decisions based on performance metrics like latency or packet loss.

Instead, BGP uses metrics such as router hop count and the number of autonomous systems the traffic passes through. Normally, the route with the fewest number of hops wins, and the traffic is sent down that link. However, the route with the fewest number of hops is not necessarily the fastest link due to congestion or other problems. IR platforms measure the latency and packet loss of each link and determine routing decisions based upon these performance metrics. This ensures that the traffic is always sent down the ISP connection with the best performing connection, which is necessary for some applications such as video or storage that require high-performance connections. IR platforms can also optimize traffic for cost versus performance if the enterprise chooses.

The most obvious benefit to IR involves cost savings. Following are some of the areas in which enterprises can benefit by deploying intelligent routing platforms:

  • Optimizing traffic by ISP. Enterprises that use different billing methods for their multiple ISPs can optimize their contracts. By sending the bulk of nonmission-critical traffic over lower-priced ISPs, companies can minimize the amount of traffic sent over their Tier 1 ISPs.
  • Reduction in network engineering operational expenses. Since the IR platform makes many of the decisions about where to route traffic, the network staff will not need to spend as much time managing routes. These high-level engineers can focus their energies on other network issues.
  • Billing remediation. Since the IR platform determines how much traffic is sent over each link, the IR platform vendors should be able to provide tools that allow the enterprise to determine how much their Internet bills should be. The enterprise can then use this information to ensure the accuracy of billing information from the ISP.
  • Greater use of the Internet for corporate traffic. By using Internet-based VPNs with IR platforms deployed, companies can experience high-quality VPN connectivity.
  • Increased revenues. Since the IR platforms address not only network failures but also brownout situations, Internet-based access to e-commerce sites, VPNs, and extranets would all be improved.

Conclusions

Enterprise demand for intelligent routing solutions will continue to grow as companies continue to deploy Internet-based applications and services, and as IR solutions provide enhanced functionality and analysis capabilities to improve network availability. The solutions available to the enterprise are dedicated hardware platforms, IR service providers, and hybrid solutions. We expect that sales of the hardware platforms will be limited to the Fortune 1000 and service providers that wish to offer an IR service, primarily due to the large price tag. It is doubtful that an enterprise will pay an enterprise to be willing to pay over $100,000 per device for a technology that is in its relative infancy. The price point of these products needs to drop to about $50,000$60,000 in order to be more widely adopted by enterprise customers.

This is good news for service providers offering this service and hybrid solutions available today. By using a service or hybrid solution, companies can try the service and measure the value before purchasing their own platform. The larger enterprises that opt for service will want some control and will choose the hybrid solutions, allowing them to try the service with very little risk. For larger mid-sized companies wishing to use this functionality, purchasing this as a service from their carrier is the best strategy until the price points of the products drop to an affordable level.