Feature Articles What American, European Experiences Can Teach Us About Meeting WAN Optimization Challenges

By Peter H. Schmidt

European and American enterprises have been taking differing approaches to WAN optimization, but there are signs that they are starting to converge. The historical differences were due to a number of factors, including bandwidth price differences and dissimilar relationships with their carriers. The trend has been that Europeans take a longer-term, strategic approach to WAN optimization in partnership with their carriers, while American firms have generally focused more on the short-term, using their own resources to solve critical problems as they arise. The trend now is for U.S. firms to start thinking more strategically about WAN optimization, while European companies are adopting many of the tactical solutions first offered in the U.S.

Why the differences?

There are two primary reasons for the different continental approaches. First, bandwidth charges are higher in Europe - in fact, there is still much usage-based or MPLS class-based pricing. At the same time, flat rate pricing is much more prevalent in the U.S., and some enterprises here are even obtaining CoS at no additional cost from their carriers.

The second difference is the relationship between enterprises and their carriers. In Europe, many more enterprises buy managed services from their telecom providers than in the U.S. Analysts estimate that carriers manage 70% of enterprise routers in Europe, while that is true for just 30% of U.S. firms.

The pricing model differences probably stem fundamentally from Europe’s history of “Postal, Telegraph and Telephone” companies (PTTs) as government-sponsored monopolies. Each country had their own PTT with its own tariff structure, and usage-based pricing was the norm. When the Internet came along and IP won the war versus ATM, the PTTs hung onto their usage-based pricing models, and the relatively more regulated European market did not experience the race to the bottom on bandwidth pricing that we saw here in the U.S. in the ‘90s.

In contrast, the breakup of the Bell System eliminated the U.S. monopoly PTT at the same time that TCP/IP began its conquest of the enterprise WAN. The U.S. venture capital markets did what they do so well in the ‘90s, creating thousands of new ISPs and overbuilding backbone capacity. Flat-rate pricing was born out of rabid competition and pools of venture money. WAN provisioning was quickly driven towards being a commodity, and the lack of profitability inhibited most service providers from being able to innovate with their service offerings beyond just bandwidth. Enterprises in the U.S. thus came to view Internet and WAN services as commodities and cost centers.

When the venture money eventually ran out for ISPs in the late ‘90s, and finally when the dotcom bubble burst in late 2000, only the players with the lowest cost of capital were left standing. Economies of scale dictated those would be the incumbent “phone companies.” So now, we have a handful of integrated telecomm/datacomm vendors in the U.S. that have recovered from the bubble and are looking for innovative ways to offer value to enterprises beyond just “bit hauling.”

While U.S. service providers were wearing each other down with price discounting, European carriers were building cooperative relationships with their enterprise customers. They could afford to invest more in managed services because they had more profitable and more predictable revenue streams. European service providers have earned the trust of their business customers and have built up expertise in managed service provision that the U.S. carriers are only now starting to develop.

Of course, the flip-side of this is that U.S. enterprises have had to be more self-reliant and have built up their own expertise in managing their own WANs. Some organizations have made astonishing investments in their WAN and their staffs. One of the world’s largest financial services companies, headquarterd in Boston, has a staff of six people dedicated to application performance, and three of them have Ph.Ds! Clearly, it will be hard for any outside vendor or service provider to enhance its capabilities much.

Another important difference is that IT outsourcing has had a larger impact in the U.S., leaving staffs with a much heavier per person operational and support workload than was true a decade ago. As a result, many network professionals are constantly in fire-fighting mode, with little time to work on longer-term projects.

Tactical Approach

This has led to a tactical bias among U.S. enterprises, in which pain-points are addressed with point solutions. While this has resulted in short-term benefits, tactics alone don’t win in the long run. Organizations require a sound strategy to guide IT investments over time. We learned from the dot.com era that buying the latest hot network device doesn’t directly correlate to improved business functions. CFOs have taken that lesson to heart and are pushing IT for more strategic investments.

Acceleration is a hot technology, and for good reasons. It can be very effective at solving tactical problems. Caching, compression, local CIFS acknowledgments and TCP window manipulation can each be very effective pain relievers. However, with time, the pain comes back in new places, and it doesn’t take long before the lack of coordination in the leading acceleration products starts causing headaches of their own.

A More Strategic Approach

A more strategic approach is to provide application acceleration and control it within the framework of achieving application performance objectives. It means going beyond reporting on applications and beyond accelerating them - it means guaranteeing application performance, especially for an enterprise’s most critical applications. It takes a unified perspective and product architecture to achieve that.

Acceleration that speeds up both recreational Web surfing and the critical ERP application, but which still leaves the ERP application under-performing during periods of congestion does not meet business needs. The first priority has to be to ensure that the ERP application always meets its performance objectives.

This leads to the concept of application SLAs. While European enterprises may have been slower to exploit some short-term benefits with acceleration, they have certainly been paying attention to the long term. Several European service providers already offer enterprise customers application SLAs. Moreover, some enterprises are even offering SLAs to their internal customers. Application SLAs enable a strategic approach to the IT infrastructure. They bring IT, service providers and users together around the critical issue - providing users with good quality of experience so they can get their jobs done. Ultimately, all that matters in IT is good user QoE - of course, at a reasonable cost and with high availability and reliability.

While economic realities are pushing both enterprises and service providers to deepen their relationships and invest more strategically, there are plenty of short-term pain points that need addressing now. Thus, we can take away from the respective European and American experiences that enterprises should seek to achieve good application performance both now and in the future. The best way to approach this is to deploy acceleration solutions within the planned framework of achieving application performance objectives and application SLAs. IT

Peter H. Schmidt is technical director for Ipanema Technologies (http://www.ipanematech.com), a provider of application traffic management systems for wide-area networks.

» Internet Telephony Magazine Table of Contents