• End-to-End Network Reliability: Fact or Fluff?
  • October 18, 2007 | Authors: Mark J. Maier; Stephen Gold; Michael T. Hepburn
  • Law Firms: McGuireWoods LLP - McLean Office ; McGuireWoods LLP - Chicago Office
  • Operating a business in today’s global marketplace puts significant demands on a company’s telecommunications network infrastructure. Supporting remote offices and users can require a substantial investment in technology to ensure that the needs of those offices and individuals are met. This is especially true considering the current trend for companies to consolidate applications and data onto fewer servers and into fewer data centers, via approaches such as virtualization (See “Virtualization: Another Reason to Read Your IT Contracts”). As available bandwidth is congested with the demands of technologies such as streaming media and voice over IP (VoIP), end-to-end network reliability is a promise that customers want to make sure is not just vendor marketing hype. Below are some key considerations for corporate IT professionals and the lawyers who support them to consider when interacting with potential network service providers.

    1. Understand the Technology. The “network” circuits that companies rely upon to interconnect their offices are most often made up of a number of links and segments operated by different organizations rather than single circuits. Each supplier organization is typically only willing to be responsible for its own piece of the overall network and measures only the performance of its own individual link(s). For example, synchronous optical networking, or “SONET,” is a very fast fiber optic network that can offer circuit speeds starting at 51Mbps (million bits per second, roughly thirty times the bandwidth of a typical home internet connection) for an “OC-1” and 155Mbps for an “OC-3.” However SONET networks are typically made up of the local loop at the customer’s premise, a long haul trunk and another local loop on the distant end. Since each of these loops is installed in a different place, different regional telecommunications providers and/or separate business units operate each one. As such, each loop may have its own independent service levels and performance standards that do not measure overall “end-to-end” network performance and reliability from the customer’s perspective.

    So, even if each individual loop is operating at peak performance, if these links are not interconnected properly and measured across the entire circuit, true end-to-end performance and reliability may not be achieved. Unless sufficient service level agreements and other contractual provisions are present, an end user may lose the entire end-to-end connectivity because a “link in the chain” went down, but the provider may only be liable for the single link outage.

    2. Understand Your Business Needs. In addition to understanding the network technology, there is also a need to understand how the technology fits into a company’s particular business needs as dictated by its IT / telecommunications strategies. For instance, if a company has instituted a centralized, as opposed to distributed, server strategy consolidating its email, file, print and application functions onto a reduced number of servers located at a central data center, then fast, end-to-end network reliability becomes extremely important as remote users will no longer have a local server with which to communicate. Instead, the user’s computer will rely on the telecommunications network to communicate with the server located at the central data center. In short, if the telecommunication network is not fast and reliable enough, remote users and offices will not be able to work efficiently or maybe not at all.

    Another consideration is whether the applications that are operated across a telecommunications network require real time access, e.g., video or VoIP transmissions, or if the applications are more “store and forward” like email. The more users need real time access to information and applications, the faster and more reliable the telecommunications network needs to be.

    One further factor always impacting a company’s needs analysis is the budget a company has available to spend on telecommunications infrastructure. Although everybody would like to have super high speed, 100% reliable connections across their network, each incremental increase in speed and reliability will result in additional costs. Companies should consider the amount of down time certain applications can sustain. Although 99% uptime may seem like a high degree of reliability, this allows for 864 seconds (over 14 minutes) of downtime per day. Compare this to 99.9%, which allows for 86 seconds (1.4 minutes) of downtime and 99.99% for only 8.6 seconds (0.14 minutes) of downtime per day.

    3. Focus on Service Levels. Customers looking for true end-to-end reliability from a network service provider should negotiate telecommunication service agreements that clearly measure performance and reliability across each end-to-end circuit. Measuring each individual link may not meet the needs of real time applications, such as VoIP and streaming video, for consistent and fast throughput with limited latency. Corresponding contractual service levels and performance credits should also be considered.

    For example, here are sample provisions constructed to address end-to-end performance and reliability of a SONET circuit. Note that the specific standards have been omitted; those would be negotiated based on customer-specific requirements and vendor capabilities.

    Speed. The speed between the customer premises equipment between two sites connected with OC-1 local loops shall be at all times equal to or faster than ____ Mbps.

    Latency. The total latency between the customer premises equipment between Site A and Site B connected with OC-1 local loops shall never be greater than ___ ms.

    Reduced Performance. If the circuit between the customer premises equipment between two sites connected with OC-1 local loops results in a 0.000001 or greater bit error rate for a cumulative twenty (20) minutes or more during any consecutive thirty (30) day period, provider shall credit the customer _________.

    Downtime. If the circuit between the customer premises equipment between two sites connected with OC-1 local loops results in a 0.001 or greater bit error rate for a cumulative ten (10) minutes or more during any consecutive thirty (30) day period, provider shall credit back to the customer _________ and customer may terminate the affected loops without payment of a termination charge.

    By using approaches like these and focusing on end-to-end performance and reliability, customers can better address the requirements of modern telecommunications applications.