|Issue:||Asia-Pacific III 2001|
|Topic:||Businesses Challenge Carriers to Deliver a Reliable E-business Infrastructure|
|Title:||Vice-President of Marketing|
Reliability of e-business networks and infrastructure is now a key consideration for providers and customers. Customers are increasingly seeking the near perfection of 99.999% (‘five nines’) availability 24 hours a day. Every lost minute means lost business and lost profits. Downtime is no longer acceptable, says Alex Dobrushin, Vice-President of Amber Networks. The clever e-business users are moving to real-time network management and negotiating clear service level agreements with the third-party vendors that operate their systems.
The rapid pace of business dictates a rapid response. To gain it, businesses are moving to the ‘net’. In relying on the Internet to communicate information that is critical to their success, businesses are demanding the same high level of availability from the carrier’s IP network infrastructure as they get from their telephony services and leased line networks. The benchmark for service availability is today’s Public Switched Telephone Network (PSTN), which delivers 99.999% (or ‘five nines’) or better network availability. A rating of 99.999% translates into a maximum of five minutes of downtime per year. FCC regulations in the United States, for example, mandate that Class 5 telephony switches should be down no more than 40 minutes in ten years-an average of just four minutes per year. With increased competition in the realm of e-business, any network outage can have a big impact on the bottom line. As far as the businesses are concerned, downtime is not an acceptable option. Downtime translates into lost business, lost customers, costly delays, or even idle production lines. Five nines availability is essential for a growing number of mission-critical e-business applications such as: · E-commerce Transactions: Bad customer experiences hurt the bottom line. The goal is to sell goods and services online 24 hours a day, seven days a week without interruption with a secure reliable Internet storefront. · Enterprise Resource Planning (ERP) (e.g. Baan, Oracle, PeopleSoft, SAP and SNA): These WAN applications must run in real-time to keep businesses operating efficiently. · Voice services based on IP (VoIP): IP-based telephony must always provide dial tone and performance that is consistent with PSTN. E-business Maturing As e-business grows in importance, the need to maintain fast response times and high levels of service availability is also rising in importance. Service failures and services lags affect the bottom line. What makes executing a successful e-business implementation so daunting is that so many of the sub-processes of the e-business transaction have been virtualised and outsourced. Activities such as credit card authorisation, VPN services, billing, fulfilment, shipping, and customer service call centres are remotely accessed. Working with them in real-time requires rock-solid communication links. Increasingly, IP services are a mission critical link in an e-business solution. With success linked to maximum uptime, non-stop infrastructures and processes have become a priority for e-business executives. To attain high availability, businesses with mission-critical applications are investing in a new generation of non-stop infrastructure products and services. This includes best of breed equipment and software (e.g. firewalls, premise routers, load balancers, web servers, application servers, and databases). In order to succeed, these companies also observe best practices when it comes to monitoring and responding to network performance and fault recovery. But it is not enough to manage the boxes that the company controls on its own premises. Managing the hosting service providers and their connections is required. This means that managing Service Level Agreements (SLAs) with service providers is a key element of a high availability e-business solution. As a result, these businesses will seek the best available outsourced services, including, Application Service Provider (ASP) services, VPN services, and web hosting services. Savvy e-business users are moving to real-time network management and to SLAs with all of the third-party vendors that operate their infrastructure (e.g. ASPs, data communication service providers, ISPs, and web hosting companies). If companies are doing cash transactions, they are also looking for the most reliable connections to External Certificate Authorities and Secure Electronic Payment systems. Whether they provide the communication channel or perform processing, online providers are getting the squeeze from their customers to make their networks and systems ultra-reliable. For service providers, this means that the IP network connections that underlying business processes must perform with high availability. The Service Provider Challenge Faced with this challenge, service providers are looking for ways to meet the critical customer need for high availability. As the Internet grows in popularity, service providers continue to deploy highly scalable edge routers that support greater subscriber densities at the IP edge. A single edge aggregation router typically terminates hundreds or thousands of dedicated access lines to the Internet. It is at this network edge where the customer’s mission-critical services are delivered and the service provider’s revenue is generated. Service providers are investing in fault-tolerant edge routing technology in order to maximise data network reliability. Service providers are also rapidly moving to Quality of Service (QoS)-enabled IP networks and considering Multi-Protocol Label Switching (MPLS) to shore-up the e-business-related network performance needs, including: · the ability to offer Service Level Agreements (SLAs) that can guarantee that each customer gets the appropriate levels of service, including parameters such as Committed Information rate (CIR), latency, jitter and uptime guarantees; · the ability to offer premium IP services to gain incremental revenue from their existing IP network investment; and · the ability to take advantage of internal efficiencies in conserving network resources through aggregation of traffic flows and traffic engineering. As a result, edge routers that aggregate customer traffic are rapidly becoming deployed in QoS-enabled configurations. The edge router is becoming a key component in the process of enforcing SLAs, and maintaining and growing customer revenue. With high-availability routers, service providers can work towards the delivering PSTN-grade reliability in the e-business-focused IP infrastructure. Legacy Router Downtime Previous generations of edge routers that are currently deployed present a barrier to the proliferation of reliable network-based e-business environments. A conservative estimate of the typical legacy edge router’s annual downtime sheds some light on the network availability challenge. Typically, edge routers are taken out of service three times a year for software upgrades. The typical router also experiences at least one hardware or software failure per year. Recovery from an outage typically takes a minimum of ten minutes until routing service is restored. The typical legacy edge router is down at least 40 minutes per year. That’s if the recovery process goes well. The ten-minute figure for router downtime assumes that the system reboot and routing protocol recovery is handled automatically. Significant manual intervention is often required. If a routing engineer has to get involved with additional troubleshooting, the outage can balloon to several hours per incident. High Availability Edge Routers This goal of PSTN-grade availability will remain elusive until a new generation of routers that are purpose-built for service provider edge network requirements are available. Bringing resiliency to IP edge requires a ground-up approach to router system design, including a resilient hardware architecture and fault-tolerant operating system that together provide a fast and reliable data infrastructure. A new class of fault-tolerant routers is now available to provide protection at the service layer and support redundancy of routing states and tables. This capability essentially restores services with optical-like speed of less than 50 milliseconds. The realisation that upper layer protocol behaviour dominates restoration time in IP-based networks has led to a new generation of carrier-class routers that feature full redundancy. They feature truly ‘hot’ standby processors, mirrored routing session and information states, and packet forwarding tables that can be swapped in tens of milliseconds. Besides the local significance of being able to quickly recover service, the introduction of fault-tolerant edge routers eliminates a multitude of harmful network side effects. Most notable is the shielding of the core network from the potentially negative effects of edge router failures. In a network with a traditional edge router, the customers will be out of service and the network will have to go through multiple re-convergence states, each introducing a new possibility for a fatal crash. Fault tolerant routers eliminate this undesirable behaviour, making network availability easily predictable for the service provider. Deterministic prediction of network downtime allows service providers to offer SLAs that satisfy customer demand for availability without exposure to the financial risks associated with downtime. This is a good model for providing a reliable and cost effective e-business infrastructure that benefits both subscribers and providers. Conclusion Building a rock solid IP infrastructure is a desirable strategy for service providers who hope to respond profitably to the challenges of e-business. The benefits of attaining high availability in IP service networks are immense. Service providers improve their return on investment (ROI), and benefit from uninterrupted revenue-generating capability and customer satisfaction. For e-business, the difference between hours of downtime and no downtime can be measured in revenue or lost productivity.