University at Buffalo - The State University of New York

                           
 
     The data communications network is a critical resource supporting the continuity of many University services. The underlying backbone infrastructure supports every application and IT service that UB offers. We rely on the backbone to connect our users, both local and remote, to the many services and applications that they need. We have spent the last couple years upgrading the different parts of our IT infrastructure from edge to applications. A result of these changes is that network traffic across our backbone is increasing and changing. It is time to upgrade the backbone so it will continue to reliably support the services and applications of today as well as the future.
     A chief reason to upgrade the backbone is to provide the bandwidth needed to support applications. Our current backbone is made up of several single gigabit links that connect the main routers. With the completion of UBNET we will have installed gigabit ports down to every user’s desktop. As the edge ports are used at gigabit speeds we need more bandwidth in the backbone to handle the load. We need to upgrade the backbone links as well as the links to each building to 10 gigabit. The increase in bandwidth will support the aggregation that occurs at the building switch where hundreds of gigabit ports are consolidated into one uplink.
     A second reason to upgrade is to improve reliability by increasing the redundancy and fault tolerance. The network has become a critical resource required to support he business functions of the university and uptime is becoming increasingly more important. To provide the uptime needed for applications like VOIP, UBLearns, E-Mail we need to add more redundancy into the backbone. We are proposing backbone router redundancy as well as adding redundant links to each building. This will ensure that even in the unlikely event of a backbone router failure, we will not lose communication with any of the buildings or users. Maintaining functionality for users is a critical goal of the new backbone.
     A third reason to upgrade is support for new features and functionality. While the backbone already supports critical features like Quality of Service (QOS) and DHCP relay. It does not fully support IPV6. With the IPV4 address space exhaustion problem, IPV6 is the future for network device addressing. We need to upgrade the backbone so it fully supports IPV6 as well as IPV4. In this way it will allow us to start getting ready for the transition and support the new IPV6 network as we build it.
     This will be a long transition but the first step is adding complete end to end network support for IPV6. Implementation of the new backbone infrastructure is estimated to cost about $5 million dollars and will take several years to complete. The high cost of the project and the time to complete is one of the factors why we need to start now so that we can spread it out over time. The upgrade will be done in several phases over the next three years. Phase 1 is designed to construct a 10gig core that supports all the functionality listed above. Once the core is built we will start upgrading the building switches to support 10gig connections into the backbone. This will take several years and will probably be sequenced by router. We are decided to use Cisco for phase1 based on their market position and quality of products they offer. Phase 1 will cost roughly $750k and will be implemented over the a year. A schedule of phase 1 attached.