Balancing cost and reliability in the design of IP backbone using agile optical networking

To address the reliability challenges due to failures and planned outages, Internet Service Providers (ISPs) typically use two backbone routers at each central office with access routers connecting to them in a dual-homed configuration. At the IP layer, providers deploy redundant backbone routers and redundant transport equipment interconnecting these routers, with the goal of providing reliability through node and path diversity. However, adding such redundant resources increases the overall cost of the network. It is desirable to examine ways to reduce this cost through a fundamental redesign that avoids the use of such redundant resources. We seek to achieve this by leveraging the capabilities of an agile optical transport network. In this paper, we propose a design using only a single router at each office but exploiting the agile optical transport layer to carry traffic to remote Backbone Routers (BRs) in order to survive outages (both maintenance and unplanned failures) of the single local BR. Setting up a connection to the remote BR only after a failure has the undesirable effect of suffering from lower performability (a metric combining performance and availability). This is because of the time it takes for the connection to be set up and for the IP layer (e.g., OSPF) to converge and utilize this link. On the other hand, setting up a ’full capacity’ second connection to the remote BR wastes this extra capacity during normal operation and translates to higher cost. Our design makes the trade-off of setting up a low bandwidth connection initially, and then uses the capabilities of the agile optical transport to dynamically re-size the capacity upon a failure. This allows us to multiplex the restoration bandwidth for different router failures and for carrying normal traffic under non-failure conditions. Optimal mapping of the local Access Routers (ARs) to the appropriate remote BR is determined by solving an Integer Linear Program. We evaluate network designs for cost and performability. We show significant reduction in cost for approximately the same level of performability as of current designs.