Monday, March 12, 2012

Integrated IS-IS Network Design Considerations

Optimizing networks depends on careful planning and design. Although each network is constrained by physical, technical, and financial limitations, network architects must strive hard to design a network to fulfill the needs of its users and the demands of various applications. Network designs are often required to compromise between the trade-off of reliability and speed.

Designing Integrated IS-IS networks with the hierarchical design requires the consideration of the data flow in addition to the bandwidth and CPU resources required by the routing protocol. Tuning the routing process to be more efficient allows the databases to converge more quickly, but often results in the compromise of resources and reliability.

Additional resources, eg: CPU, memory, and network bandwidth for propagating IS-IS PDUs, are required for L1/L2 routers that handle the processing of L1 and L2.

Below are some typical IS-IS network design considerations:
  • A totally flat network that uses only L1 routing. This design is not scalable because any change upon the network requires flooding of LSPs to all routers and trigger frequent router computation. However, this simplified design has some advantages, in which there will be only 1 link-state database and no problem of suboptimal routing between areas.
  • A totally flat network using only L2 routing. L1 areas can be easily added as the network expands. The L2 backbone has complete knowledge with the advantage of running a single SPF instance. L1 areas can be viewed as organizations connect to an ISP, in which new L1 areas would be added as new customers come online.
  • A totally flat network using the Cisco default of every router running L1/L2 routing. This allows for easy migration to a hierarchical design with multiple areas and avoids suboptimal routing when an L1 router always defaults to its nearest L1/L2 router. However, this design requires resources to maintain 2 separate L1 and L2 LSDBs.
  • A hierarchical network where the core is running L2 routing, with L1 areas connecting into the core. L1/L2 routers are used to connect the L1 and L2 areas. Although this is an excellent design, but there are still concerns that should be considered. This design results additional design and implementation efforts and the possibility of causing suboptimal routing. It requires an extensive knowledge of the network topology to ensure that a routing problem will not be compounded. The route leaking feature which introduced by recent IS-IS developments allowed allows L2 routers to pass specific routing information into L1 areas for facilitating the optimal routing decision.

Scalability is achieved by minimizing the size of the LSDBs and routing tables, the number of network updates, and the amount of processing. Implement route summarization whenever possible. Route summarization can be accomplished only where the careful address planning permits grouping addresses by a common prefix. This condition is true for OSI and IP. Therefore, it is very important to carefully plan the IS-IS areas, NET addresses, and IP addresses.

No comments:

Post a Comment