Management, Provisioning, and Troubleshooting

Scalability affects management, provisioning, and troubleshooting design best practice guidelines such as knowing the minimum, average, and maximum number of equal cost multi-paths (ECMP) in the network and the frequency of Interior Gateway Protocol (IGP) changes within a given topology. Often, customers ask, "How scalable is LSP Ping?" The response depends on a variety of factors, such as the topology characteristics of a network and the device sending the probes with available CPU and memory. We look at a detailed analysis of scale factors a little later.

Management, provisioning, and troubleshooting assume access from the network operations center to the CEs. Although all CEs (customer owned included) should be under the management umbrella, the accessibility of CEs depends on the management model. There are multiple ways of managing CEs. For example, static routes can be configured to reach the CEs in a VRF or all CE and PE loopbacks can be in a network management VRF, called the VPN_Network_Management. Each approach has its advantages and disadvantages, and a detailed network management design is beyond the scope of this book.

The Service Provider Network Management station(s) originates from this VRF. Conversely, each customer VRF should contain the service provider network management station(s) to permit bidirectional communication between the management workstation and the CE router. By creating a management VRF, all CE routers can be managed from a single spot, and by virtue of the transitivity rule (that is, only routes that originate from the VRF are exported and routing separation is guaranteed between CE routers).

In terms of troubleshooting, a distinction exists between detection and diagnostics. Detection can involve using a mechanism such as bidirectional forwarding detection (BFD), and diagnostics implies the use of tools such as LSP-Ping or virtual circuit connection verification (VCCV) to locate and repair a problem in the MPLS network.

For design considerations, we note the following BFD factors:

• BFD over VCCVThe ability to send BFD packets for VCCV; this induces a requirement on BFD to scale to a large number of sessions.

• Distributed BFD implementations are needed to deal with the scalability requirement.

• An aggressive detection interval with a large number of BFD sessions increases the chance of false positives.

• If MPLS LSPs is fast-reroutable, the BFD fault detection interval should be greater than the fast-reroute switchover time.

Fast detection via BFD for MPLS LSPs and VCCV still requires additional study and operational input to recommend best practice guidelines.

We recommend you use tools such as LSP-Ping or VCCV for diagnostic purposes rather than at high frequency.

These tools are designed for troubleshooting and therefore can impact the performance of the network depending on the platform executing these tools, the number of ECMP paths in the network, and the frequency of IGP changes. The next section focuses more on equipment and network scalability factors that impact network design and the execution of tools and applications.