Automating Network Optimization, Part 3

Stephane is a customer of Packet Design. In this three-part blog series, he discusses how traditional approaches to network optimization can be improved upon with automation. In Part 1 of this series, Stephane explained two temporary network optimization scenarios requiring traffic engineering. In Part 2, he showed how the troubleshooting situation can be handled using automation. In this final part,  Stephane discusses how a customer request for a temporary increase in bandwidth between two sites can be managed using the Packet Design SDN Platform and Cisco XTC.

In Part 1 of this series, I outlined a customer request for a temporary increase in bandwidth between two sites for a major event. This use case must be handled differently from the troubleshooting issue I discussed in Part 2, as it is part of a customer service order request. It can be achieved in a similar way to the previous case, but this will require involving some backbone teams and manual actions within the customer provisioning process that may not be completely suitable.

Another way to achieve this in a more automated way is to couple the customer service orchestration/provisioning tool with the Packet Design Explorer SDN Platform (the Explorer products plus SDN extensions).

Consider that a customer has a pseudowire Ethernet service connected between PE1 and PE2. The service has an initial bandwidth of 100 Mbps, which is easily achieved by carrying the traffic on the shortest path (no specific routing needed).

Packet Design PCE with Cisco XTC

To address a specific event that occurs during a busy week, the customer requests a bandwidth upgrade to 2 Gbps for this particular week. The customer sends the service upgrade order to the operator using a portal or an API. The customer service provisioning tool or orchestration tool (e.g. Ciena BluePlanet) will be responsible for handling the complete service delivery. It will also drive the update of the shaping parameters on the CEs and PEs for each access connection. But it will also need to ensure that the network will be able to carry this traffic.

As the customer service tool does not have such deep knowledge of what is happening in the network (in real time), it needs to delegate low-level tasks to a real-time network analytics solution like the Packet Design Explorer Suite. Explorer’s northbound interface can be used by the orchestration tool to request this bandwidth upgrade. The SDN-TE app will perform the new path computation and place the tunnel on a path allowing the 2 Gbps of bandwidth. It will still interact with Cisco XTC to update the LSP parameters on the routers.

If anything changes in the network, such as traffic patterns, topology change, etc., the Explorer products will detect this and update the path of the LSP through XTC, if necessary (service orchestration is not involved anymore).

Use cases like this one can be fully automated while keeping control of the network. The real-time network optimization use case I discussed in Part 2 is real and can be addressed now by combining multiple technologies and products. It may still require human action to check if the proposed solutions are appropriate or not. This is mainly a question of trust between the human and the artificial intelligence that increasingly is embedded in the network analysis tools. We can expect to have more and more fully automated processes (self-healing) as this level of trust increases.

Explorer Suite | More Resources | Request Demo | Follow us on Twitter