TechEX 2019 Sneak Peek: How SURFnet in The Netherlands is Reimagining its R&E Network with Automation and Orchestration
au·to·ma·tion: The sequential running of a simple task on one system.
or·ches·tra·tion: An iterative process which uses the plan, do, check, act cycle to maintain consistent state.
It runs complex tasks across many systems.
Various drivers are pushing national research and education networks (NRENs) and other networks into automating the provisioning of their networks. This is due to many reasons, including: a brain-drain in network engineering, faster provisioning, and more demand in flexible and different types of services compared to relatively standard (non- or semi-automated) layer 2 circuit and layer 3 IP services.
In achieving the best possible service, NRENs and networks in general are facing a choice. Do we automate or do we orchestrate? Or can we do both?
SURF believes that to stay relevant and provide the best possible service to our end user we must change the traditional network of manual and semi-automated provisioning, with tools like ansible, Perl scripts, or napalm and build a software stack that is capable of meeting the future demands of our customers.
Just automation, even though it is an important part of day-to-day operation, is not enough. In our long-term vision, we foresee automation far beyond only the network domain, including multi-resource orchestration such as storage and compute. We must have the capability to do more and be more intelligent. Making this possible is part of the reason why SURF has chosen to take a more holistic approach in provisioning network services.
Next to a fully automated network we also want to be able to keep track of the complete lifecycle of a customer facing service. In doing so we make sure all changes to the service, whether they are initiated by the customer, customer support, or our network operating center, are permeated through not only our network, but our complete set of operating and business supporting systems.
When enacting a change on the network the SURF Orchestrator manages the complete cycle of Create, Read, Update, Delete and at the same time provides a way in which we can plan the change, do the change, check the change, and act upon unexpected issues.
After about two years of work, SURF now has a fully orchestrated set of customer products. A case in point in the effectiveness of the orchestrator is how we migrate customers from our legacy network to our new network. With the orchestrator we are able to effect the change with workflows that cause minimal downtime for our customer along with a reliable and fault-free provisioning of the service on our new network for our network engineers.
On average a complex migration of an IP service takes about 6 minutes. This includes an email to the customer, physical migration of production optical fibers to new equipment, creation of the new service, and visibility in our customer portal.
We hope you can join us at our 2019 Internet2 Technology Exchange session, Automation vs. Orchestration, on Wednesday, December 11 at 1:40 p.m. During this talk we will explain our philosophy in more detail, and provide a demonstration of the software showcasing its capabilities on our production network.