Hi Team,
we're expecting a large migration of 7600 PE nodes (IOS) to new A9k PE nodes (XR).
Customer realized manual config translation from IOS to XR will take ages, but automated approach also not trivial, due to config exceptions.
So the rough idea is:
1/ develop the NSO templates for all services flavors
2/ reveal anomalies by matching the NSO-generated configs against current IOS configs (can this be done with NSO?)
=> iterative process: (feasible or too cumbersome to modify yang models/templates/java code several times?)
- update the templates to cover the exceptions, or,
- cleanup that part of IOS config to comply with template
3/ import existing inventory into NSO
=> adjust existing HW inventory to match new A9k nodes (interface names and numbering, hostnames,..) (can this be automated somehow by querying external HW translation repository?)
4/ deploy the service to new XR A9k nodes with NSO
Does above make sense at all? :-)
Eg. not sure if red text above is feasible, or we better first cleanup existing configs with help of ansible, jinja, ..?
Any practical experience or pointers welcome.
The NSO approach (compared to any kind of migration scripts) is the future-proof solution.
Ie. all further service provisioning will be automated through NSO, hence no more exceptions and frequent human error…
