Intent Based Automation Is Complex, But It Shouldn’t Scare You
In the last year the SDN/NFV hype peaked and people moved on to an automation focus in general. It always comes back to automation eventually, but this time a new term has gotten hot. Intent Based Networking, which is something I discussed in a blog about 2 years ago.
The beauty of the emergence of Intent Based Networking as a hot topic is that it is a byproduct of automation done correctly. Now, what do I mean by that?
If you do automation correctly you are moving to a model-based approach. You still may have scripts and other old-school automation in place, but you are migrating to the more flexible, dynamic capabilities that models bring. You are also introducing cross-domain tools that allow you to manage physical and virtual elements in a programmatic way. These tools leverage the models to abstract any complexities of the automation away from users, thereby allowing less skilled people to execute on MOPs and provisioning tasks, while your highly skilled engineers focus on actually being engineers.
Guess what it takes to effectively do Intent Based Networking? See if this sounds familiar. You need models that abstract the actual networking (i.e. complexity) from the Intent the user is concerned with. You may want it abstracted to the point a person just says, “I need a video connection from A to B that can handle HD video.” The model and toolset performing the automation translate this to the config required to satisfy the Intent.
The thing that concerns many is that the introduction of Intent to the automation picture GREATLY increases the underlying complexity. Your models get significantly more complex, even as your user base may get less so.
Shown here is what I think happens as Intent Based Automation gets put in place:
As you can see, the complexity escalates quickly, but a benefit is an almost equal decrease in the number of people that need to lay hands on the network to manage it. This complexity is still visible, but it is visible to a very small set of highly skilled engineers that will manage the models involved. The complexity those folks deal with is totally abstracted from a majority of your team.
Today’s staff is large and expensive because EVERYONE has to know how to manage the network via CLI. You are paying highly skilled engineers to be provisioners. As you get Intent Based Automation in place this role will shift to order management personnel, for example, and the customer can begin self-servicing if you choose. You will get more value from your less skilled employees, you will get more efficiency from your highly skilled engineering staff, and the return should be:
- Higher Quality, due to less human error as entry of key data is automated versus typed in
- Faster Execution, due to the automation of the process
- Higher Revenue, as you can turn up services faster and better which means billing sooner and happier customers that turn into your best marketing resource
- Lower Cost, as you can likely reduce staff, or, at the least, shift staff around to eliminate open headcount slots and not hire for them
- Accelerated Innovation in the network, due to engineers being able to focus on this versus provisioning customers
So, bottom line? It won’t be easy. Behind the scenes there is a TON of work to do to define models, train your staff to maintain those, implement the correct tools to enable an Intent Based Automation approach, and a shift in corporate mindset to trusting the automation you put in place. The work here is not trivial…at all. The results, however, will be worth it.