Why intent matters
Be on guard, we have a new buzzword; “intent”. Appledore have been advocating such approaches for many years, long-before the industry coalesced around the term intent. Why? Because we believe that this core operational approach is essential to achieving the industry’s elusive goals of agility and true, hands-off, automation. “Intent” fundamentally changes how we structure operational processes, resulting in far more latitude for automation to do it thing, but, conversely, far less certainty about what the precise implementation will look like. One of the industry’s biggest challenges will be coming to grips with this uncertainty – “letting the machines do their job”.
Intent is a useful handle, but also a misleading buzzword
Like most buzzwords, intent is both conveniently short, and misleadingly open to interpretation. Intent seems to be a fairly clear word in English. Yet, as we emphasize out in our recent Market Outlook, it is perfectly easy to define “bad intent”. The real challenge is to define a service at the proper level of abstraction. Too high and it’s useless. Too low and it becomes hard-coding.
A longer, less sexy, but better definition
Back in 2016 we discussed an orchestration approach that had three characteristics. Little did we know that later they would be called “intent based”. In it we argued for:
- Defining models that described the service in terms of results – we suggested SLA parameters
- Using rules or algorithms to translate these models into actual, physical instances
- Guiding the rules and algorithms with the current situation, which we called “context”. Examples of current context might be
- The address of the customer order
- Available inventory that meets the requirements
- Current failures and congestion.
This definition of “intent” demands that we define the outcome in abstract terms, sufficiently flexible to allow rules (or fancier algorithms perhaps based on machine learning) to do their job. It specifies an automated, yet non-deterministic method to translate from “intent” to “reality”. It allows reality to be guided by practical considerations such as the customer order, and the status of the network, including failures.
There will no doubt be a continuous string of technical improvements to how we model services, to the rules and algorithms we use to find solutions, and to the intelligence we apply to learning from past experience. Yet those three recommendations will likely stand.
The Business Value is enormous
This concept is crucial to both cost and agility. Abstraction allows for automated orchestration, which increases speed and lower costs. It also reduces the costly, bespoke engineering needed to anticipate every condition and pre-define every action (nearly impossible, and certainly costly). Finally, it allows the flexibility to use underlying resources in the most optimal manner, increasing the utilization efficiency of multi-billion-dollar hardware and facility investments.
So how well is the industry progressing?
Watch this space for an upcoming blog on how well CSPs and suppliers are implementing intent, and for our “Implementing Intent, Status in 2020” Market Outlook that covers both trends and the specifics of leading suppliers’ products.
You can read in more detail on “Why Intent Matters” here