Apstra pioneered the use of intent-based reasoning for network operators. Now, intent-based-everything has become all the rage, with widespread analyst acclaim and lots of vendors jumping on the bandwagon.
Over the past 3+ years in the business, we think we’re getting the hang of what it means to create real value from intent-based technology. We’ve deployed in the world’s largest service providers and enterprises. We have learned from customers in multiple vertical markets including tech, finance, media, insurance, manufacturing, mobility, services, oil and gas, etc. It seems like everybody is engaging and learning to benefit from intent-based network systems.
Our take-away is that it’s less about the underlying technology, and much more about how it is applied. Here are five ways Apstra has applied intent-based reasoning to networking.
Number 1: Infinite visibility. By this we mean the ability to define and collect any type of telemetry from thousands of devices on demand. A few lines of code is all it takes. See any data from any device, any vendor, any time, all at once. Normalize your data for consumption throughout the organization. It sounds a little hyperbolic no matter how we write it, but it’s true.
Number 2: We work with your intent, not ours. Vendors cannot foresee what data you choose to collect. Vendors cannot define the intent model for each metric. Only you can. If our experience makes anything clear, it’s that intent, when baked into a solution designed by a supplier, nearly always reflects only the supplier’s intent. Not yours. Telemetry and intent are not canned things. Expect the telemetry you choose to collect, as well as the intent-based reasoning you apply to it, to change a lot.
3) Multi-dimensional workflow. When you know what’s actually happening in a system, and you know what should be happening, you have the knowledge to confidently operate things. Natural workflows fall directly out of this knowledge; operations, troubleshooting, configuration and inventory management, documentation and compliance, design, etc. When combined, all these workflows form a kind of reference design for everything you do. We get you started with a lot of it, but again, workflow is yours - not ours (think Salesforce; works out of the box, but continuously customized by customers).
4) Multi-vendor, multi-system solution. Building an intent-based reasoning system to operate a box built by one vendor is just not that interesting. It’s more interesting to use intent to operate a complex system with lots of parts from lots of suppliers. For example, consider a self-driving car; all the intent a car needs is the address of your destination. It doesn’t ask the passenger their intent on how to operate a lot of separate sub-systems like the transmission, steering control, engine or tires. One system should service all types of devices from all vendors. And as for your network, we don’t know what devices you want to operate, who your selected vendors are, or how you choose to assemble things. We give you the tools to operate your entire network your way.
5) Respect your legacy as well as the future. Existing equipment and applications do not just go away because of a new telemetry and reasoning strategy. We need to instrument all the old stuff too. We need to support old L2 applications, old topologies, and old switches (think IOS, Junos, and all the stuff that you currently support with SNMP). Intent-based operation is never the reason you need to buy something new. Intent-based reasoning is an important tool to help you operate your complex network, with lots of equipment, both old and new.
So I guess our perspective comes down to this: intent-based reasoning technology is really cool. But value comes from how you apply it, not how you buy it.