CRN Interview: Apstra Global Sales Chief Butler On The Future Of The Multi-Vendor Data Center And The Partner Opportunity In Intent-Based Networking
Dave Butler, Apstra's global head of sales and business development, said 100-percent channel Apstra offers customers a streamlined solution for notoriously complex data center environments.
Poised For Growth
With a significant new customer, and the hire of a networking security veteran as vice president of engineering, Apstra Inc. is positioning itself for growth in the burgeoning market for intent-based networking.
Menlo Park, Calif.-based Apstra's AOS platform differs from other intent-based networking solutions in that it works to automate data center operations across the network in large, enterprise data centers built on the equipment of various vendors, including Cisco, Dell EMC, HPE Aruba, Arista and others.
The company's approach won over Yahoo Japan, which has deployed the Apstra AOS platform in what the company calls the first multi-vendor deployment of intent-based networking ever.
Venture-backed Apstra came out of stealth in mid-2016, and last month made a key addition to its management team when it hired Manish Sampat, a Palo Alto Networks veteran, as vice president of engineering.
Dave Butler, the company's global head of sales and business development, said 100-percent channel Apstra offers customers a streamlined solution for notoriously complex data center environments, and the Yahoo Japan win is clear validation of the company's strategy.
What follows is an edited excerpt of Butler's conversation with CRN.
What's the revenue potential for partners with the Apstra solution and how does it compare to what they had been making?
Their markup should be very material compared to what they're used to with standard networking products. They're competing with commoditized products between vendors at incredibly low margins. We would be an order of magnitude higher in terms of margin, plus lots of professional services. The channel can add huge value in terms of adding intent-based reasoning probes that check things that the vendor community would never have thought of. They can build libraries of tools that are completely unique in the marketplace. Customers are developing them faster with their partners than we can ship them.
Why is the channel important to Apstra?
We're 100 percent channel. We need to go through the partners, and there are lots of reasons for that. Everybody is thinking about their next data center revision. Partners bring us a presence at the right time, when people are looking to do this, so we're not in protracted periods of discussion with customers when they really need to be thinking about what's most important to them at the moment. The larger value is that all these areas: deep intent-based analytics, extensible telemetry, intent-based fabric management and multi-vendor management; these things still need to be thought out. Up-leveling engineers is a huge value for the partners to add distinguished from what they were doing in the past. We're just partnering with a few companies. We need to ensure we do a great job in every account.
What problems is Apstra solving that other intent-based networking platforms do not?
Networks have lots of problems. They're incredibly complex. We've built a simple solution that allows you to build a data center network and run it. Initially this was a way for us to allow multi-vendor devices and solutions to be deployed together in a meaningful way so people could extract the best value from each of the vendors. Over time, a number of other areas became paramount to what we were doing. One is fabric management. You'll see fabric management solutions from a number of vendors, but only for their products. We added continuous validation and monitoring to all products using telemetry that hadn't been anticipated by the vendors. This is a big deal and it's in an important area where customers are looking for help. Otherwise, people have to do these things by hand. The next thing is deep analytics with an understanding of the intent of what operators are trying to do.
What does that mean for partners, and how can your partner program help?
They have two things: One thing, when they come to us, they have insight when people fall into the traps associated with building new networks and trying to integrate fabric management with monitoring and continuous validation and deep analytics. Those things require programming. They require the construction of and specification of intent, and that's a big deal. The combination of that is why we need to have a strong partner program.
And when all that comes together, you get a deal like the one you just signed with Yahoo Japan?
Those are the things that allow Yahoo Japan to go from spinning up a data center pod in months to minutes. The level of agility change is that large. The ability to solve problems with continuous monitoring and validation against intent allows them to have a mean time to insight that's reduced by 50- to 90 percent. People have always looked at data center networks as a collection of switches that are stitched together by brilliant people to solve problems. What we've added is a framework whereby we can understand what this stitching should look like and we can stitch through all that fabric management, monitoring, continuous validation, deep analytics and that's the differentiation that people are looking for.
What problems was Yahoo Japan trying to solve?
A multi-vendor solution that does all of this and ties it together is a big deal. That's why Yahoo chose us. Yahoo is an interesting company. I think it's the largest SaaS company in Japan. They have the kinds of scale problems and agility problems that are consistent with the overall issues I was trying to describe. For us, it took a long time. They really, really investigated whether they could scale and bet their entire company on this. They're doing that. It's a proud moment for us. The level of scale in such a rapid fashion is important.
Big SaaS companies like Yahoo Japan, or Facebook aren't buying gear from the big legacy vendors. Are they even looking at intent-based networking platforms from those vendors?
Yahoo will buy products from name vendors, but not just one of them. They won't be locked into a solution that forbids them from considering other things. We consider all these switch vendors to be partners, and we're trying to accentuate the advantages they've got, whatever it is. Some of them have come to market with standard protocols quicker than others. We try to use those as quickly as we can. Yahoo does buy commercial equipment, but if you're using commercial vendor 'A' for a while in one area and 'B' in another area and you have a simple platform that runs them both in exactly the same way, the next area that gets retrofitted could go out to any vendor. You could use open source, you could use bright-box solutions. They'll interoperate seamlessly right next to the Arista stuff, or whatever is there. It allows them to compress the price of the whole networking stack. We can reduce the number of failures and we can do it in a multi-vendor environment. We fully embrace the Ciscos and the Aristas of the world.
In order to have Cisco's intent-based networking platform, you have to have its Catalyst 9000 switches, and you're saying you can have whatever switches you want.
Yes. The Catalyst 9000 is an awesome switch. They're really good switches, as are a number of other vendors' switches. There are basically three big areas in terms of innovation in the data center beyond just getting this stuff to work together. It's how do I treat the fabric as a single system. Cisco uses ACI to do that, but it only works with Cisco switches and only when they're running in ACI mode. The second thing is continuous validation and intent-based monitoring, and Cisco has a bunch of new products and of course they only work with Cisco stuff. They're great products. Third is deep analytics, checking for things beyond just whether this CPU is overloaded compared to that one. It's really thoughtful things that engineers design and change on the fly. They have an important product called Tetration that does this stuff, or some of it, but we've made it generic and more programmable. It's a play for the channel as well.