 Live from Palo Alto, California. It's theCUBE at Pier 2.0. Brought to you by the Pier 2.0 Foundations. Learn, connect and grow. Now here are your hosts, John Furrier and Jeff Frick. Welcome back, Jeff Frick here with theCUBE. We are at the Pier 2.0 Foundation event in Palo Alto, California. As you know, we like to take theCUBE out to the events, extract the signal from the noise, get the smartest people in the room. We can find, bring them on, ask them the questions that you'd like to ask them and really share the knowledge. So we're joined here on our next segment by Andy Davidson. He is the CTO of Allegro Networks and a self-reported network automation and peering geek. So welcome to theCUBE. Thank you, Jeff. So you just got out of your session, right? You gave a little session. So for the people that weren't able to join, why don't you give us a little background about what you covered in your session. So in this session I talked about the important aspects that a service provider would need to go through if they wanted to do automation on the network. Automation's really, really important because service providers want to give a consistent experience for their customers. They need to ramp up the number of customers they can deal with at scale and doing that through traditional means leads to inaccuracies. It leads to processes that don't result in happy customers. Automation's really, really important. And we were the first people to explore automation as the reason for the company, as a reason to experiment and innovate and try and reach new customers. So it's been a really, really interesting journey. I wanted to share that with the service providers. So how did you parse out that challenge, right? Because everyone talks about automating or automating everything. And at what point can you automate? At what point do you start to automate? Did you look for really problematic errors that are causing problems? Did you look for low-hanging fruit that was just easy to do? And then the other thing, once you started down that path, it generally happens is you automate one thing and that just kind of moved your problem to a different spot. We talked a little bit about how you set a course. So we started out by, when we established it, we wanted to build network services. We looked at why buying network services from the organizations that will become our competitors really, really sucked. And I've been buying these kind of products for around 15 years of my career and every single time it's been painful because you have to deal with 20 different people inside an organization. The way that the services get rolled out is something called a ticket moves between different departments. Each department goes out incorrectly. You're months from the point in which you need a service to where it can be delivered. And we worked out that with consistent business processes that lead to software, you can get rid of all of that. You can replace it with automation so that services are rolled out there and then for customers in a consistent manner. And that means they can unlock revenue today or do innovation today. That was the reason. And so how many steps did you guys automate in what was kind of the process before? So we took a lean methodology to the way that we started automation. So we worked out what's the least amount of work we can do to test the concept. So we started very clearly with one service which was a provider in a data center and one town needs to connect their network to a data center in another town. Can we get that stage completely automated? Can we allow them to go to a web portal, define what they want, and roll out that service for them? So we had this full suite of services that we wanted to automate like IP transit, interconnections and internet exchange points and plugging into other parties over the network. But we started with that simple piece. Can we take this product and make it a simple step for automation? And just in terms of a scale and the importance to your business, how often were you doing that type of a task? Well, it would be more than once for every single circuit on the network because you have a point in which you make the circuit for the customer between building A and building B. But the customer's requirements change and evolve over the time. They need more capacity or they need to do maintenance and change it to a different technology. So every time you have a circuit, there's a minimum of one thing that you have to do with that circuit. There's normally two, three, four over the course of two or three years. So it's literally every single time we come into contact with a customer. And how many steps were in that process when it was the ticket process and how many people did it touch? So you'd normally have to speak to a salesperson, first of all, who would spend a week dancing around the handbag until you really, really needed that circuit and then wouldn't try and get a discount. And then it would go to some kind of assurance team who agreed that the circuit was possible to deliver. There was capacity on the link, even though you'd agreed a price and a service. Then it would go to operations who would actually install, roll out the service. And then you'd end up dealing with a network operator center to go and put right all the things wrong that the provisioning and operations team had done when they built your circuit. So it could such, a circuit provision could such five or six people on a small service provider and it could such five or six teams on a big service provider. And now you can such one by portal and have the service live in five minutes. And you've got a lot of interconnects and interdependencies, right? With other departments, other entities, you've built that all into your software. Yeah, a perfect example here, somebody who maybe sells cloud services, cloud hosted applications, or people who want to interconnect to the big cloud providers like Amazon, Rackspace, Microsoft, they need to turn up circuits quickly because they're fueling innovation inside their organization with their ability to move data into the cloud. And yeah, that's exactly how we approach the subject of dealing with multiple providers. As long as you have a simple portal and you behave in a consistent manner every time, then those business processes turn straight into software that delivers the services people need. So talk a little bit about your CTO, it doesn't say, VP engineer, you're not, I'm sure you guys have plenty of software in your products but you probably don't have kind of IT services software. So talk about, did you have to kind of become a little bit more of a software company within your service provider arm? And how was that, do people like that? Did they grasp it? Were they excited about it? And then how does she actually execute and deliver a successful commercial software product? So it becomes, it's a great question that because there's absolutely no way you can build a project like this, build a product like this without really, really good cross-department discipline because network technologies, network configuration, network services that can be complex because the rules you have to follow if you're going to have a service that's compatible with what every customer needs. And there's a lot of things that are possibly not logical to people that don't really understand internet plumbing, internet core engineering, like when a peering decision will be a yes, when it will be a no, when a decision to interconnect will just use a VPN over the internet will take place. And so we hired some really, really great software engineering people that have worked in web portal design and complex application construction but not in the networking space. So we had to work really hard to make sure that knowledge moved between the network engineering team, the service delivery team and the software development team, absolutely. Otherwise we couldn't have built the platform that we did in the time that we did. And I know our software developers joined in June of last year and we launched the minimum viable product in September of last year. So we rolled out an entire automation suite in the time it takes some carriers to roll out a circuit. Right, right. And how many people in that team just oars of magnitude and the software development team? Software development team is two people. It's two people. And across the organization we have 20 people and we look after a data center spanning the length and breadth of the UK. Right. So you talked about delivering the, kind of the laying fruit, the minimum level. Let's get it started. Let's proof of concept. So how many more services have you automated via this path so far? So. In September you said, right? Yeah. It was the first one rolled out and so now almost a year later. So we looked at the core service provider building blocks that every service provider needs to buy so that we could start to push those out automation as well. So connections to internet exchange points. You can do that through a process that's automated as possible. Connection to IP transit, which is a wholesale connection to the internet at large. But also the really, really interesting thing that we wanted to work on, the reason we did all of this work first of all was to allow different providers, either two different companies, provider A and provider B or service provider A and cloud provider B to plug into each other over our network. Which in terms of engineering complexity isn't a great deal different to building a circuit between a single provider's network. But of course there's so much business logic involved making sure that the technical parameters that two competing organizations require for the circuit come together. And all that has to be done through the portal and we mirrored the type of conversations that different service providers have when they come to agree to make an interconnection through the stages through our portal. So using our portal to make a provider A, provider B circuit is a lot like having the conversation with the other provider, which way you'd agree parameters for turning up the circuit. And how well were you able to replicate that in software? That sounds like much more of a people negotiation and necessarily a trade off software. Yeah, well it's a lot like social networking in some ways. So when we first started demoing to our software development team just how this conversation works, essentially two guys who work for service providers go and meet in a bar or a conference like this and say we need to plug into each other and enable this service. We drew out the steps, we role played the steps, we wrote down the steps and our software development team looked and said this is like a Facebook friend request. So we could borrow concepts from social networking to use for provider A, provider B interconnection because the conversation is quite a lot like hey shall we hook up on Facebook? Right, that's terrific. So how has the success in this kind of toe in the water if you will process in terms of developing software to enable this connection between companies and underlying technology there's really more about connecting companies. How has that changed within your group and your team and maybe your senior management in terms of opening up new opportunities that maybe nobody ever even thought of before? So that's another key reason why we had the vision of let's cause circuits to happen at the pace in which people can innovate because that will maybe cause them to innovate faster and we certainly see it's always been the case that you can get cloud big data services available in an instant or CPU compute power available in an instant and having the network capacity to join all of these different instant delivered services together at really, really high speed has unlocks the abilities of push data into those environments and we certainly see customers doing this. We have some customers that work in e-commerce who wanted to move data from their own databases which are stored in a big tin inside traditional data centers. They wanted to move that data into a big data cloud provider and instant deliver big data company and they found that they had a challenge how they were going to get the data out of their infrastructure into this cloud service provider so that they could analyze it and better understand their customers and prioritize their own work patterns. They wanted to innovate in the way that they did business and they struggled to do that because they just couldn't figure out how to move the data around and we met them, we showed them how they could do it really, really quickly and tie into a big data platform that's on our network and they could see right away, this is how we can move the data around this is how we can do it and they bought the additional service straight away and they used that to power innovation inside their own organization in ways that we didn't necessarily think would be what people use it for when we started for this but we always knew that we would end up having these conversations over and over again and finding new ways for people to work. We knew we'd have them over again. So is this service provided as part of their core services that they buy from you or is this an additional product? So are you actually launching a new product line with the service? It's, every customer sorry is able to order every product so if you're a service provider, if you have a network you can log on to our portal and that allows you to buy all of the services. But the portal itself is not for fee, it's just part of what they get when they're doing business with you, right? So it's an enable or it's infrastructure. Absolutely because we encourage and we want people to use the portal because then they can deliver the services themselves right away. It's not business hours in the UK right now but one of our customers who needs an interconnection to another customer or somebody who needs extra capacity between two data centers can log on to our portal and enable those services right now. It doesn't matter that it's not business hours, it doesn't matter that our sales people are at home. Hopefully they're at home having a lovely evening, right? Because the portal's doing all that work for them. So a customer, we specifically want a customer to have access to the portal and to use the portal because that enables the instant delivery so they get it for free so they have any service. Right, so you touched on it a little bit about this service that you're providing which is really just a simplify and remove human errors and expedite things at the speed that people want it done. Saves money, but we talked a little bit about Off-Fair how it's really empowering your customers now to innovate in ways that they couldn't do before and you talked briefly about this company now being able to take advantage of a big data analytic process or service so that they could better react to their data. What if you can just share any other anecdotes about how what feels like kind of hardcore plumbing really is enabling innovation at a different level and not really, it's not a cost savings, right? It's an unlocking additional value and the other interesting story is you can share. So it's really that we have, for example, hosting companies on our network that are part of the solution and there are customers on other hosting companies who have done other innovative things and plugging those together is the innovation. So somebody who has a portal that can analyze web traffic logs, we have customers, customers who have services like that that can do that in real time and look for new insights to do with web performance or customer trend analysis and having the private link that can be built over our network means that some totally different hosting company who never had that idea can plug into that service provider and start to use that service the day that they've discovered them. That's the thing that allows people to do it. The innovation right now is a time-to-market thing, I guess rather than fueling new ideas but I know we have customers who are working on brand new ideas that having instant delivery network can enable. Right now it's a time-to-market innovation. It's a time-to-market matters, right? I mean, at real time, whatever real time means we talked about that a little bit in our last interview. It means something to everyone, not necessarily the same thing to everyone. So let's shift gears a little bit. Why are you here? Why is this an important event and what brought you here? So I was really interested to find out what people who are new to the peering industry wanted to know. So because I've been doing peering for over 10 years and from my perspective it hasn't changed enormously. The speeds have got bigger. The number of people you can interconnect to has increased but the really, really interesting thing here is what is it that the guy's working on the brand new networks, the brand new service providers? What questions do they have? And that allows us to put that into our peering products because the questions they have, if we can answer them through our portal then it becomes easy for them to do business with us. So simple. And are they significantly different than what you expected or what you've seen over the last 10 years? What are some of the differences with the new guys? The new blood that's coming up? So I see people looking for greater simplicity because for me it didn't seem, it didn't seem odd when I joined the industry that there were the complex lead times for circuits and it didn't seem odd that it was the case that you had to literally go and meet 100s of people to get the interconnections you needed to make your network perform optimally. And some of the best questions that I've asked people or the people in the audience have asked from academic or people who are moving into interconnection roles at service providers, some of the best questions they've asked is, but why? And when you start to answer that question, you start to realize that actually there are things that we can do, there are steps that we can take out of the interconnection process and logic that we can build into the portal that answer that but why question on the lines of, well maybe you don't need to do that anymore, maybe it should be the case that certain things are taken as forgiven and I think it'll inform the next stage of our portal development so that we can start to ask, but why more often? And I think that's been the most interesting thing. I just love how we often talk about the consumerization of IT and it's really all about expectations, right? And now you're talking about a new generation and as we said a few times before there's a huge representation here of students and young people who have a very different expectation about how things should work, what types of response time they should get and just how easy things should be without necessarily the full appreciation of what's happening behind the covers, but that's okay because they know best of breed application delivery as evidenced by Facebook and Amazon and Google and the things that they interact with this sets the standard for what they hope everything else will act like. Absolutely. Interesting. So I want to shift gears one more time before we let you go. So your soft-spoken guy came over from the UK. I just want to ask, you know, we think of the internet as global, right? We have global companies, we're on the news, we see what's happening all the time, but talk a little bit about what's different in the UK and in Europe in general from what's happening here in the US. Get your perspective on that. Well, that's, I could, I can really unlock the whole half hour conversation if I mentioned things like net neutrality and the tendency to interconnect in general. A lot of the, there's been a lot of time in our press about conversations that big networks in the US have with big content players in the US about who should pay and who shouldn't pay for access to each other's network. An interesting difference between the UK and the US ecosystems is that it's much more common for a UK or a European player to agree to make an interconnection with one another much more likely that they'll make an interconnection to a content delivery network provider because they understand that both sides are having a big win in terms of the customer experience. Right. And I think that that's, it's led to a market in Europe whereby it's really, really easy to build service provider. It's really easy to build an ISP. It's really, really easy to introduce content and serve. And I think that it's done great things for competition and therefore pricing for consumer services. It's done great things for increasing the speed by which content innovation can happen because there is this greater tendency for competing organizations to work together and build an interconnection. I'd love to see the US market open up in that way and it's become more likely for people to build bigger interconnections between the networks because it will result in performance win for end users in North America. Now what's interesting here is that, the Flashpoint use case is Comcast and Netflix which is getting all the pub. But what's ironic is the cable industry used to be extremely distributive. There was a ton of mom and pops because nobody knew what was going on when there was the big network system and Comcast and Time Warner and a few others have kind of aggregated that over time. So now there's not that many providers as there used to be. So has that not happened in the UK in terms of the service providers? Have they just been able to maintain a much more diverse population and they haven't had that roll up? I mean do you think that's kind of a big difference? We did have a very, very similar cable TV roll up just as you've described here. There was, he used to buy cable TV from your city's cable TV provider and that's not the case anymore. But in some places in North America you have the option to buy DSL from a relatively small number of providers but pretty much in every town in the UK you have a choice to buy service from over 100 ISPs. 100? So essentially that traffic has got to get between all of those ISPs somehow and there are so many people that sell on the incumbents wholesale platform. There's so many ISPs who've built their own circuits, their own last mile. There's the cable company, there are wireless providers. There's a great choice in most cities and there are literally that many ISPs that you can choose from. And did that cover the density of the population? I mean it's always kind of the last mile problem. I mean there was the regulations where a lot more ISPs had access to the wires into the home. I mean it seems like an environmental structure that would enable that. Yeah there's definitely relatively open access to last mile connectivity that's owned by the incumbent but it's also been relatively simple not simple, it's always been expensive and difficult to build networks touching so many households in the city but there's always been a culture that's allowed people to dig new fiber into the streets, dig new last mile into the street and in most cities you also have a choice of a number of wireless providers. I don't mean cellular providers here, I mean people who can do very high speed Ethernet services over Wi-Fi style equivalents. Dedicated Wi-Fi network. Yes and of course we have fierce competition in the cellular space as well. And we're seeing a lot of young people buying just cellular service, not buying fixed broadband into their first homes or to their student households. We're seeing the substitution of fixed line services. Yeah here the kids are back at slide TV. They got a bunch of kids downstairs. Yeah. Well the point is they'll be buying, if they were in the UK they'd be buying different services to what we were buying to get into the access and telecoms into the homes. This is interesting so we're gonna have to get you back on we're getting the hook here to go to our next guest but this is a really interesting conversation because it's really more of an infrastructure situation that enables what's now becoming a pretty hot topic here in the US. All right so Andy thanks for coming on Andy. Davidson CTO of Allegro Networks. I'm Jeff Frick you're on theCUBE. We're at the peer 2.0 foundation event in Palo Alto, California. We'll be right back with our next guest after this short break.