Phil Scanlon – Solace – Cloud Expo Asia – Singapore
Phil Scanlon from Solace speaks LIVE to Andrew McLean about event mesh during Cloud Expo Asia Singapore.
Andrew: We are joined today by Phil Scanlon head of Technologies APG at Solace. Phil, welcome.
Phil: Thank you.
Andrew: Let’s start from the beginning. We’re primarily a UK audience. We’re in APEC and we want to find out about Solace. Tell us what you do.
Phil: So Solace, I’ll start from who we are. We’re a Canadian company, a bit different from most technology firm these days, have a different view on the world. We’ve been around about 10 years. We’re based in a very very cold place called Canata in Ottawa, which is just North of New York very different temperature from here. So it’s about 50 60 odd degrees of temperature swing when you go there. We basically create Advanced Event Brokers. So we’ve been doing this technology for about 10 years. And we cut our teeth in the trading for the platform’s very high volume, very low latency. We built some very very cool technology. That’s basically done us very well over those 10 years, but we’ve also more recently now been as that technology is more applicable to more use cases, we’ve now moved into software and cloud and kind of the hybrid Cloud kind of area so events like this to kind of come and talk about it have been very useful for us.
Andrew: It surprises me. Most people ventured into trading. You started right at the start with the trading technology that’s terrifying. The amount, I mean, we’re talking milliseconds this technology has to remain up and remain stable and solid. So you had a very very quick schooling to begin with and now you’ve gone into that I would say slightly easier thing of the general public. But, so I was looking around on your site and it mentioned something called event mesh. What is event mesh?
Phil: Yeah, so the concept starts with events. So we’re an event broker company. We don’t do these kind of events. This is not what we do, an event to us is something that changes in the world. So some places are trades, someone picks up a telephone, someone clicks on a website, you know, all of these things are natural things that occur so, you know those to us are events. Now the thing with the mesh is those events as companies get bigger and bigger or they do things like in a new cloud platform or their IT estate grows. Those events are very distributed so they can happen anywhere. So the whole mesh concept is really to allow an interconnected series of Event Brokers which we provide to allow the participants, so the people that are clicking on websites the back end systems that need that feed of data to be completely decoupled. So you talk to your local event broker and then Solace takes care of getting that information to where it needs to get to so the mesh is really around making it simpler for people to work on the problems like the applications rather than worrying about data distribution, latency, buffering, moving lots of information around at scale.
Andrew: We’re obviously at the Cloud Expo in Asia and would, although cloud seems to now be a bit of a, it’s an accepted term and people have moved on and started talking about AI and things like that. Hybrid Cloud. Still a prominent conversational point. And you deal in private power, don’t you? That’s essentially one of the things you do. How have you found the trends around hybrid cloud. Why have you been involved in it? Why not just put everything in one place. What’s been the logic behind that?
Phil: Yeah, so I mean a lot of its around innovation and so, you know, we started off purely in the internal data center. Our customers then had the opportunity, we’ve got a lot of customers that have got a very big data centers and they’re now thinking that you know the cost of running those the cost of operating those, there are some Cloud providers that are giving them a better option there. So they’re wanting to run some of their compute now in that cloud environment. So our role is to move data around so we need to make it easier for people to move into those cloud environments. And we do that using standards. So Open Standards, the idea being that when you move into a cloud environment there are native event technologies in these clouds, but they’re all proprietary to each cloud. So if a customer invests in that cloud provider, invest in that technology if they look to move that compute back into the data center because people will over time or they’ll move it to another cloud provider. They can get stuck. They got to rewrite systems. So the whole idea with using Open Standards is you use an Open Standards Broker in these environments that decouples the actual applications that are running there from the actual platform. It means that you can move things into that cloud very quickly, but more importantly if you want to get out you realize it’s cost you too much. You’ve moved the wrong workload into the cloud. You can get it out or you can move it to another cloud so you don’t get locked into a single cloud provider. You’ve got that arbitrage, you can do that. And the other piece there as well as that you’re being Canadian and being very very open and very very, not wanting to lock people in our technology is replaceable. So we use the Open Standards. You can take our broker replace it with an open source or another technology. So if we don’t innovate if we don’t remain the best broker, you can replace us without having to rewrite all the applications which is historically what’s had to happen in our space. So that’s a real driver for innovation and allowing people to get in and out of cloud environments. The other piece is around with things like AI and ML you have your normal flow of your business. You want to tap at that data and get it in and push it through into an AI environment. So Google cloud has some great AI ML technologies. So does Azure also does AWS and they’re all kind of competing and growing at the same time. So what we’re seeing is people kind of they’ll move some of their computer into there, but it’s all the data is the same. It’s the same event so you can tap into the events that are happening the rest of your business and bridge them across into the cloud environment bridging into one environment for AI in a different environment for ML or if you decide that, you know invest in technologies, like pivotal Cloud Foundry bring that on-premise and do private cloud where you know, I kind of thought the infrastructure will be cheap, but actually it’s cheaper if I run it myself. People are doing that and we’re very close with people like pivotal to be able to run in their environments too.
Andrew: So you’ve got event mesh in one hand. This is my event mesh. This is an API. Is there a difference?
Phil: Yes, so APIs historically have taken a you know, you request some information via an API. So historically that data sort of sitting in a back-end database or sitting in an application and you kind of build this API around it. Very request response, but you run into challenges as people evolve and a lot of the traditional API vendors are starting to see this as data isn’t at the center of the world. So Gartner have kind of done some work on this and they think about 70 percent of all new technology will be event first, event driven. So, you know, you do need to get data into some system of record but really events happen and you push them to where they need to be processed. So whereas an API takes a kind of a query kind of mode, eventing takes a kind of a push but it also supports that the query type as well so you can kind of have the best of both worlds. With an API, you typically don’t have the ability if a back-end systems down you kind of you push that all the way back to the client the client gets told sorry that something’s broken and it can be quite embarrassing to someone who’s API is down. With eventing and with using an event broker in the middle. You basically can have an API type request response, but you can actually hold that data in motion. So if the back ends down or if it’s busy you can actually queue it up and then handle bursts. So we had a very interesting use case with the tax department in India. They had an API centric approach to filing your taxes as everyone knows comes to the last day of the month or last week everyone tries to file at the same time. India has a very large population. So that’s a huge load on some pretty old systems. So what they basically did was put a layer of an event mesh in front of all of that, allowed the front-end web sites to submit. We hold on to that data so we can take data at the rate that the front end sending that. We cache it up. We hold onto it and we drip feed it to the back end systems. So the last couple of years what used to basically cause a lot of chaos in the last three weeks of tax filing now there’s no impact to the citizens and there’s about three weeks of catch up on the back end that no one really sees but it’s actually you’ve got a better quality of service. And critically without having to rewrite all of your back-end systems to kind of change for this new world. So there’s a kind of just an example.
Andrew: Okay. Well, okay, so we read lots of examples lots of theoretical approaches. What about live data? Can you tell me some case studies? Some clients have got, things like that.
Phil: So we do, we have customers. To us an event and data it’s just information we’re moving around so. We have some customers that are taking things like video streams and then using those to fan out. So using our technology to kind of buffer that fan-out allow one publisher many subscribers. We have in Singapore live data, so if you’re traveling around on the transit systems here when you get on a bus those buses are all basically IOG sensor devices, so they have a whole load of metrics, position, speed, location, number of people on board, when the doors open, the doors close. So the LTA who run that traffic system, they have real-time visibility of pretty much the bus systems. That information is used though to power the information displays so when you’re a citizen sitting at a bus stop and you want to know what buses are coming. You can go onto your app. You can say I’m at this bus stop and you can see not just based on the schedule, but you can actually see the arrival times whether they’re wheelchair enabled, whether they’re full or not, whether it’s worth waiting for another one. So that’s kind of running now live here in Singapore. The next phase of that ERP2 is to basically take that technology and move it into the vehicles. So all of the cars will have real-time tolling so rather than having gantries in and out of the city.
Andrew: I saw this we were in a taxi and went was like a flash something out of the X Files.
Phil: Okay, so hopefully it wasn’t a speeding ticket, but so what they have is gantries and as you pass through them there’s a dumb device in the car that basically pushes back to say, you know, this is my vehicle. So you have a cash card in it and it deducts the card you got to go off and recharge that and all sorts of things so that system has been in place about 20 odd years, its up for renewal, so rather than having these fixed entry point and extra exit points of the city. They’re moving towards having basically real-time segments based, this segments busy. There’s no point in charging tolls for other parts of the city. This is the bit that I want to charge a toll on but I can actually warn people that are moving towards that that hey this bits busy, so if you take an alternate route, you’ll avoid the toll so really giving that much more granular visibility of the traffic conditions and providing a better service to the citizens.
Andrew: Amazing. Well, I also heard mention of American Express.
Phil: Yes. So we’re a very horizontal technology. I mentioned we kind of cut our teeth in financial services mostly on the trading floors, but more recent years company like American Express have they’ve got a massive estate of event technology based on some legacy from another vendor and they’ve managed to replace about three thousand servers. They’re halfway through replacing those. They’ve replaced 1500 servers with 12 of our appliances. So the data center space just to do that compute, but if you use American Express you swipe your card, you can pay with points every authorization every transaction goes over the top of Solace. We’re doing about three billion transactions a day with them. So it’s very very high scale, very low latency. Now, if you’re paying with points, you expect to see the point straight away. So we provide that kind of distribution of information a global scale for companies like Amex and we also do something very similar with Nets here in Singapore. So the payment exchange you can pay now for food at the food court with a QR code. So you take a photo of it it’ll basically ask you how much you want to pay the vendor, you pay the vendor and the receipt pops out literally within a second. So it’s that but that’s the QR code is independent of the backend banks. So my bank the merchant’s bank the whole payment system is all kind of choreography by Solace sitting in the middle providing that connectivity.