OGC Member Snowflake Discusses Benefits of Testbed Participation

Contributed by: 
Richard Rombouts

On January 12, 2015, Richard Rombouts, Senior Technical Consultant and Training Lead at Snowflake Software, talked with Ron Exler, Senior Consultant at the Open Geospatial Consortium (OGC) about Snowflake’s participation in the OGC and the OGC's Testbed 11, and how the company has benefited from participation.

Ron:                Why OGC? Why did Snowflake get involved with OGC and why is this process with the testbed important?

Richard:         For Snowflake, open standards enabled us to build our business. We're founded on open standards. 14 years ago Ian Painter and Eddie Curtis set up Snowflake, after careers at Ordnance Survey where they built the systems behind the first commercially available GML dataset – a dataset called OS MasterMap which still exists today.

Back then Ordnance Survey led the world and was the first big adopter of OGC standards.  Using open standards brought huge benefits that Ian and Eddie could see being replicated across many domains. From there, they founded Snowflake, and the company became an OGC member. Primarily we are a software products company. We build commercial data exchange software products that use OGC standards to join systems together.

Ron:                It sounds like it's an ETL (extract, transform, load) system for geographic data.

Richard:         ETL is a very techy term, we tend to use ‘data exchange’, that said we also have a little bit of data visualization via our free GML and ATM Viewers. Our GML Viewer is very popular - I think we’ve lost count but last time we looked it had well over 60,000 downloads.  The idea of our free viewers is to take away some hesitation people have about this big scary XML standard.

Our focus has always been on making data exchange simpler and that's essentially what we do, without the need for custom engineering.

Ron:                Why OGC? Why is that important to Snowflake?

Richard:         Mainly the “G” for Geography. Certainly, for most projects we come into, Geography plays a major role. That said we don’t define ourselves as a GIS company; we consider ourselves a data exchange company and geographical data is part of that exchange. One thing I like to say is, “spatial is not special, but it's not normal.”

Ron:                What do you mean by that?

Richard:         It's all data, but at times the geography part can be complex and needs specialist knowledge, for example, coordinate reference systems or geometric densification. That's where we come in and our software takes care of all the complexity - keeping it simple for the user.

Ron:                What do you get out of participating in the OGC testbeds?

Richard:         It's a number of things. First, the collaboration. You have the opportunity to work directly with other organizations and even your competitors. For us as a small business it's great to collaborate with larger organizations; working closely with them and showcasing how our technology solves their problems. Outside of OGC it can difficult to get noticed, whereas inside an OGC testbed it’s one team.

Ron:                You’re involved in the Aviation area of Testbed 11. How did Snowflake get into that?

Richard:         We started working in the aviation domain four years ago as a result of our involvement with the OWS-6 testbed. OWS-6 was the first testbed to address the Aviation problem domain and all credit to the FAA and Eurocontrol for sponsoring research outside of its traditional industry base.  For Snowflake the OGC helped pave the way to a new sector which today is a key part of our business.

Ron:                Right, but there's an opening for you. There was a need in OGC’s standards development efforts for what Snowflake does. That enabled you to step in and fill that role. They needed your expertise with the technology and you were able to learn the domain, participating in the process.

Richard:         Exactly. That's where the expertise from Snowflake around open standard solutions come into play, and it's something we see a lot in our training processes. We are one of a few providers in, I would say, the world, which have very specific training around open standards like GML.

We've seen it in the last year or two where we've started training customers in the aviation domain about open standards and we include a big section on OGC standards, the stack of standards of OGC and W3C and the world meteorological office (WMO). All of those standards bodies come together and paint the big overall picture. For many, it is an eye opener because quite often people have been working in an industry specific silo, defining their own standards. But once you go outside the silo, it just doesn't work.

Ron:                So the training is an important element of how Snowflake addresses standards compliance?

Richard:         Yes, training is usually the first step, educating the customer on the benefits of open standards and best practice is very important.  Participating in testbeds enables us to keep up and influence those standards, all of which make us a credible source of knowledge, not just theoretical but practical.

Ron:                It sounds like OGC lends you some credibility in the marketplace.

Richard:         Definitely. By contributing in the testbeds, it shows we are willing to collaborate. We’re specialists in our field of data exchange so we don’t do everything. We prefer to work in partnerships with other organizations to combine the best skills and we want to work with other organizations.

Ron:                Why open standards?

Richard:         Data sharing using open standards can result in huge savings. We’re currently running a project called XMAN with NATS (the UK equivalent of the FAA) where we use OGC Web Services to share flight arrival information from Heathrow with the surrounding air traffic control centres in France, Belgium, The Netherlands, and Ireland.

Ron:                The other European countries were sending information?

Richard:         In this case, NATS has a system in which they manage all the incoming flights into London’s Heathrow airport.  NATS only handles a flight once it enters UK airspace and given that we’re only a small island there’s often not enough time to organise all the flights efficiently. When this happens flights often get put into a ‘stack’ circling around until a landing slot is available.  This is complicated further when a flight gets delayed or arrives early.  Heathrow runs at 98% to 99% capacity so delays are extremely costly in regard to time and fuel.

Ron:                That seems to leave little room for error.

Richard:         There’s not, so the XMAN system is sharing arrival sequence data outside of the UK to stop the congestion before entering UK airspace. For example, for flights coming in over French airspace the air traffic controller in NATS could slow the plane on arrival much earlier to avoid having to put the plane in a queue and enabling the flight to land without the need for prolonged stacking. In the past this level of information was not available. The adjacent airspace ATCs did not know which flights might arrive when and at what time into Heathrow. That information is now being sent to neighbouring ATCs via OGC standards.

Ron:                So it's more efficient, so you don't have circling and waste of time in the air.

Richard:         Exactly. The trial has been running since April 2013, and NATS recently published some impressive operational savings.  Using OGC open standards has made the needed interoperability quicker and easier.

                        The OGC’s standards stack is a mature set of standards. It has good pedigree, tested all over in many different domains, so it made sense to implement these instead of custom engineering.

Ron:                What about Testbed 11 in particular, what's grabbing you about that, why is that important for you to participate in?

Richard:         It builds upon the last two testbeds, where we also participated in the aviation thread.

Ron:                What specifically is Snowflake offering to the aviation thread?

Richard:         In Testbed 11, one of the things we're working on is the Digital NOTAM enrichment service. NOTAM stands for Notice to Air Men. We work with Digital NOTAMs in our day-to-day business but the quality of the data could be improved on. When the requirements call came out for an enrichment service, but also for a validation service, we jumped at the chance to participate.

Ron:                What will you be doing in Testbed 11 that you see adding value to the testbed as well as something that you'll be able to leverage later?

Richard:         Our value add is the speed in which we can get things up and running. We have a lot of experience in providing OGC web services, so we’ll get those setup straight away to enable other participants to get started quickly.  In this case, we will be working with the sponsors to quickly get their data into the testbed for the clients to consume.

In terms of future leverage, we’re excited to be working on a new exchange model developed by Eurocontrol called AFX. We’ll be making it more aligned to OGC standards and seeing if it’s fit for purpose for other major users such as the FAA. If all goes well another global standard will be on the market and that’s good for data exchange software!

Ron:                You think through this testbed, you'll be able to show that?

Richard:         We certainly hope so. At the moment, a NOTAM message is, if you break it down to its very basics, a text message saying, for example, this particular runway is closed. But that message does not have any geometry, so we don't really know where the runway is on a map.  One of the things that the service will do is to make that connection between message and geometry. Then a participant in the testbed will use a client to visualize this bit of data, but we only have the data available using standard web services, standard encoding, and standard data models.

Ron:                So you're essentially taking information that exists along with the other information to enrich it so the stakeholders have new information faster -- either the pilots, the air traffic controllers, or whoever is involved in that.

Richard:         Exactly. It’s a value-add service, adding context and making it easier to interpret.  The old NOTAM service is only text based with no geography, the current Digital NOTAM service adds some geographic elements; we’re taking it a lot further and making it much more valuable to the end user.

Ron:                I see. An airline might need to know if a runway is closed, for example, because of wind or something like that, and the airline might have to adjust its schedule to accommodate that closure.

Richard:         Absolutely, but not just airlines – also airports, controllers, ground handers, logistics, planner – there’s a huge amount of stakeholders impacted by NOTAMs.

Ron:                So normally it would be done manually?

Richard:         Yes, paper-based data exchange.

Ron:                What are some examples?

Richard:         For example, a pilot’s flight bag contains all the information the pilot needs for the flight.  In that flight bag are hundreds of messages with information about the flight. Maybe a new construction crane has been put up or there might be a storm front coming up over the Atlantic. The paper NOTAMs contain advance warning messages of these changes.  The problem is they’re on paper and most of them are not relevant for the flight so the pilot needs to read through them one-by-one before taking off.

Ron:                Right. The human element is still quite intense. So Snowflake’s work to be able to standardize NOTAM messages could be a big benefit.

Richard:         The number of NOTAMs on a given flight can easily be in the hundreds, so the end goal is to take the human out of the loop and send only the relevant NOTAMS (based on the location and time of the flight) directly into the cockpit avionics.

Ron:                Is there a difference between airlines?

Richard:         Currently it’s early days for Digital NOTAMs and I’m not aware of any airline sending them directly into the avionics.  A lot of airlines are converting all that paper into pdf and putting it on an iPad but in reality it’s still the old way of doing things. Changing avionics is a very expensive and time consuming thing so hopefully adopting open standards will speed things up.

Ron:                It sounds like aviation, as an industry, has been highly dependent on the people, the expertise of the pilots and the air traffic controllers and the routing people at the airlines. You know, there's a lot of people expertise that goes into it, plus the paper maps and the paper sources that provide them with the information, but it's really more up to the people to put it all together.

Richard:         Quite often it is. Although one thing I’ve learned as well is that there's one big driving factor in aviation: safety.

Ron:                There needs to be some balance between the automation and the people element?

Richard:         Yes. I think that's another reason why these testbeds, like the OGC testbeds are so important. It's to showcase that these standards can be implemented. They’re mature. The industry can use them to solve problems. It's not science fiction.

Ron:                Part of what you're doing here is coming up with a scenario to show that for the testbed, right?

Richard:         Indeed, an OGC testbed always has a cross-thread scenario to bind all the testbed efforts (aka threads) together.  In Testbed 11 our NOTAM enrichment work will be part of a flooding scenario in the San Francisco Bay area.  For example, if San Francisco airport floods or one of its runways floods -- how do we notify flights inbound to San Francisco not to land or use an alternative runway instead?. Maybe San Francisco airport as a whole is closed, so planes are rerouted to Portland.  It’s all very well experimenting with standards but cross-thread efforts ensure those experiments work over many domains, otherwise you end up with standards built in isolation.

Ron:                Well, and in a real world situation, that's what has to happen, right? If there is a flood, all these pieces do need to work together.

Richard:         Silo approaches mean that each thread is created using standards but the threads themselves don't communicate. So it will be good to showcase that overall the Testbed 11 testbed actually binds the different threads.

Ron:                Right. Okay, so it sounds to me like Snowflake Software benefits in a number of ways by participating in the OGC and the Testbed 11 testbed.

Richard:         Yes, we're able to have a voice and have some influence on the standards themselves.

Ron:                Any areas of improvement that you would say … I mean I'm sure you've already probably communicated this to the right people, but are there any things where you would say, you know, “ If I was OGC I would do this different or that or ...”

Richard:         The atmosphere, which is very open certainly needs to be preserved. One thing I have noticed here is that there’s a very open atmosphere, even within the OGC, which encourages collaboration between participants. That’s very important.

I've been positively surprised. From what I've seen, it's very pragmatic, with the right amount of process. That is a key word. ‘Pragmatic’ is something I personally like, but also it fits in very naturally with how we operate within Snowflake.

Ron:                Thank you, Richard. Best wishes on your Testbed 11 testbed work.

Richard:         Thank you, too. Snowflake Software looks forward to continuing its contributions to the OGC.