WTRS Executive Interview

Interview with Robert Van Andel, President and CEO of Allegro Software Development Corporation.

October 24, 2008

George: Can you tell us a little bit about yourself, your background and what led you to found Allegro Software?

Robert: I had a position as engineer and sales systems engineer for a long time in the high tech industry. Gradually was looking to start a company but was looking for the germ of a good idea. My last job prior to founding the company had been as VP of Engineering at a small company building widgets, Data communications widgets, and in some cases we did our own development and in some cases we bought outside consulting services and in some cases we bought software components. When I was looking to start a company the notion of software components became pretty interesting.
In 1996, when we founded the company, the World Wide Web was just starting to gain attention. Browsers had just gone graphical and were starting to be used pretty widely inside Enterprise environments but not that much still on a personal basis. The number of sites that were out there that were public sites compared to what there is now was dramatically different.
The Internet had of course been around for a long time being used with ftp, email and things like that, but the web was still a relatively new phenomena. My cofounder and I got the idea of building software components to enable devices to talk to the web. The notion of being able to use this relatively new thing, the graphical browser, to manage devices was sort of a key founding concept.

George: Was the intent from the start to provide these web service to embedded applications or did you have a broader scope?

Robert: It was all aimed at embedded devices and it was all focused on the thought that tying these small limited resource devices into the larger web would be a benefit in terms of ease of use.

George: What is it important for us to understand about Allegro Software and the company's products, now that we are twelve years down the road?

Robert: We started with an embedded web server product aimed for building comprehensive web-based management applications. So not only was there an HTTP server, there was a page compilation kit and it was intended to build very complex web based management applications but in a very small resource limited environment.
That was the initial product. From there we have brought into a greater variety of web products an email product, security products and more recently consumer electronics and home networking products.
The thrust of all of these has been aimed at connecting embedded devices to the Internet. The thrust of the company has been providing software components. We don't provide end product solutions, we're not really a services company, we are a software products company.

George: If I am involved in the design and development of almost any kind of embedded device and I want to enable web services on the device I would come to Allegro Software to get the tools to accelerate my ability to do that. Is that a good summary?

Robert: That's a great statement.
When you go to market with a product you have a lot of choices to make and very few people write their own compilers and very few people write their own operating systems and very few people cast their own silicon.
At some point you do tradeoffs and you say where are the key things where I can add intellectual property and where are the key things I can buy off the shelf? And in many, many cases there are still people who write their own compilers because somehow they think that is part of their critical expertise. The average widget manufacturer isn't going to do that. And in some cases they will do specialty silicon where they are building custom silicon on chip. More and more they are buying system on chip devices from merchant suppliers and adding their own value.

George: In the software side of the equation this is really the same approach as someone that goes to a merchant semiconductor manufacturer for their chips can come to Allegro Software for the base software technology that they are using in their products.

Robert: Absolutely. We are in parallel with people that provide operating systems, TCP stacks and other sorts of specialty software components. There are still people who roll their own but they can generally get a device to market faster if they are using pre-built components.

George: Your products operate across a broad number of platforms and on a broad scale. They range from fairly small memory constrained, power constrained devices to things that would operate in a data center or some other industrial environment where that power and size and memory constraint isn't as critical.

Robert: We have some applications where people are running thousands of simultaneous connections through our software but more typically we are in smaller devices that are only running fives or tens of connections.

George: What is the key to enabling this functionality?

Robert: One of the things that are the kiss of death in the embedded environment is memory leaks. From day one we designed the product to be configurable in terms of its performance and memory usage. We avoid dynamic memory allocation like the plague.
There are really a couple of design constraints that we approached with the initial product and then applied to all of the rest of the products.
The first constraint led to some design solutions but it was really more of a market based analysis that said there was no one processor in the embedded space that was dominant. And therefore to be successful as a company we couldn't focus all of our efforts on one particular processor model or one particular operating system model.
In conjunction with building the first deliverable product, the web server, we also built a very powerful abstraction layer to make sure that this web server could run on any device. When we built other products, web clients, an email client and so on, all those things were built on that same abstraction layer. In the first year or so we had to tune that abstraction layer to prove that it really met its design concepts but it's been rock solid ever since.
The second design principal was highly configurable memory usage so that somebody who only had a few tasks to run was not going to pay the memory penalties of someone that had a lot to run and vice versa.
We came from a communications background and had some knowledge about writing state machines and predictable performance. The real key to scaling up is to be able to add more memory in a discrete fashion.

George: The conversation we had at ESC was around the forecast that "Trillions of devices will be connected to the internet by 2010" missed by a couple of orders of magnitude but that doesn't matter because the real number is still a very big number and represents a very high growth rate. How has Allegro Software been riding this growth wave and what do you see this meaning for the future of your business and the future of the industry?

Robert: I think that when some of those predictions were made they were looking at what was called Internet time and how quickly people were adopting new ways of doing things on PCs and sort of missed some of the issues related to both infrastructure and the costs and conservatism associated with building reliable devices.
If you are going to build a device that has to operate 24/7, and doesn't have a human that can recover from the blue screen of death, your design principals are a little different and it takes a little while to roll those things out.
The key thing as we mentioned at ESC was the deployment of world wide infrastructure happened a lot slower than many, many people forecast. It's now starting to ramp up again and you are getting dark fiber lit up.
The nature of an infrastructure in place that is always on, and the lack thereof, really slowed down the proliferation of devices.
That is changing now. The definition of what broadband is keeps going up and up, the number of people with high speed broadband keeps going up and up.
We've been focusing on CE devices, Home networking; because that's a pretty exciting and interesting area to work in but it is just (now) being enabled by the high speed infrastructure.
YouTube is still not all that attractive in terms of video quality. If you want to watch YouTube on your brand new big screen HDTV you will be pretty disappointed. But it illustrates a concept of the huge amounts of content streaming into the home from a variety of different places. The Apple TV box, Microsoft Xbox Live Video movie services are early, early efforts still. This is not an ingrained way that people are grabbing movies.
There was just an announcement, though, the other day that the Apple TV box now has the rights through the music store to get movie releases at the same time as DVDs.
What we are seeing is a real trend in media delivery using Internet which leads to a whole set of media related devices.
That's what they call the killer app. That is the thing that really gets the Internet solidly in the home for more than just PCs to the outside world and connecting printers. Once you start distributing media around the home you are going to start saying "Well, why not alarm services, why not meter reading services, why not a whole bunch of other things leveraging on that distributed network in the home?"

George: From a technology adoption standpoint or the first enabling application into the home, the consumer electronics, DLNA home media focus is just a start. There is a host of other services that can piggy back along on that existing infrastructure and it seems that all of those other services will require some form of embedded web services to be running on those devices.

Robert: Or it could be embedded Internet services. It may not be just a web connection. You can think about a device that sends you an email alert when something critical has happened and that can be either a message that comes to you in your home or something that comes to you on the road. Again, assuming that you have that infrastructure in place there are a variety of protocols that are possible.
Those networking applications have been being built for a long time. The UPnP forum and the initial sets of protocols was formed late 1999 and really got going in early 2000, so people have been thinking about these problems for a long, long time. But it is just starting to hit for consumer adoption now. Once again, when that hits then you really have a whole set of framework laid, in many cases the tougher problems will have already been solved and adding on a coffee pot that provides an alert when it's used is a relatively easy thing to do.
I pick on that particular example: Why do you want your coffee pot to let you know when it's being used? You don't necessarily but its one of those things where you may want to give the coffee pot to an aged relative and use that as a sanity check; as a heartbeat pulse to sort of politely monitor their health. Or have a refrigerator door opening transmit an alert. So that if in fact you don't see signs of activity at an aged parent's house you can follow up with other kinds of things.
Those networking applications have been being built for a long time. The UPnP forum and the initial sets of protocols was formed late 1999 and really got going in early 2000, so people have been thinking about these problems for a long, long time. But it is just starting to hit for consumer adoption now. Once again, when that hits then you really have a whole set of framework laid, in many cases the tougher problems will have already been solved and adding on a coffee pot that provides an alert when it's used is a relatively easy thing to do.
That's a pretty obscure application and it doesn't really make a lot of sense on its own, except it becomes a very easy incremental add and an easy way to add a pleasant monitoring activity if that infrastructure is already in place.

George: You wouldn't necessarily deploy the infrastructure to do something of that nature just for that function. If the infrastructure is already in place, it's easy to find additional functions that would use the infrastructure. To tie it back to what you are doing at Allegro Software, any of those devices are potentially a candidate for your tools.

Robert: Yes. The other thing that happens is homeowners with very large amounts of money on the high end can pay for and build out customized security systems. What Internet technologies tend to do is allow displacement of custom things by using these commercially proven protocols that are easy to add in increments.
Adding a web cam into an existing home Internet infrastructure to be able to add security protocols and mapping becomes much, much easier once that infrastructure is there. Again well-heeled homeowners can get all those things now but they are highly expensive and highly proprietary. Here you can tie those things in and because of the leverage of the infrastructure, actually obtain some services that aren't easily available now, such as being able to monitor your home remotely or get intruder alerts remotely.

George: Are there any questions that I should have asked that I didn't? Can you help me here and make me look a lot smarter than I really am?

Robert: We've talked a lot about the home and the home is certainly a blossoming area, and there are things that are sort of ancillary to the home such as medical equipment and home monitoring where you take advantage of that infrastructure in the home.
The real issue is that we are starting to see this with mobile devices and infrastructure build-out all over the place. I know you guys are heavily focused on the wireless infrastructure, WiMAX and all that. That really enables a different way of looking at devices and that network of devices.
Microsoft just announced Live Mesh, a set of services in the cloud for all sorts of things attached not just PCs. What is interesting is you start getting those kinds of back end services on the Internet and you start getting physical layers like WiMAX; you suddenly have a cluster of devices that can be everywhere not just home. We've been focusing on using standard Internet protocols to enable devices to connect to the net and add a variety of different services. That's been our focus.
Other people have been focused on building improved infrastructure. Other people have been focused on building backend IT services and this Live Mesh announcement is an interesting one. All those things together, each company doing its part, add up to a vastly changed environment for devices and that's where we see things going.

George: And that in turn represents a huge opportunity for Allegro Software?

Robert: We're pretty happy these days about the way the world is moving.
We started the company 12 years ago and the premise was that the Internet was going to be everywhere - I'm not sure that we necessarily believed the trillions of device forecast but we thought it might be nice - I think it will be trillions of devices, just not by 2010.

George: I think that it doesn't matter that it's not trillions of devices by 2010. Even numbers in the billions is still an interesting number.

Robert: Absolutely!
We're having fun. It's really starting to move out of the realm of science fiction into the realm of real useful applications for people and when people find things useful they pay for them and we are a supplier to that eco-system.

George: Thank you.

More information about Allegro Software Development Corporation here...

This interview ran in our October 24, 2008 newsletter issue.