KDE and Business: AEI Interview
Continuing in a series of interviews with businesses that benefit KDE and benefit from KDE, we investigate AEI (Analytical Engineering, Inc), a Midwestern engineering firm founded in 1994. In an interview originally conducted by Aaron Seigo, AEI's design engineer and author Caleb Tennis discusses AEI's IT needs and KDE's involvement.
What sort of products or services does AEI offer, and what is your role at AEI?
CT: We're a research and development firm. Right now we are heavily focused on diesel engine technology, particularly emissions. We offer in house durability testing and performance development. We also make a number of products which measure various engine characteristics, such as lubricant oil soot particles. My role is facilities design engineer, but I'm also head of the IT department.
What are AEI's IT requirements?
CT: We have a number of rooms (called test cells) which house the engine that is being tested. The engine is coupled to a large amount of instrumentation that monitors various parameters, such as temperatures and pressures. All of this instrumentation is connected back to a computer, which maintains the engine running at various conditions. Most importantly, it has to be certain that everything is running correctly. If any problems happen, it has to get the engine shut off and notify someone to fix the problem.
We also need a computer to watch over the facility globally. To support each of the cells, we have fans that must provide air for combustion. There are pumps to bring fuel into the building. There is a process water that must be cycled to provide cooling capacity. All of these things must be monitored and maintained.
Furthermore, it's very important from a cost standpoint to minimize how hard the facility is running - we only need to provide air, fuel, and cooling for what the current needs are at any snapshot in time. Providing extra is wasted electricity.
How is KDE helping AEI meet its IT needs, and how long has AEI been using KDE?
CT: Having a very easy to use GUI for the test cells is very important to us. Our test cell computers operate in what I call "pseudo-kiosk" mode. That is, most of the desktop features of KDE aren't used much, but they are available. Instead, all of the operation is done via a few custom written applications. The widgets that are available, and the ease of customizing new widgets, is a huge plus.
A polished look helps a lot too. Not only does it make life very simple for the operator, but eye candy is actually quite important from a business perspective. Potential clients on plant tours tend to remember catchy things they've seen, and almost every computer in our test facility that controls something uses KDE.
What were the key factors that led you to choose to deploy KDE at AEI?
CT: We chose Linux because we felt the need to have something that would be rock solid, customizeable, and affordable. We chose KDE because the API and documentation for it and Qt are unbeatable. In the end it really did come down to a Gnome/KDE decision. Some people still seem shocked that we didn't choose Gnome, because it's "free". That may be so, but from my perspective after having tried both, Qt's design fits our needs a lot better. There will always be an argument for which side is better; after having evaluated both, Qt/KDE worked best for us.
When it comes down to it, we cannot afford to have a mistake happen because the program crashed. Many desktop users are now completely accustomed to having to reboot every few days. They're also used to programs crashing haphazardly. That may be liveable on your girlfriend's computer, but for one that's monitoring a $1,000,000 development engine, it becomes pretty important to not have problems because "Windows crashed".
From our experience, Linux/KDE uptime and reliability are second to none.
How smoothly did the initial deployment of KDE go, and does KDE integrate with any other systems at AEI?
CT: Not that bad. We initially started by using Redhat. I fell in love with KDevelop for my development and started getting disgrunted when trying to compile snapshots of it. This was also around the time that the KDE 3.0 series was being released and I wanted to try it out. We fell into the "rpm dependency hell" that gets talked about very frequently. Later we switched to a Linux From Scratch setup, and now we're using Gentoo. While our production systems all run stable versions, we are constantly trying to keep up with the development versions so that we can take advantage of what's new. Some people believe you should upgrade every few years. I think it's much easier if you're constantly upgrading, every few months. You don't get behind the game this way.
We started using KDE in one of the units we sell to customers. This is as a replacement to software that was running on Windows 95. So far the feedback has been good. Once we're all comfortable with it, we plan to offer more embedded KDE type products in the future.
How many and what sort of machines are you using KDE on?
CT: We have 12 test cells, 2 database servers, development servers, production machines, and various emissions racks. We probably have 15 computers running KDE and another 15 running just Linux.
Has AEI realized any special business or technology advantages from using KDE? (Alternatively: How has KDE been useful/instrumental in solving problems or addressing special needs that AEI has?)
CT: The cost savings has been phenomenal. Our test cell code is used in house, which means that we didn't need to buy a Qt license. We did anyway - because we believe that the value was there. Our code probably doesn't have much value to the greater open source community, anyway.
Linux and KDE are allowing us to make use of older hardware. When someone needs a new desktop computer for their office, we can take their old desktop, put KDE on it, and give it another 5 years of useful life. This means that instead of having to upgrade every computer in the building every 3 years, we can "hand down" computers and get more life out of them. Obviously, this makes management very happy.
Most obviously, since it's all open source, we can customize anything we need. We can also get in and see how things work. We can get ideas from existing code.
And it's not all about us. I've attempted to give back by contributing to projects that I've used. I've worked on KDE, both in bug fixes and documentation. I helped as the maintainer of KDevelop through a large portion of the 3.0 release. Currently I'm helping maintain KDE within Gentoo.
I think if more companies realized that this method works, and that it's substantially less expensive than what they're used to, they would be begging for more information. We've been working more and more with outside companies who are experienced in Qt and KDE development for building up some of our products.
What are your favourite and least favourite aspects of KDE?
CT: My favorite part of KDE is the strength of the community. There are people who thoroughly enjoy their work and their contribution. They want to make the product better. If you have a problem, there's someone who wants to help you.
My least favorite is the strength of binary compatibility. This is also one place where the library shines brightest. I would personally like to see binary compatibility broken more often, particularly when it adds tons of functionality. But I move a lot faster than the rest of the world - and I'm sure 95% would disagree with me. I want the latest and greatest. I don't mind recompiling.
Where would you like to see the future of KDE go, and what would you like to see in future releases?
CT: It's in the right direction now. Continuous improvement - listen for feedback and make things better. The people who care about KDE are doing this anyway.
Better interoperability with Gnome is important, and I think that's heading the right direction. freedesktop.org is a wonderful idea.
Currently I fail to understand why so many people are scared of Qt's licensing scheme. My prediction is that over the next few years, a lot of companies getting interested in Linux will jump to Gnome to avoid paying for Qt. Right now the Gnome folks are dangling this carrot by saying "We're LGPL. You can use us for free!". This will translate to higher costs in the end as the key Gnome players charge more and more for their services. It's like buying razors. The razor costs almost next to nothing; it's the refills that are really expensive.