The third quarter was a good one for Iona Technologies, a player in the rapidly growing SOA [service-oriented architecture] field. The company saw revenue jump 19 percent over the same period last year, achieved Microsoft Gold Certified Partner Status and rolled out its Fuse open-source offerings. Eric Newcomer, Ionas chief technology officer, sat down with eWEEK Senior Editor Darryl K. Taft to talk about the industry, the company and the future.
Whats with the SOAP [Simple Object Access Protocol]-versus-REST [Representational State Transfer] debate? Have we put all that behind us?
I think so. It looks pretty clear that theyre going to coexist, and probably for a long time. I think all but the most fanatical REST-afarians can acknowledge that SOAP and WSDL [Web Services Definition Language] provide a lot of value to a lot of people. Web services gained adoption first, and then we have had a big push for REST, which unfortunately gained the most attention when it was characterized as anti-Web services.
Im convinced the seemingly endless debate derives from the fact that the SOAP and REST folks make fundamentally different assumptions about the world. The REST folks assume that all enterprise computing requirements can be met using HTTP, and they kind of skip over the fact that you have to do a complete redesign and reimplementation to accomplish that. Sure, the big Web sites have done it, but they started with a clean slate.
On the other hand, the Web services people assume that companies want to preserve and extend the designs and implementations they already have. Both are correct and there is no ultimate answer; it always depends on the situation at hand.
Fortunately were now seeing convergence, with vendors starting to implement both. We support REST in the latest version of our products, and I know that IBM and Microsoft are doing the same, probably among a lot of others.
Whyd you guys do a [registry/repository] offering?
It started with a something we were hearing from our customers. They like the distributed approach to SOA infrastructure that we provide, and the lightweight microkernel-based containers, but they kept asking us for better support for configuring, deploying, and managing it. So we built the reg/rep to manage information about the services and their configurations and deployments.
The best thing about it, which gets all the attention, is that you can discover and assemble services, attach policies to them, choose a deployment container, and literally push a button to generate deployable code.
None of the existing reg/rep products on the market did this, so we needed to develop our own.
Whats your view of the state of Web services standards? Are we there yet?
Im not sure were there, but I think were basically done. I think its time to work on other stuff. Last year at this time I wouldnt have said that, and when the Microsoft folks started saying it I admit I wasnt convinced.
But SOAP, WSDL, WS-BPEL, and WS-Security are widely adopted, and I think thats realistically about all were going to get. Some of the other specs have achieved a level of adoption, but most surveys show that the core specs are still the ones most widely adopted and available. We do support several of the others, as do many other vendors, but its not clear at all that all the vendors are going to deliver the same set-except for the core ones.
So I guess Im saying lets move on, a lot has been accomplished with Web services, but I think the industry now needs to tackle the question of how to standardize what the large Web sites are doing around their infrastructures.
Page 2: Iona and SOA
Iona and SOA
What do you mean?
At the HPTS [High Performance Transaction Systems] conference this year, it became pretty clear that the current designs for enterprise software are not meeting the requirements of the large or even some of the small Internet businesses. There were presentations from Google, Yahoo, eBay, PayPal, Salesforce, KnowNow, Second Life, and others, and folks from Amazon.com were there as well.
The way I characterize it, and this is probably an oversimplification, is that the traditional designs for middleware and database management systems were based on a kind of mainframe-centric set of requirements. People were thinking about centralized systems with centralized control structures and centralized guarantees of consistency. But if you think about how these things have evolved, the need for different solutions is obvious. You can only scale up to the biggest computer system, but with the new scale-up architectures, you can scale without limit. And you can use the cheapest hardware and disk systems. Once you get past the idea that failure is inevitable, you can accept the idea that failure is common and potentially frequent. You have to deal with it anyway, why not save yourself a lot of money and plan to deal with failure as often as possible?
How do you see the open-source ESB [enterprise service bus] world shaking out? What about open-source SOA in general?
Open source is a growing trend in general, and the ESB category is no exception. We are seeing a growing interest in ServiceMix, and although Mule has achieved some initial success, we expect significant competition to emerge from Apache Camel.
I have no doubt open source is the wave of the future. Its hard for me to explain exactly why I believe it, but it has to do with the significant change in the software industry. The software industry has reached a turning point. Twenty or 30 years ago, all we thought about was the next process to automate. Should it be inventory management, order entry, office automation, packaging and shipping?
Today we are really reaching the edge cases of automation, most everything has been done.
The industry has to shift between developing new features and functions to enable more processes to be automated, to improving that automation. One of the ways to improve it is to lower the cost, to commoditize the features and functions, so that you pay for the value of the work-how well the feature has been implemented-rather than for the value of the idea, whether you are the only one using the feature or function. Thats kind of a long way of saying that I think open source is here to stay, including for SOA.
Regarding the open-source ESB question in particular, I think the jury is still out. But I also think its getting to be a kind of two-way race between ServiceMix and Camel and Mule. I think Mule definitely started out with a great idea, that took a while to recognize-and I can definitely speak for myself here, it took me a while to catch on, longer than it should have, to be honest-but now that its out there, I think things like Camel will catch up and do well, especially because its Apache-licensed and Mule isnt.
Its been some time since Iona moved to spin out Celtix as an open-source effort. What has that experience shown you?
Well, its definitely taught us a lot, including the fact that we didnt know as much about open source as we thought we did. Thats why we acquired LogicBlaze, and that has worked out very well, so much so that we have retired the Celtix brand and are going forward with the Fuse brand that LogicBlaze was using.
The history is that we started Celtix at ObjectWeb and after V1 moved it to Apache, where we worked with the XFire guys to create Apache CXF. Initially we released this under the Celtix brand.
With ServiceMix, we brought into the company some of the leading committers on Apache ActiveMQ-the leading open source JMS [Java Message Service]-Apache ServiceMix-the leading open-source ESB-and Apache Camel, a fast rising star in the field of integration patterns.
Page 3: Iona and SOA
Iona and SOA
We have that, and rebranded it to become Fuse Service Enablement. The other parts of Fuse are Fuse ESB, which is based on Apache ServiceMix, Fuse Messaging, which is based on Apache ActiveMQ, and Fuse Routing, which is based on Apache Camel. We have created a separate business unit to ensure that the open source strategy can be pursued independently of the commercial software strategy, but I think its safe to say the whole thing has been more difficult than we expected initially.
We are very committed to open source because we strongly believe it represents a significant future line of business, but getting the business model right has been trickier than we expected. We have incorporated CXF into Artix/Java and built Artix Java on top of
Whats your bet on OSGi and where it will take the industry?
A couple of important things here-modularity and dynamism. OSGi helps break up applications into compatible modules that can be loaded and unloaded dynamically. I think customers are going to look at Java applications in terms of what they can deploy onto OSGi. I mean, theyre going to start with an OSGi platform and then buy things that run on it, from different vendors, and develop some modules themselves. The magic of OSGi ensures it will all work together.
Why did you guys buy C24? Whats it bring to Iona?
We had been working with the C24 folks for several months ahead of the acquisition, and it just felt right, culturally and businesswise. C24 brings us an industry-leading SWIFT [Society for Worldwide Interbank Financial Telecommunications] network integration, which is one of the most important interchange standards in the financial services industry, which is our second biggest vertical after telecommunications. The C24 approach is very compatible with Artix and CXF, and weve already delivered an integrated solution to some customers. In an acquisition, you always hope for synergy, to take a smaller organization and increase the volume of its sales by plugging their product into a larger organizations sales channel, and thats exactly whats happening. We also acquired some world-class individuals who have fit in well.
How is the Eclipse SOA Tools Project going?
It could be going better. We had hoped for a V1 delivery in Europa, but we missed that. We did ship a beta, but we would obviously have liked to have been a bit farther along. Theres been some change in membership in the project, and a change in focus. We had to do a bit of a replan following Europa, and theres been some new additions to the project, so we are optimistic now for completing the project, getting something out by the end of the year, and hitting the next major Eclipse release.
Eclipse is also home to the industry leading OSGi platform, Equinox. We are looking to also align the Eclipse work with the OSGi enterprise expert group work.
Anything happening with AMQP [Advanced Message Queuing Protocol]?
Version 10 is due out in a few days and there will be an announcement about that. Several new members have joined the working group, mostly from the customer side, so thats pretty good. Unfortunately, the spec keeps changing and evolving. Thats good, since its improving, but it seems like its potential impact on the industry continues to sort of hang in the balance. The idea is great-no one can argue with the need for a standard message format-but the extent of its adoption is still somewhat in question, and with the specification continuing to change, it may take longer to sort that picture out. We are still participating, still supporting the effort, but I guess whats unfortunate is that a lot of the question marks about it remain unanswered.
In the meantime we are supporting the open-source FUSE Message Broker, based on Apache ActiveMQ, and doing well with that. At some point we expect to include AMQP as a transport for FUSE, but given the uncertainty around AMQP, its hard to say exactly when.
Check out eWEEK.coms for the latest news, reviews and analysis in programming environments and developer tools.
The third quarter was a good one for Iona Technologies, a player in the rapidly growing SOA [service-oriented architecture] field. The company saw revenue jump 19 percent over the same period last year, achieved Microsoft Gold Certified Partner Status and rolled out its Fuse open-source offerings. Eric Newcomer, Iona’s chief technology officer, sat down with eWEEK Senior Editor Darryl K. Taft to talk about the industry, the company and the future.
What’s with the SOAP [Simple Object Access Protocol]-versus-REST [Representational State Transfer] debate? Have we put all that behind us?
I think so. It looks pretty clear that they’re going to coexist, and probably for a long time. I think all but the most fanatical REST-afarians can acknowledge that SOAP and WSDL [Web Services Definition Language] provide a lot of value to a lot of people. Web services gained adoption first, and then we have had a big push for REST, which unfortunately gained the most attention when it was characterized as anti-Web services.
I’m convinced the seemingly endless debate derives from the fact that the SOAP and REST folks make fundamentally different assumptions about the world. The REST folks assume that all enterprise computing requirements can be met using HTTP, and they kind of skip over the fact that you have to do a complete redesign and reimplementation to accomplish that. Sure, the big Web sites have done it, but they started with a clean slate.
On the other hand, the Web services people assume that companies want to preserve and extend the designs and implementations they already have. Both are correct and there is no ultimate answer; it always depends on the situation at hand.
Fortunately we’re now seeing convergence, with vendors starting to implement both. We support REST in the latest version of our products, and I know that IBM and Microsoft are doing the same, probably among a lot of others.
Why’d you guys do a [registry/repository] offering?
It started with a something we were hearing from our customers. They like the distributed approach to SOA infrastructure that we provide, and the lightweight microkernel-based containers, but they kept asking us for better support for configuring, deploying, and managing it. So we built the reg/rep to manage information about the services and their configurations and deployments.
The best thing about it, which gets all the attention, is that you can discover and assemble services, attach policies to them, choose a deployment container, and literally push a button to generate deployable code.
None of the existing reg/rep products on the market did this, so we needed to develop our own.
What’s your view of the state of Web services standards? Are we there yet?
I’m not sure we’re there, but I think we’re basically done. I think it’s time to work on other stuff. Last year at this time I wouldn’t have said that, and when the Microsoft folks started saying it I admit I wasn’t convinced.
But SOAP, WSDL, WS-BPEL, and WS-Security are widely adopted, and I think that’s realistically about all we’re going to get. Some of the other specs have achieved a level of adoption, but most surveys show that the core specs are still the ones most widely adopted and available. We do support several of the others, as do many other vendors, but it’s not clear at all that all the vendors are going to deliver the same set—except for the core ones.
So I guess I’m saying let’s move on, a lot has been accomplished with Web services, but I think the industry now needs to tackle the question of how to standardize what the large Web sites are doing around their infrastructures.
Page 2: Iona and SOA
What do you mean?
At the HPTS [High Performance Transaction Systems] conference this year, it became pretty clear that the current designs for enterprise software are not meeting the requirements of the large or even some of the small Internet businesses. There were presentations from Google, Yahoo, eBay, PayPal, Salesforce, KnowNow, Second Life, and others, and folks from Amazon.com were there as well.
The way I characterize it, and this is probably an oversimplification, is that the traditional designs for middleware and database management systems were based on a kind of mainframe-centric set of requirements. People were thinking about centralized systems with centralized control structures and centralized guarantees of consistency. But if you think about how these things have evolved, the need for different solutions is obvious. You can only scale up to the biggest computer system, but with the new scale-up architectures, you can scale without limit. And you can use the cheapest hardware and disk systems. Once you get past the idea that failure is inevitable, you can accept the idea that failure is common and potentially frequent. You have to deal with it anyway, why not save yourself a lot of money and plan to deal with failure as often as possible?
How do you see the open-source ESB [enterprise service bus] world shaking out? What about open-source SOA in general?
Open source is a growing trend in general, and the ESB category is no exception. We are seeing a growing interest in ServiceMix, and although Mule has achieved some initial success, we expect significant competition to emerge from Apache Camel.
I have no doubt open source is the wave of the future. It’s hard for me to explain exactly why I believe it, but it has to do with the significant change in the software industry. The software industry has reached a turning point. Twenty or 30 years ago, all we thought about was the next process to automate. Should it be inventory management, order entry, office automation, packaging and shipping?
Today we are really reaching the edge cases of automation, most everything has been done.
The industry has to shift between developing new features and functions to enable more processes to be automated, to improving that automation. One of the ways to improve it is to lower the cost, to commoditize the features and functions, so that you pay for the value of the work—how well the feature has been implemented—rather than for the value of the idea, whether you are the only one using the feature or function. That’s kind of a long way of saying that I think open source is here to stay, including for SOA.
Regarding the open-source ESB question in particular, I think the jury is still out. But I also think it’s getting to be a kind of two-way race between ServiceMix and Camel and Mule. I think Mule definitely started out with a great idea, that took a while to recognize—and I can definitely speak for myself here, it took me a while to catch on, longer than it should have, to be honest—but now that it’s out there, I think things like Camel will catch up and do well, especially because it’s Apache-licensed and Mule isn’t.
It’s been some time since Iona moved to spin out Celtix as an open-source effort. What has that experience shown you?
Well, it’s definitely taught us a lot, including the fact that we didn’t know as much about open source as we thought we did. That’s why we acquired LogicBlaze, and that has worked out very well, so much so that we have retired the Celtix brand and are going forward with the Fuse brand that LogicBlaze was using.
The history is that we started Celtix at ObjectWeb and after V1 moved it to Apache, where we worked with the XFire guys to create Apache CXF. Initially we released this under the Celtix brand.
With ServiceMix, we brought into the company some of the leading committers on Apache ActiveMQ—the leading open source JMS [Java Message Service]—Apache ServiceMix—the leading open-source ESB—and Apache Camel, a fast rising star in the field of integration patterns.
Page 3: Iona and SOA
We have that, and rebranded it to become Fuse Service Enablement. The other parts of Fuse are Fuse ESB, which is based on Apache ServiceMix, Fuse Messaging, which is based on Apache ActiveMQ, and Fuse Routing, which is based on Apache Camel. We have created a separate business unit to ensure that the open source strategy can be pursued independently of the commercial software strategy, but I think it’s safe to say the whole thing has been more difficult than we expected initially.
We are very committed to open source because we strongly believe it represents a significant future line of business, but getting the business model right has been trickier than we expected. We have incorporated CXF into Artix/Java and built Artix Java on top of
What’s your bet on OSGi and where it will take the industry?
A couple of important things here—modularity and dynamism. OSGi helps break up applications into compatible modules that can be loaded and unloaded dynamically. I think customers are going to look at Java applications in terms of what they can deploy onto OSGi. I mean, they’re going to start with an OSGi platform and then buy things that run on it, from different vendors, and develop some modules themselves. The magic of OSGi ensures it will all work together.
Why did you guys buy C24? What’s it bring to Iona?
We had been working with the C24 folks for several months ahead of the acquisition, and it just felt right, culturally and businesswise. C24 brings us an industry-leading SWIFT [Society for Worldwide Interbank Financial Telecommunications] network integration, which is one of the most important interchange standards in the financial services industry, which is our second biggest vertical after telecommunications. The C24 approach is very compatible with Artix and CXF, and we’ve already delivered an integrated solution to some customers. In an acquisition, you always hope for synergy, to take a smaller organization and increase the volume of its sales by plugging their product into a larger organization’s sales channel, and that’s exactly what’s happening. We also acquired some world-class individuals who have fit in well.
How is the Eclipse SOA Tools Project going?
It could be going better. We had hoped for a V1 delivery in Europa, but we missed that. We did ship a beta, but we would obviously have liked to have been a bit farther along. There’s been some change in membership in the project, and a change in focus. We had to do a bit of a replan following Europa, and there’s been some new additions to the project, so we are optimistic now for completing the project, getting something out by the end of the year, and hitting the next major Eclipse release.
Eclipse is also home to the industry leading OSGi platform, Equinox. We are looking to also align the Eclipse work with the OSGi enterprise expert group work.
Anything happening with AMQP [Advanced Message Queuing Protocol]?
Version 10 is due out in a few days and there will be an announcement about that. Several new members have joined the working group, mostly from the customer side, so that’s pretty good. Unfortunately, the spec keeps changing and evolving. That’s good, since it’s improving, but it seems like its potential impact on the industry continues to sort of hang in the balance. The idea is great—no one can argue with the need for a standard message format—but the extent of its adoption is still somewhat in question, and with the specification continuing to change, it may take longer to sort that picture out. We are still participating, still supporting the effort, but I guess what’s unfortunate is that a lot of the question marks about it remain unanswered.
In the meantime we are supporting the open-source FUSE Message Broker, based on Apache ActiveMQ, and doing well with that. At some point we expect to include AMQP as a transport for FUSE, but given the uncertainty around AMQP, it’s hard to say exactly when.
Check out eWEEK.com’s Application Development Center for the latest news, reviews and analysis in programming environments and developer tools.