Thomas Erickson recently took over as president and CEO of Systinet, one of a growing number of companies focused on the web services and service-oriented architecture (SOA) management spaces.
Prior to joining Systinet, Erickson served as an executive vice-president at webMethods and also held executive positions at Baan, Filenet/Watermark Software, and MRO Software. Among Systinet’s products is Systinet Registry, a Universal Description, Discovery and Integration (UDDI)-compliant registry for SOAs.
Earlier this month, Erickson met with InfoWorld's staff to discuss his views on the phases of web services, web services standardisation, and alliances with other suppliers.
What is your view on web services?
We are looking at the web services world or SOA world as a three-phase evolution. The first phase involved people who, rather than building this application programming interface and remote procedure calls with some proprietary protocol, build it using web services.
The second phase is what we call the business services phase - moving from a web service to a business service. The difference is that people are now starting to say, rather than just build these interfaces onto existing applications, we are going to design something from the ground up with a notion of reuse, with a notion of providing real value.
The best example is one of our customers which unfortunately will not allow us to share their name. It is a financial services organisation, one of the big five globally. It has a compliance requirement where to prevent money laundering it wants to do compliance management of wire transfers, and it operates in 120 countries around the world.
However, it wants everybody to use the same service. Compliance is a big issue across all kinds of corporations. The enterprise architects in these companies are starting to realise that creating a business service where compliance might be a compelling need seems to make sense.
Web services management is what we call the third phase, and that still needs some more standards to get into play. That needs things like WS-Addressing. So the first phase has just really done web services and then created some business services.
But what they have not done is helped us with what an SOA’s supposed to be about. This is interoperability between these endpoints or applications, if you want to call them that, to the extent that not everything has to be known by the person building the endpoint at the time they build it.
Do you think there are too many standards efforts going on in web services?
It is confusing, and it is confusing for us as a company trying to keep up with all of them. Are there too many? I am a perpetual believer that dialogue is good, but I believe some standards will survive and others will not. You need to have dialogues between parties to create some kind of interoperability capability.
These draft standards are evolving. You quickly realise that to create a true web services, there is a lot of information you have to have about it so that it can be considered for reuse and used in a broader context. There is a whole variety of, if systems is not the right word, interested parties.
One customer called it the SOA cloud. Interested parties in the SOA cloud, the management vendors, the security vendor, the identity vendor, all want to know information about that service so that they can interoperate.
Very quickly you start to ask, where is that information going to be stored? That is one of our main value propositions, and in discussions with the analysts and our customers, they are starting to come to the realisation that there is a notion here of a registry repository where this information needs to be stored.
Would that be UDDI?
UDDI has its issues, but perhaps UDDI is the alternative that exists to start us along the path. UDDI is only a schema you know. And it is a way to look into the database.
So when you say this registry repository is standards-based, you mean you can query it using open standards like XQuery?
Like XQuery, like UDDI. You have to have some kind of facility to put a taxonomy in there that other applications throughout the enterprise can recognise.
How much can a business service really describe itself, and what it can do via WSDL? How far can you push WSDL, because web services is just an interface? And if you are talking about data types and all this other kind of stuff, then you are talking about getting into XML schema and all these other things that do not really directly relate to web services.
The context around what you are talking about is in fact our whole supposition for a different need for tools to enable this business services evolution. You have just come up with a lot of our value proposition around our go-to-market strategy, which is, how much can you do in a WSDL and where else, what else do you need to get the job done?
Because you do need other things. You need to be able to provide policy information with that, you need to be able to provide different types of management about how it s operating.
You need to be able to provide information about security and identity. Web services management from our perspective will include things like: if we are going to revise it, what types of things we need to know to revise it. If you are on the runtime side, you need to know who is using it, what is the state of it, all these other factors.
And our point, and the reason for cooperating with AmberPoint - and we have got some dialogues with the other players in this space too - is that we want to ensure that we collaborate with these companies to do the right thing.
One of the first things we are doing is creating a technical brief on web services distributed management (WSDM) to UDDI mapping. So in other words, how do you take the WSDM standards and put them into a repository? How does that look?
Haven’t you also had an alliance with Actional?
We have had an alliance with Actional in the past.
Do you see any other alliances happening?
We do see some more, we are in discussions with some of what we call the “classic management vendors.”
Are you watching for any potential acquisitions [of Systinet] by a large company that might want to buy a technology like yours instead of developing it in-house? Do you see the company as a whole as maybe an acquisition target?
I don’t know how much you know about Warburg Pincus, who are our investors. They are a rather large, even perhaps you could say old-fashioned type of private equity firm rather than a true venture capital, in that we’re part of a large $5bn (£2.73bn) fund that they have.
I asked them that question because in some of our discussions with our partnerships, which we can’t announce today but we are starting to accelerate on. Invariably when you get into this kind of discussion with other vendors, they are going to say, "Well, rather than partner with you, what if we buy you?" What are your motivations?
I joined the company five months ago and I didn’t come on board to flip the thing. I moved my family from Australia. But I needed to check with them as well. They said, "Not only did we just hire you, but we just put $10m more in Systinet back in May.
Our motivation is not to sell Systinet. The motivation is to try to create some real value in the SOA space. We believe there is going to be a play, and we are going to use Systinet as the vehicle to achieve that. We believe there is an opportunity that when a paradigm shift happens there is an opportunity for a new company to emerge.
If you go through time you can see this happening repeatedly. There has been a new major player in the software industry appearing when there are paradigm shifts because some of the older players fail to make the transition. Whether it is in the hardware transition from mainframes to minicomputers to micros, whether it is been software transitions from mainframe to client/server to all along the way. Because Pincus was involved in BEA Systems, they are the company that funded BEA, they have a similar vision - that we can do this again.
So the quick answer to your question is, no, we are not trying to find a company to buy us.
Paul Krill writes for InfoWorld