I see in today’s Irish Times that Steeltrace was sold to Compuware for $20million. Steeltrace was a company that was salvaged from the wreckage of Ebeon. They have a requirements management tool (a pretty good one I might add) that gives you a soup to nuts requirements process based on Use Cases. I’ve known Fergal (the CTO) since his Ebeon days and it couldn’t have happened to a nicer guy.
Another nice exit for TVC on top of Similarity Systems (that Caulfield fella seems to be everywhere these days:-)) so good times all round for the Irish Software Industry…
7 thoughts on “Steeltrace exits for $20million”
Any clue on how well Steeltrace works for worldwide distributed development? Could I collaborate with you in Ireland from here in Portland, OR? My concern from my very brief look at it is it might require developers to be on the same LAN to perform (i.e., like Rational).
Tell me it ain’t so.
My guess is the application to database protocol is JDBC (i.e. not great for WANS). I’ll ask Fergal to respond directly.
The architecture of SteelTrace is client – server (n-tiered & MVC under the covers but lets not go there). Anyways, the server sits on top of an SQL database and handles concurrency, persistence, connections etc.
Now, many of our customers operate in geographicaly distributed locations. (LogicaCMG, GE Assett Management etc.). There a number of options for this. If you can connect to a network over VPN or in a LAN/WAN environment you will be fine as we allow concurrent access by applying automatic locks (controlled by the server) to any elements you touch.
On the other hand, ST also allows full projects to be taken offline and worked on completely disconnected from the network. Projects can subsequenly be merged back online. This is excellent for consultants or SIs in general as frequently they are on the road. Offline projects can be merged then back in once you’re back on the network without locking the central project.
Therefore, if you’re in Portland for example and you send me your offline project as an attachment in a mail (footprint is v. small as a zip cause it’s a text file) I can load it here in Dublin, make changes and you can accept that back, merge it and review those changes. This is completely unlike CaliberRM from Borland or ReqPro from our Rational friends who have an ‘always connected’ model. They’ll possibly say otherwise but thats more less the reality.
Also, as our latest release version 4.1 (due early July) we have introduced better client server configurability so that you could fo instance have a real-time cnnection from a hotel room if running a VPN for instance. You can also configure ports for firewalls etc.
hope this helps, rgds, Fergal.
Sounds pretty impressive.
What about if I use Linux as my OS. Am I canned?
You mentioned Rational and the “always connected” model. They also have the “must be on the same LAN” model because it is so RPC intensive.
Please tell me that you scale significantly better. Will I gauge my eyes out due to poor performance if the server is in Ireland and I am here in Portland?
Thanks – happy Friday.
On the Linux question, we don’t support Linux so you’re kinda out of luck. The client and server are win32 however you *could* possibly run the SteelTrace server on a win32 box and have the DB MySQL for instance run on a dedicated Linux box. I can’t standover that but we’ve heard of folks doing it. (BTW, while we’re on it, we also support MS SQLServer and Oracle 9i as the DB if thats’ your poison.)
Our model for 4.1 is client to periodically poll the server. We used to have ‘callbacks’ from server to client in 4.0 but we’ve simplified that whole area for 4.1
On the scaling side, under reasonable project partitions we have never had scaling problems reported. Scalability is pretty bad on Caliber for instance cause they use Versant OODB which for so many reasons is a very bad idea but most noticably on the scalability side. I won’t mention Req Pro as it’s a whole different ball of wax due to it’s clumsy word interface where Word tries to pretend it’s a front end. As we in SteelTrace know, word is pretty tricky to code in that manner and not for the faint hearted. In fairness to reqPro, it’s still chugging along after many years of neglect and nice to see Smalltalk lives (just about). But I digress….
If the server is in Ireland and you are in Portland you won’t die a slow death cause we only shunt deltas over the wire on saves and the polling interval can be configured.
Form a latency perspective, the proof is in the pudding, so for instance, I know for sure we have at the moment, the guys in Amsterdam in LogicaCMG shunting back and forth to their server in Bangalore and the GEAM guys doing the same from Connectuicuit to their offshore arm, also in India. In both instances the server lives offshore and both companies have pretty heavy offshore fulfillment models. ADP Canada work between Toronto, Montreal and a number of folks in the west of Canada. I’ve not heard of any issues. I work closely with these customers but can’t speak for all the sites we’re deployed on as they number close to 500 but I’ve not seen scaling noises/chatter to date.
So, I’d say you should be fine hooking into Dublin if that’s the scenario.
enjoy the Friday too, the sun shines here in Dublin and thats always the biz.
PS: if you want to see more on SteelTrace check out our forum at: http://forum.steeltrace.com/index.php
Cool thanks Fergal. I’ll use the forum for any further questions.