I'd say that, regrettably, my experience of the root causes of failure has not been altered by recent experience. Seems like the common theme of most failing programmes is their overly technical focus and the creation of a responsibility gap. However we, at iergo, have been doing a lot of work to make the assessment of failing projects so much more objective, and I've shared some of that thinking in this article. Check it out and let me know what you think.
On to Talend. I think it's taken me three weeks to write about these guys. And now I'm not sure the brevity of this posting will do them justice, but check them out at www.talend.com. Talend have gone down the open source route to develop and build their products. They have a drag and drop ETL development environment (Open Studio) that can be downloaded for free from their web site. Plus a landscape analysis tool also free to download. Together, these give you all the standard features - metadata management, data interrogation, ETL flow designer etc., with a rich pallet of components. My examination of the product was too brief to comment on their 400-plus components but I'm guessing that with the flexibility of the open source model they should have a widget for most things.
However, in addition to the free to use Open Studio, they also sell, under licence, an Integration Suite. This includes the (very necessary on all but the smallest projects) shared repository to allow development to be booked in and out, shared, re-used. The Integration suite also has additional functionality - like a metadata importer - plus the necessary scheduling and load balancing that an industrial strength ETL tool needs for deployment.
However, for the purposes of our cloud discussion, what interests me at the moment is their Software as a Service (SaaS) model that offers the integration suite as Talend On Demand. A subscription service located on the Web, it is suited to creating a collaborative environment for Data Migration projects without the hassle of configuration and access issues. It still relies on Open Studio for its design and runtime environment. As envisaged, these would reside locally; however, as both Open Studio and Integration Suite offer web services as a connector mechanism, there is nothing to prevent us mounting the data (source and target) in the cloud as far as I can see.
The Integration Suite comes in five flavours, four of which offer both a monitoring console and dashboard. Returning to an early blog on architecture, it would be interesting to see how far these could be integrated into a workflow engine that would manage the whole migration process. I don't know the answer to this. I will endeavour to find out and let you know, but I think we can see, at least from this vendor, the direction of the first steps of the use of SaaS in data migration will be putting collaborative services on the Web as an easy, common location. One could envisage the use of a common Integration Suite stacked with the metadata for popular COTS packages as a useful service, but the next step? That's where we should be looking.
Finally, just a note to say that we have settled on the date of the next Data Migration Matters event - it will be 1st October. Details of where to book etc will be made available shortly. We will be making a call for papers, with the opportunity of presenting to the conference, probably managed via our relationship with Data Migration Pro. Look out for more news in the next week or so.
Johny Morris
jmorris@iergo.net
