Semantic Web and Big Data: Moving past the Relational Data Model
John Randles CEO Bloomberg PolarLake
A common complaint from buyers of Data Management Platforms is they have difficulty differentiating between competing commercial Data Management offerings. This is often down to vendors saying the same words and competing along the same lines. For example the “the 25,000 fields in my data model has a home for your entire set of Vendor Data” pitch is often used as is the “my list of adaptors is a superset of all you will need and longer than my competitors’ list”. When everyone says the same thing, offer what seem to be the same solution and have the same number of clients then buyers become frustrated.
But there is one area seldom mentioned which is a key to successful Enterprise Software implementations – the upgrade path. With most successful Enterprise Software systems the initial implementation is a small fraction of the overall lifetime of interacting with the product. Therefore it’s surprising that the issue of product upgrade gets so little airplay when trying to find differentiators between Data Management products. This is a huge issue when you consider a lot of the value proposition of commercial Data Management offerings is to maintain connectors / adaptors / feed handlers to Data Vendors. As Data Vendor feeds evolve more and more rapidly over the lifetime of a Data Management Platform then you can see why the upgrade path becomes the key issue, not an afterthought. This is especially significant with vastly more feeds and more rapid feed changes happening than 10-15 years ago, when the current relational Data Management approaches evolved.
Upgrades are crucial in understanding the real value delivered by annual support and maintenance, particularly in the little understood world of “Data Vendor adaptor support”. The client pays for updates to “Data Vendor Adaptors” which change when the Data Vendor feeds change (new fields, feeds, renames etc.). But what good is your new upgraded adaptor when you don’t take the relational data model upgrade, which is configured to the new adaptor? And when you look at upgrading the “Data Model” you discover that it has been invariably so customized and localized that any “upgrade” becomes a re-implementation of the product. This causes the Data Vendor adaptor support value proposition to fall through the floor as clients end up doing the adaptor maintenance themselves. The key problem then becomes the underlying technical infrastructure isn’t upgradable and therefore can become obsolete over time.
This is why a lot of seemingly straightforward Data Management product upgrades turn into a fully blown RFP processes. It is also why many traditional support and maintenance contracts end up being cancelled as the value proposition becomes more and more tenuous with each passing year from the original implementation. Maintenance ends up being done in house when the upgrade becomes a daunting new project. Some even try to re-implement with the vanilla “out of the box, un-customized newer version” only to re-discover that it is completely and totally unusable in the real world “un-customized”.
This issue is really a continuation of a theme we have been running on why it is important for large organizations to own, control the core intellectual property of their “Data Model” or “Data Semantics” as we prefer to describe it. The difficulties in upgrading and getting true value from the commercially supported Adaptor and Data Model solutions is one of the main reasons why a lot of firms are looking to new approaches to Data Management or end up building their own. The idea that a large firm can migrate its business to a 3rdparty Data Model and wait patiently for upgrades is just plainly unworkable and has resulted in much frustration from the end users. And all the market’s focus is now on scalability in many directions and dimensions – volumes, business requirements, regulations and operations – and all with never ending demand to decreasing costs at the same time. This is why the old 3rdparty Relational Data Model approaches are no longer sustainable.
We believe that firms need to concentrate on the Semantics and Business uses of their Data and not be distracted by technology infrastructure. Technology infrastructure is what software vendors, like PolarLake, look after and allow firms to concentrate on the core intellectual property and what makes their business different to their competition. The next generation of Data Management Platform based on Semantics and Big Data Technologies do not suffer from these legacy limitations and have learnt from the past. In future the client should be able to concentrate on the Data Semantics, which is what their business is really about and leave the plumbing to the software vendors.
So the next time you get a sales pitch on a Data Management Platform ask the vendor to set aside the glossy slides and talk to about upgrades, not just implementations.
To learn more about how Semantic Data Management can address the challenges of the traditional problems of Relational Data Models please contact us.