Data Access Governance

In some conversations that Project X has had with vendors as well as customers over the past couple of months, I have found some interesting insights as a result of Service Oriented Architecture and it’s import on good data access governance and some of the opportunities that present themselves.  Here are three I would like to share and I welcome more:

  1. Teradata Views
    As a generally accepted practice in teradata you create views of data when you want to start leveraging the data in the warehouse.  Now these views are virtual so they take up next to no space like a play list on your iPod.  One of the great advantages of these views as that by their very nature access to these views can be set on a permission basis, so by a somewhat default (though Teradata would say by design) you get data access/usage governance and control.
  2. Enterprise Information Integration Platform Views (like ipedo)
    In a similar manner on the EII platform the user gets access to disperate enterprise data through views which can be highly controlled both in access and design.  Once again this gives us an opportunity for data access/usage governance.  The added value in some of the tools is that you can then publish these views as services which can then be managed by the registry.
  3. Data as a Service – in general this is very similar to the tail end of EII story without it necessarily being EII.  This allows the service registry to manage who has access to the data.

This in contemplation created some interesting thoughts so I did a bit more research and came up with SOA is useless without good data by Joe McKendrick and he added the ever elusive issue that still reins supreme "Metadata" and "Data Model".  He is so right on the importance of these, but sadly just like many other IT value creation issues, these are often overlooked.

My last thought was as we build services and then services on those services at what point may we by mistake share information with someone who isn’t supposed to get the data or are we always building subset groups smaller and smaller until only 3 people can look at the results.

Leave a Reply

captcha *