Denodo relies on data virtualisation, but isn’t defined by it

Denodo is a long-established software company that is probably best known for its data virtualisation techniques. Data virtualisation – or data federation – or federated querying – or whatever else you want to call it, is of course the ability to query data that is in multiple physical locations at the same time, with the same query. This is certainly a useful capability, and there’s no doubt that Denodo’s long-standing data virtualisation technology has won it some significant acclaim over the years. So it’s interesting that the most prominent takeaway I had the last time I talked with the company was a firm-but-polite insistence that Denodo is not a data virtualisation platform.  

At first this might seem counter-intuitive, but when you dig deeper into the company’s logical data platform and what it offers, it’s easy to see why the perception of the company as “just” a data virtualisation platform rankles so much – although it may have been true at one point, it is an impression that is now long past its sell-by date. As of now, the company is more focused on providing a robust data fabric to its customers, one that delivers data governance, metadata management, data integration, data access – and yes, data virtualisation – among other things.   

Although it’s not hard to see that data virtualisation is a core pillar of the platform, it is by no means the whole of it. Nor do Denodo take a hard-line approach when it comes to data virtualisation and its opposite, data materialisation: if on occasion you need to physically move your data in order to query it, go right ahead, says Denodo. Moreover, the other features on offer aren’t merely stapled on, but are genuinely effective at what they do. For example, the platform’s governance layer uses semantic models to classify and tag your data assets, then limits access to them according to both those tags and any policies you’ve laid out, more or less exactly as you would expect from a high-end governance solution, with the models themselves being built via a web-based, self-service design studio. As a matter of fact, the platform is deeply invested in self-service in general (a fact additionally demonstrated by its data catalogue). 

This isn’t to say the company has abandoned data virtualisation. It categorically has not, and continues to develop techniques to both enhance it (such as via ongoing development of smart query optimisation and accelerated caching) and take advantage of it (such as delivering data extremely reliably – even, say, during a data migration). In addition, the company has ambitions of usings its data fabric to support data mesh architectures via the creation of domain-specific virtual models (“logical data fabrics”) that would presumably be impossible (or at least much harder) to develop without data virtualisation. 

The end result of all of this is that in some ways Denodo is a victim of its own success: it has become so thoroughly recognised as a data virtualisation solution that the market continues to think of it as such, even though it has moved far beyond that at this stage. At the same time, data virtualisation is still the crux of what it does, and I would wager it is its most significant differentiator from other data management platforms. The company must walk a fine line, making it clear that data virtualisation is not the whole of what it does, but merely one technology among many, while also acknowledging the fact that it may well be the leading reason for adoption of its product and is clearly still an area of interest (and research) for it. I, for one, do not envy its marketing department.