Database popularity

In case you don’t know it, there is a website at www.db-engines.com that purports to monitor database popularity both across different types of database (relational, key-value, graph and so on) and between individual databases within each category. It’s quite interesting and some vendors (those that are doing well) use it in some of their marketing. So, while I say it is interesting, I want to sound a note of caution: you should take the results posted on this site with a pinch of salt.

The most egregious example of this is that Db-engines rates Oracle (ranked the most popular database) as around 8 times more popular than DB2. If you make the (realistic) assumption that this simply isn’t true then it is important to understand how this figure has got so inflated.

Db-engines develops its ratings based on analysis of social media: how many mentions there are of a particular database in, say, LinkedIn or how many job adverts are there for a particular database, or how many comments there are on monitored message boards. That’s fine as far is it goes but there are at least a couple of caveats one should place around that.

The first is, that companies with eponymous database products are likely to fare better than those that don’t have the same names: if your company is Oracle and your database is commonly referred to as Oracle then the two will likely get conflated, at least some of time, whereas this will not happen with IBM and DB2.

Secondly, suppose that you need twice as many database administrators to manage one particular database compared to another specified database. Then wouldn’t you expect as least twice as much (actually more because it won’t be a linear scale – probably logarithmic) message-board traffic and job advertisements?

So, DB-engines is interesting for those us that are in the database space but it is the trends that are perhaps more fascinating than the actual comparisons. For example, graph databases – my next major project – are shown as having the highest growth in interest over the last 12 months, albeit from a low starting point. But, whatever you do, don’t take the findings on this site as gospel.