Increased complexity is generally a good thing for managed service providers because it adds more stress to IT environments, which usually creates demand for additional managed services. Lately, the greatest complexity involves the number of places data gets stored. Once upon a time, all there was were relational databases accessing block storage systems or one of handful of file systems requiring network-attached storage. Since then, so-called NoSQL databases based on platforms such as Hadoop, document databases, and various flavors of key/value stores have proliferated, and there are also object-based storage systems residing in the cloud.
A new survey of 875 IT professionals conducted by O’Malley Media on behalf of Zoomdata, a provider of tools for analyzing data, suggests that more than two-thirds of analytics applications are accessing data sources other than a relational database. That’s a long way from when the only real option for building a data warehouse was a relational database.
Emerging data platforms
The study finds that the most popular mechanism for storing data is still the relational database. But, it’s clear that in terms of data platforms MSPs need to master multiple solutions. According to the study, the top places data is being stored other than relational databases are analytic databases, Hadoop, NoSQL databases, cloud data stores, in-memory databases, and search databases. In fact, the study finds that while data warehouses and data marts are still the most widely used means to blend data from different sources, many organizations are increasingly blending data on the fly in memory within an analytics application.
To make matters even more interesting, there are also more types of data than ever. From video files to graph databases and time-series data generated by Internet of Things (IoT) projects, organizations are trying to mix and match disparate data types in ways that were never imagined. More challenging still is that only some of that data is structured. Most of it is either semi-structured or completely unstructured.
Securing data on different platforms
The study finds that the IT skills most sought after span expertise in Python, SQL, relational databases, Hadoop, and finally Java. But MSPs would should also take note of the fact that all this distributed data needs to be managed, secured, and protected. Each type of data being stored has not only different security requirements, but also different recovery time objectives and recovery point objectives.
Developers and independent software vendors don’t always inform internal IT about what type of databases and data stores they are employing. It often comes as a surprise to internal IT just how many have been installed. MSPs might want to consider providing free data discovery just to make the point. Once that’s completed, the next logical question is how is the data residing in all those data platforms being secured and protected. Chances are high it’s not. Once that understood, it’s usually not too long before the IT leaders are asking the MSP what should to be done about it.