Databases 2019: what will happen in a hybrid, cloudy world?

What can we expect from databases in the year ahead, and why does the move to cloud matter?

This is a contributed article by David Waugh, Senior Vice President, Market Development at DataStax


Staring into the crystal ball can be challenging - what was magical, impossible technology even a few years ago is now business as usual, and markets that have developed in specific niches for years are now blossoming based on access to funding, market interest and new use cases. What never changes? Data.

Data sits at the heart of everything that IT does for the business. Where does data live? Within databases. So, how will the move to cloud, to more agile digital strategies, and to new ways of providing service affect the range of database technologies? How will venerable technologies that have been with us for decades fit alongside new approaches, and where can they all play nicely together?


Prediction #1: Hybrid cloud is going to grow even more

Businesses are moving more and more to hybrid IT models, making use of cloud across their own IT and public cloud services. Around 71 percent of companies are now hybrid according to RightScale research, while 81 percent said they ran multi-cloud environments. This move to multi-cloud is only going to increase in 2019.

Why is this? Put simply, going all in on cloud seems like it should be simple, but for enterprises it is hard to achieve in practice. Forrester's Lauren Nelson describes it as: " … completely switching from today's current infrastructure reality to one based 100 [percent] on public cloud is so prohibitively expensive that no one's doing it."

Running everything in public cloud - when you have existing IT infrastructures that work, work well, and are already integrated into your business processes - doesn't yet make sense for enterprise IT teams. Why? Because replicating only what currently exists is not enough of an incentive, based on the cost model that can be involved. Aside from this, going all in with one cloud provider can limit options and choice in the future, regardless of how many services or new options may come through in the future.

All of these trends mean that hybrid approaches will carry on growing.


Prediction #2: Data will get more distributed as it grows

Alongside the growth of hybrid cloud, databases will have to follow suit. If your computing environment is spread across multiple locations and multiple service providers, but your data is held in one central location, then your applications will suffer. And as many enterprises are discovering in this digital first age, when their applications suffer, their business suffers.

Distributing data solves two problems - it removes some of the latency issues that can otherwise affect performance at scale. It provides more resiliency for the application, as it can carry on processing while specific sites are down. However, it does mean more thought around how to link up hybrid cloud and data is needed.

By 2020, most data and analytics projects will need access to some form of distributed data in order to provide the results that the business requires. When data underpins those applications, deploying data close to the application and making it flow properly will be essential to design. Hybrid cloud designs will therefore have to include more thought around data layers and database designs too.


Prediction #3: Data portability will become an issue for more companies

Migrating to cloud has been a common trend for the past few years. Ask many CIOs, and they will be proud of the new services that are up and running. However, ask them where the data goes, and the smiles may falter a little. It's all too easy to treat data so it sticks to where it is being created - so if you run that application on one cloud, the data stays there as well.

Effectively this becomes a silo in its own right. While one of the main reasons to move to cloud is to accelerate the development and deployment of applications, that benefit is often diluted with the creation of additional data silos. In fact, moving to the cloud has often accelerated, not mitigated, the creation of data silos. This is an unwelcome consequence of limited planning around the data layer and data architecture. It also means less communication between different applications. When one of the biggest challenges around traditional applications is the integration between different systems, this is an area that can't be ignored.

Data portability covers two areas: one, that data can be easily moved from one cloud back on-premise or to another provider. Two, this movement should be possible at any time, and with no impact to the application itself. If we don't achieve this level of flexibility around data, then we will remain stuck with data silos in more places.


Prediction #4: There will be more focus on database design

Over the past few years, the interest in different database designs has led to a plethora of new market options. From traditional relational products supporting cloud deployment, through to open source projects like Apache Cassandra™ offering multi-cloud support, to new graph database designs, there are now many more options when it comes to supporting and understanding data.

Getting the right data infrastructure in place around your data means understanding what you want to achieve with that data, and how your application should be best supported. If you are looking to scale up in one place, then more traditional database designs will still be the best fit. Looking for that more hybrid approach to IT? Then distributed data modelling will be important, especially for companies that want to run across multiple sites and cloud providers.

Databases continue to be at the heart of new applications and moves to cloud. However, much of the thinking and planning around databases is based on traditional data management ideas. It is long past the time for thinking about databases to focus on that issue of distributed data, and how it affects everything around applications under the covers. Without this thought, however, cloud deployments will continue to be more of the same, rather than meeting their potential.