Greater variety of database platforms increases IT options

Nmedia - Fotolia

Cloud-managed database services change data administration

Managed data services are growing in use, as types of data stores proliferate and the cloud becomes the home for more data. DevOps is a driver behind the changes, which bring new duties and needed skills for DBAs.

Many changes descended upon database administration in recent years, as developers selected new database types that challenged the relational database's enterprise dominance and cloud computing grew in use.

Among the more recent changes is added reliance on managed database services that relieve user shops of some administrative tasks, while supporting developers' needs to connect varied data sources. The changes can be chalked up to the drive to create applications more quickly, and to use DevOps techniques to do so.

According to Donnie Berkholz, an analyst at 451 Research, the role of the database administrator (DBA) is undergoing the same kind of transition that systems administrators had to go through during the past 10 years. At the center of that transition is DevOps, the Agile methodology that embraces new, often open source technology and makes developers more responsible for how their applications work in production.

The surfeit of data stores is also a learning challenge for the DBA, both in their numbers and their diversity.

"With DevOps, you can't be static. The technology is changing very fast," Berkholz said. "Data administrators now must serve all kinds of data to the business, and few will get by simply maintaining a relational database."

At the same time, the managed database services capabilities of cloud providers and others continue to expand, along with the number of available database as a service (DBaaS) options. Players here include Amazon Web Services, IBM, Microsoft, Oracle, Google, Rackspace and others -- and activity is on the rise.

For example, NoSQL document database maker MongoDB recently joined the competition with its Atlas DBaaS offering. And just last week, Google moved some of its long-brewing beta technologies into general availability. Moreover, Google Cloud SQL, Cloud Bigtable and Cloud Datastore now come with service-level agreements, the hallmark of enterprise database administration.

Waiting on the platform

Increasingly, database tooling centers on REST-oriented APIs tuned to meet developers' needs to connect diverse data stores. Berkholz cited IBM's homebrewed Bluemix platform as a service, as well as its purchases of DBaaS companies Cloudant Inc. and Compose Inc. -- in 2014 and 2015, respectively -- as examples of the trend.

The data store types Compose supports are quite numerous, including PostgreSQL, Elasticsearch, RethinkDB, Redis, at least three versions of MongoDB and more.

"Since the data is easily available in the cloud with database as a service, developers can focus on doing cool things," said Larry Weber, a program marketing director at IBM. That means more building of new applications and less administration of existing ones.

Meanwhile, IBM research showed DevOps-era stalwarts may be ready to give up some of the newer data administration work they've had to take over. A survey of Cloudant executive and technical users in late 2015 showed 41% of enterprises found their biggest database management challenge was the required management time. This was cited by respondents as one of the drivers behind their moves to the cloud.

Blurring versions and instances

As data options in the cloud have proliferated, the need for expertise in connectivity has grown, according to Chris Lalonde, vice president and general manager of managed data services at Rackspace Inc., based in San Antonio. The cloud hosting company was early to up its data services capabilities, purchasing DBaaS specialist ObjectRocket -- where Lalonde was co-founder and CEO -- in 2013.

Among the database administration know-how ObjectRocket brought to Rackspace was expertise in PostgreSQL, MongoDB, Elasticsearch, Redis and other data stores. This year, Rackspace has also moved forward support for data services on the Amazon Web Services cloud. Such breadth of support is hard for in-house DBAs to match.

"We have the expertise to spin up databases, of course, but there is a lot more than that needed now," Lalonde said. "With DevOps, there is a lot to manage.

"You need to know if an instance needs more disk space, which version of Python your developers [should] be using and so on. We do a lot of that type of work for customers. In that sense, we become the customers' DBA."

Next stop: Hadoop in the cloud

Lalonde said he anticipates more interconnectivity between more types of data stores in years to come. This will further challenge DBAs, he suggested.

When relational databases were young, developers played a more hands-on role in operations. As the technology matured, the DBA assumed major responsibility for data in production. What's different now is the wide variety of built-for-purpose data stores. On top of that, the move to the cloud appears to augur continued change in database administration processes going forward.

Changes are also being felt in the Hadoop analytics and data warehousing spaces, which are feeling the effect of efforts to tame them into managed cloud services.

Next Steps

Find out how a SAS expert defines data as a service

Learn how the role of the DBA is morphing

Look at how CDOs wrestle more value out of data

Dig Deeper on Database management system (DBMS) architecture, design and strategy