BOSTON -- A general push for agility has led to increasing influence for developers in data decisions. That, in...
turn, is altering both the DBA's role in the data lifecycle and the way open source relational databases like PostgreSQL are being deployed.
Moreover, the cultures of IT and data are changing, according to Santiago Comella-Dorda, a partner at McKinsey & Co. He said he sees a more dispersed role for databases and database administration in the organization going forward.
"We are getting to a place where database administration is not a role, but a capability, and it gets distributed," said Comella-Dorda, who this week led a panel on issues in companywide agile database development at EnterpriseDB's Postgres Vision 2017 conference here.
Database schema, once mainly the province of database administrators (DBAs) working in centralized groups, is one of the things being distributed.
"My team is not responsible for designing schema. The teams we support are empowered to do that on their own," said panelist Matthew Spilich, director of engineering operations at TripAdvisor Inc., who marks the travel website company's development style as agile.
Sea change in data roles
But distribution of data schema responsibility is not a one-way street, according to Spilich. At TripAdvisor, based in Needham, Mass., some teams work on schema independently, while others tap his centralized group for guidance. Still, agile database development is something of a sea change.
"The traditional role of the DBA can become a culture of 'no' -- where the DBA is the gatekeeper that doesn't let things into production," he said. Today, technology managers have to change the way in which that role is distributed to other people, he continued.
Another panelist, Joe Schuler, who is vice president of global data architecture at Mastercard, has seen changes in data management as agile development methods have arrived. Notably, DBAs that were part of an architecture team are now part of operations.
"It's an age-old debate," he said. He described the organizational transformation wrought by agile database development as a move "from organizations to tribes and guilds." Schema responsibility is more federated, while data auditing and compliance work remains centralized.
Data status quo in flux
If DBA roles are being distributed, so is database technology. Much of this is driven by interest in cloud computing.
Teams are moving to "build data services, not databases," according to Marc Linster, senior vice president for product development at EnterpriseDB, based in Bedford, Mass.
"As we are looking at very fast and agile deployments, we are looking at database as a service or integration with platform as a service," he said.
With database as a service, or DBaaS, "you stop deploying databases, but you start providing a service of data persistence," he said. "That is a complete change for DBA organizations."
Marc Linstersenior vice president for product development, EnterpriseDB
Over time, DBAs have prided themselves on providing a database, but developers do not really care about the database, Linster claimed. What developers want to know is that there is a reliable service available for them to write data to, get data back from, to execute a transaction and to know the data is still there tomorrow, he said.
"The fact that a database is needed -- today, it's almost irrelevant to the developer," Linster said.
EnterpriseDB has produced a framework for DBaaS, known as EDB Postgres Ark, that helps departments make the move to data as a service, Linster said. Meanwhile, developers' use of "schemaless" data architecture is enhanced by JSON support that EnterpriseDB now includes with its version of Postgres.
Growing Postgres SQL support for DBaaS is mirrored in other open source relational databases, such as MySQL and MariaDB, but Postgres SQL services have become a particularly notable part of the developer-driven move to data on the cloud. Postgres SQL has become part of cloud offerings from cloud providers such as Amazon, Google, IBM, Microsoft and Rackspace.
What seems certain is, with so many choices and so much change, additional agility will be needed by DBAs and developers alike in days to come.
Be there when data engineering meets DevOps
Learn how cloud computing challenges conventional database administration