An interview with Heroku on Postgres

Posted By: PostgresOpen on Tuesday, August 22

We got the chance to sit down with Rimas Silkaitis from the Heroku Data team and dig into some of the history behind their choosing and becoming the first database as a service provider and how the see things evolving.

PostgresOpen SV: Heroku Postgres helped drive much of the resurgence of Postgres, why do feel it was often overlooked several years ago?

Rimas: It depends on where it was being overlooked and the time frame. If we're talking about Heroku's sweet spot, building web applications, 10 or more years ago, then Postgres was definitely overlooked. Postgres took the approach of focusing on security, being standards-compliant and having a robust feature set. Other databases at the time, focused on ease of use. For many developers looking to build web applications, and for many for the very first time, ease of use was paramount. Because of the approach the community took on Postgres, it provided fewer opportunities for developers to shoot themselves in the foot as their application grew.

PostgresOpen SV: You've been running a database as a service for some years now, in fact you were one of the first to run Postgres as a service, how have you seen managed database services change?

Rimas: When Heroku Postgres started being able to run Postgres as a service or any database for that matter was no small feat. That was the value of the service. Over time what we've seen is that getting the database service running is the minimum bar. Any DBaaS needs to go beyond just providing the service and do more for the customer. This means providing more features and functionality around operating the database that may not be inherently built into the database itself. For example, Heroku Postgres has commands and functions that will diagnose potential problems with the database called PGDiagnose. It's those kinds of features that developers are really looking for so that they don't have to worry about the database.

PostgresOpen SV: For most people trusting you with their data is a big decision, what are they able from a database as a service that they couldn't achieve otherwise.

Rimas: I'm a bit biased because I lead Heroku Postgres but using a DBaaS should be a no-brainer. So much goes into running a database effectively that if you want to do it on your own you're going to need so much infrastructure to pull it off. That's just the fixed costs. What about the people power needed to keep that database running? But the real value for anyone using a DBaaS is not all of the features and processes that comes from buying that product but that time that you and your team gets by focusing on the areas that matter to the business. Will you make money by building your own infrastructure to support a high-availability setup or does it come from time your team takes to build the application that actually generates revenue? The really good DBaaS providers will make sure that they give you and your company the confidence that your data is secure by providing all of the necessary transparency and data portability. This way you can be confident that you are in control.

PostgresOpen SV: What are you most excited for about the direction of Postgres and its future?

Rimas: We're really excited about the work being done around partitioning. In postgres 10, the community is making it easier to use and making it part of the database itself. While table partitioning does not get Postgres in the distributed system realm, it does point to a future where Postgres could get there. Companies like Citus and projects like TimescaleDB are already showing us that it's possible. Postgres will become the database choice to handle all sorts of applications and their use cases, it's only a matter of time.