Percona, PostgreSQL and the Complex Database Landscape
“We’ve strived for a long time to try and reduce the complexity of our corporate systems, especially around the database space. And what has been happening is the exact opposite of a simplification, it’s been making it more complex, it’s been making it more fractured, if you will,” said Matt Yonkovit, Percona chief experience officer.
Among the findings:
- More than 90% of survey respondents use more than one database technology in their organization and 85% work with more than one open source database technology.
- Over half have so far avoided the public cloud and opted for alternatives that allow for easy deployment and scaling but in a more “hands-on” environment.
- Yet more than 38% say that they are using more than one cloud deployment, either with single or multiple providers.
- More than 60% of companies completing the survey do not use database-as-a-service (DBaaS).
- Over 25% of respondents are using containers, but not necessarily to run databases. Many respondents don’t know whether they use containers for their databases.
- Companies are testing the use of Kubernetes for their database environments, but only a third of respondents are actually using it in production. The level of reported experimentation suggests that the number will be growing.
One interesting finding, according to Yonkovit, was how many databases overlap.
“There’s always been this great divide between Postgres and my SQL. It’s been a debate that has raged for as long as those two products have existed. But what we found was 44% of those users who responded, actually use both,” he said.
Though multicloud and hybrid environments are growing, use of DBaaS remains somewhat low, which to some extent Yonkovit attributes to fear of vendor lock-in. Portability remains and issue with DBaaS and the DBaaS technologies don’t necessarily work well together, which poses a potential business opportunity for Percona.
Tony Baer, principal analyst at dbinsight, describes Percona as “herding cats” with its implementations of MySQL and MongoDB, smoothing out the rough edges of incompatibilities and spotty documentation and adding in its own enterprise bits.
Said Yonkovit: “Postgres is truly the most open of all of the most popular open source databases. So the licensing is very open, it’s very easy to contribute, it has a massive following and a massive base of contributors. … There’s a lot of awesome extensions, a lot of awesome features, a lot of awesome open source tools that are out in the ecosystem. And so there’s a lot to draw from, but the problem is, a lot of them don’t necessarily work together.
“It was obvious that a lot of these components are already there, you just need to package them in a way that is designed to work together and minimize the bugs and the friction between these components, these extensions.”
After Percona launched support for Postgres a little over a year ago, customers said they were looking for something that would really stay open and not lock them into a vendor like EnterpriseDB, he said.
The distribution, based on v11.5 of PostgreSQL, takes a few of the most popular extensions to smooth out. They include:
- pg_repack, a popular third-party extension to rebuild PostgreSQL database objects without requiring a table lock.
- pgaudit, a third-party extension that provides in-depth session and/or object audit logging through the standard logging facility in PostgreSQL.
- pgBackRest, a backup tool that replaces the built-in PostgreSQL backup offering for scaling. It uses streaming compression to help companies minimize storage requirements and delta restores to significantly lower the amount of time to complete a restore.
- Patroni, a high availability solution for PostgreSQL implementations that offers the ability to set up replication, and then have the ability to fail over and promote a standby if you need to.
All the work on the extensions will be released back into the community, Yonkovit said. And it will be tackling other extensions with new releases about every three months.
The company’s monitoring and management tool also will support PostgreSQL, as well as MySQL, MongoDB and MariaDB. Rewritten from its initial launch in 2016. It provides detailed query analytics; service-level dashboards; labeling to conveniently group servers for analysis; an API to add, change or remove nodes, security enhancements and more.
Feature image via Pixabay.