The folks at TimescaleDB have published their ”State of Postgres” survey results in a new micro-site where you can find a summary of responses, some more detailed analysis, and the source data from the survey. This survey was conducted for about 2 months during the late summer/early fall of 2019 and while I haven’t gone through all the raw data as of yet, after reading the results… well, I have some opinions. If you haven’t read it yet, go check it out, it has all the context for the rest of this post :-)
Join, or Die
Because the Postgres project has no single owner, the Postgres community has always been a little bit fractured and doesn’t always speak with one voice. As users, this means the community can look rather different depending on which vendors you work with, the country you live in, the tooling you use, or the online communities you interact with. Since these different groups aren’t always as coordinated as one would hope, initiatives like this can sometimes be harder to push forward, and I think this survey did suffer from that; it only made it out to about 500 people which is a pretty small subset, and you have to keep this in mind before making too large of conclusions about what you see in the data.
Slow and steady growth
39% of respondents have been using Postgres for less than 5 years, with 10% having started within the last 2 years. I’ve seen surveys from communities where they suddenly catch fire and 50% have used it in less than a year, and 90% less than two years (rhymes with shmocker?) and it becomes really hard for those communities to manage that, so this seems like a positive, and helps confirm that Postgres is growing at a solid pace, but not in a way that is likely to be damaging for the community.
You do what now?
Technical titles are hard, but with more than half of the survey respondents reporting some kind of developer-oriented job title, and 50% saying they work in software companies, it is again a good reminder that Postgres isn’t just for DBA’s, and that most peoples interactions with the software are coming from non-traditional outlets. I’ve spent some time coordinating between the Postgres Funds Group and The U.S. PostgreSQL Association this year to ensure a presence at shows like Pycon, Railsconf, and All Things Open, among others, and I hope to see this trend continue into next year.
About those clouds
The answers related to running Postgres on-prem vs the cloud were a bit hard to decipher. We can safely assume about 1/3 of folks are running on fully managed Postgres, but we don’t know how many of those people are also manually managing instances as well. (We do both and I expect others do the same depending on the size/scope of their deployment needs). I feel like I could make a hand-wavey argument that at least 15% of overall respondents are AWS customers, which seems like a pretty big number and will for some will probably exacerbate the rumblings that, relative to their code contributions, Amazon is not contributing their fair share. Granted that isn’t as surprising as the data on the other cloud providers; Azure/Citus didn’t even rank in the poll, which I just have to attribute to a skew based on Timescale’s reach, especially since GCP got a hefty 18%, which seems amazing considering how they have managed their Postgres offerings. (I have friends at GCP and I like the platform in general, but Postgres seems like a second class citizen the way they are currently running things)
Those quotes
Oy vey. I’m not sure if Timescale was picking quotes just to stir up some controversy (there are certainly more friendly ones in the raw data), but the quotes about NoSQL are a bit off-putting. This is an area where the community needs to continue improving because we have a reputation for sometimes being “stand-offish”. Not in all cases of course, but if you want to find people with strong opinions who are not afraid to speak out, the Postgres community has lots of them. (Perhaps this blog post is a case in point) Anyway, given at least 50% of respondents are using at least one NoSQL system in conjunction with Postgres; and based on modern infrastructure patterns that isn’t going to change; we need to learn to focus on helping people where they are, rather than where we think they should be, and being less abrasive about it in general.
All in all, I hope this information will be useful for the community, and I want to thank the Timescale folks for publishing the results (and the raw data), and I hope they will continue to do this and/or work within the community to expand the reach of this survey next year.