As a new member of the Percona team, one of the first things I was interested in understanding was our software adoption. I decided to take a look at the information we have about adoption (instances running) of Percona Monitoring and Management (PMM) and found some very positive trends. First, I looked at overall instances as of the end of January 2020, which showed that PMM had an average of over 6,000 instances daily, with 14% of those being on version 2, which is pretty good considering it has only been out for around 4 months.
I then dug in a little bit deeper to look at the adoption of our latest release of PMM 2 which was released on September 19, 2019, so I started my analysis beginning in October for the ease of looking at full month comparisons.
First, when I looked at the average total instances running by month for PMM 2, I see a nice overall adoption with an overall growth of 175% for the 4 months, with our January monthly average at 850 instances.
The PMM team has a regular sprint cycle which results in 2.x releases about once a month, so the next thing I wanted to understand was if our customers are picking up the latest releases or are they sticking with older versions. As you can see in the graph below, as new releases come out, our customers are pretty quick to download and use the latest release.
This information is important because it lets us know that customers are quick to pick up the latest release and use the new features. It also means we need to make sure that our customers know when these releases are coming, where to get them, and what’s new for them to try out. Which is why PMM provides this information on its main dashboard.
You can see the current version you are running, the last time it checked for a new version, and if there is a new version available. You’ll also notice we provide a one-click update to the latest version. This could be why the adoption of the latest release is so high.
I have worked at many companies where understanding what customers are doing with your software is a huge challenge because there was little to no data. The best data we had was the number of downloads, which doesn’t really tell you what customers are actually running. Or, we talked with our support teams about what they were seeing being run by customers they were working with, but it is hard to track and you don’t really know if that is representative of your whole customer base.
So, naturally, I wanted to understand how Percona is getting such insightful data. When a user installs PMM Server, it sends a request to our “Check Version” system. When it sends this request, it provides the PMM version and unique ID information. This is what enables Percona to get an understanding of how many users we have for which versions of PMM.
Customers can always opt-out of sharing this information with us. You can easily switch this off via the on/off switch for Call home on the PMM Settings panel off the main dashboard.
You’ll notice that we also give you a separate switch for checking for updates. This ensures that even if you aren’t enabling a call home, you can still pull the information about the latest version and do a local check against what you are running to determine if you are up to date.
One question that the data naturally leads to is, “If we could ensure full backward compatibility, would customers want automatic upgrades?”. What do you think? Share your thoughts in the comments below or on Twitter @tlschloss.