I work at Grafana, and just attended KubeCon, even when not near the booth but just in Grafana t-shirts I was stopped frequently by individuals saying how much they love Grafana, how it just works, how we should just keep doing what we're doing... there's strong signal there (though this is KubeCon, and K8s isn't the whole industry by a long shot).
But even I would rationally question the survey results, as like all surveys it would only represent those surveyed.
If you scroll to the bottom of the article there is a section "About the respondents and methodology" and to quote, "This year, 306 people participated in the Grafana Labs Observability Survey. We did not contract an outside agency to solicit responses. Instead, we reached out to our community to participate, through our website, social media, and at various Grafana Labs and OSS events".
So let's be upfront here, it's a biased and small sample, but the consistent methodology with prior years means that the directional trends can be identified, even if you cannot extrapolate any specific number to the whole industry.
The indicative signals in there for trends and direction... but it's not quite a 100K audience from all industries and representative of wider engineering.
The number is high due to who was asked to respond.
This is just a piece of marketing material - it's the survey of ~300 Grafana customers. Congrats to Grafana for getting their advertisement posted to this Website.
>About the respondents and methodology. This year, 306 people participated in the Grafana Labs Observability Survey. We did not contract an outside agency to solicit responses. Instead, we reached out to our community to participate, through our website, social media, and at various Grafana Labs and OSS events.
This should be right at the top as it indicates a substantial bias in the data.
Calling this survey "Key findings and analysis on the state of observability" is disingenuous at best with that methodology. If this team realistically thinks this small sample size and narrow grouping gives an accurate analysis of the big picture, I would be hesitant about any analysis their product can provide.
A lot of software is purchased that never gets used. Sometimes the managerial sponsee leaves the company after acquiring the software for instance. The vendor will update their CRM with a new contact. Or, the vendor is trying to expand a customer and has new contacts in other parts of the company that haven’t converted yet. Or a customer that had converted but has not yet deployed. They could still be working with customer education and consulting services and/or CSM to get their deployment squared.
There seems to be newer players both open source and commercial now. But a lot of the focus seems in metric and specially distributed traces. Does anyone in the open source do code level profiling (cpu, allocation, locks etc.) preferably sampling profiler (including Java)?
