This article is more than 1 year old
Datadog allegedly asked developer to kill open source data export tool
But the code creator has revived the abandoned code for OpenTelemetry
Updated After a delay of over a year, an open source code contribution to enable the export of data from Datadog's Application Performance Monitoring (APM) platform finally got merged on Tuesday into a collection of OpenTelemetry components.
The reason for the delay, according to John Dorman, the software developer who wrote the Datadog APM Receiver code, is that, about a year ago, Datadog asked him not to submit the software.
On February 8 last year Dorman, who goes by the name "boostchicken" on GitHub, announced that he was closing his pull request – the git term for programming code contributed to a project.
"After some consideration I've decided to close this PR [pull request]," he wrote. "[T]here are better ways to OTEL [OpenTelemetry] support w/ Datadog."
Members of the open source community who are focused on application monitoring – collecting and analyzing logs, traces of app activity, and other metrics that can be useful to keep applications running – had questions, claiming that DataDog prefers to lock customers into their product.
Shortly after the post, Charity Majors, CEO of Honeycomb.io, a rival application monitoring firm, wrote a Twitter thread elaborating on the benefits of OpenTelemetry and calling out Datadog for only supporting OTEL as a one-way street.
"Datadog has been telling users they can use OTEL to get data in, but not get data out," Majors wrote. "The Datadog OTEL collector PR was silently killed. The person who wrote it appears to have been pressured into closing it, and nothing has been proposed to replace it."
Behavior of this sort would be inconsistent with the goals of the Cloud Native Computing Foundation's (CNCF) OpenTelemetry project, which seeks "to provide a set of standardized vendor-agnostic SDKs, APIs, and tools for ingesting, transforming, and sending data to an Observability back-end (i.e. open source or commercial vendor)."
That is to say, the OpenTelemetry project aims to promote data portability, instead of hindering it, as is common among proprietary software vendors.
The smoking hound
On January 26 Dorman confirmed suspicions that he had been approached by Datadog and asked not to proceed with his efforts.
"I owe the community an apology on this one," Dorman wrote in his pull request thread. "I lacked the courage of my convictions and when push came to shove and I had to make the hard choice, I took the easy way out."
"Datadog 'asked' me to kill this pull request. There were other members from my organization present that let me know this answer will be a 'ok'. I am sure I could have said no, at the moment I just couldn't fathom opening Pandora's Box. There you have it, no NDA, no stack of cash. I left the code hoping someone could carry on. I was willing to give [Datadog] this code, no strings attached as long as it moved OTel forward. They declined."
He added, "However, I told them if you don't support OpenTelemetry in a meaningful way, I will start sending pull requests again. So here we are. I feel I have given them enough time to do the right thing."
- AWS fixes 'confused deputy' vulnerability in AppSync
- Hacker bites Datadog, finds hard-to-chew bcrypt passwords
- Container adoption still low, barks Cloud Foundation
- Security pros get ability to manually add incidents to Microsoft Sentinel
Indeed, Dorman subsequently re-opened his pull request, which on Tuesday was merged into the repository for Open Telemetry Collector components. His Datadog ARM Receiver can ingest traces in the Datadog Trace Agent Format.
Coincidentally, Datadog on Tuesday published a blog post titled, "Datadog's commitment to OpenTelemetry and the open source community." It makes no mention of the alleged request to "kill [the] pull request." Instead, it enumerates various ways in which the company has supported OpenTelemetry recently.
The Register asked Datadog for comment. We've not heard back.
Dorman, who presently works for Meta, did not respond to a request for comment. However, last week, via Twitter, he credited Grafana, an open source Datadog competitor, for having "formally sponsored" the work and for pointing out that Datadog "refuses to support OTEL in meaningful ways."
The OpenTelemetry Governance Committee for the CNCF provided The Register with the following statement:
"We're still trying to make sense of what happened here; we'll comment on it once we have a full understanding. Regardless, we are happy to review and accept any contributions which push the project forward, and this [pull request] was merged yesterday," it said. ®
Updated to add
The Register asked Dorman whether the situation has been resolved to his satisfaction. In an email response received after this story was filed, Dorman said, "There was nothing to be 'resolved,' once Grafana decided to sponsor the work I opened the PR again."
"I stand by what's in the Pull Request, which is that I closed it, no guns to my head, out of professional courtesy," he explained. "I also re-opened it, no guns to my head, for the reasons Grafana mentioned. I am happy it is merged and am happy to be moving on to other things."