Heroku confirms three-year slide in Bamboo stack performance

News Jane McCallion Feb 15, 2013
Man looking confused with his computer
Man looking confused with his computer

PaaS provider comes clean over routing system redesign

Platform-as-a-Service (PaaS) provider Heroku has admitted apps running on its Bamboo stack have experienced a steady degradation in service over the past three years.

The admission by the company’s general manager, Oren Teich, comes in response to a blog post from one of its biggest customers, Rap Genius.

Developers at Rap Genius accused Heroku of having “quietly redesigned its routing system [in 2010] and the change – nowhere documented, nowhere instrumented – radically degraded throughput on the platform.

“Dollar for dollar a dyno (virtualised Ubuntu server) became worth a fraction of its former self,” the post continued.

Rap Genius claimed the switch from intelligent load distribution to random load distribution had caused the degradation.

In response, Teich said: “In short, Ruby on Rails apps running on Bamboo have experienced a degradation in performance over the past 3 years as we have scaled.

“We failed to explain how our product works. We failed to help our customers scale. We failed our community at large. I want to personally apologise, and commit to resolving this issue.”

Teich said Heroku would be posting an in-depth technical review at some point today and would generally be working towards increasing customer understanding of how Bamboo works.

While some users said they appreciated Heroku’s honesty, Teich was criticised for not adding “fixing the problem” to the list of actions it would take.

Commenter Abe Burnett said: “I appreciate that you're taking responsibility and not shifting blame. Yet, the RapGenius story raises some serious concerns about the viability of Heroku as a platform for large-scale applications.

“Better, more accurate communication is important, absolutely; but so is ensuring that the underlying service problems are addressed.”