The importance of cloud has increased over the years and a lot of workload is shifted to it to make the most out of the available business resources. Cloud is rising high and getting bigger day by day but a serious concern of customers’ is its authenticity. It’s like doubting your own dashboard about the services it provides. This trust deficit on the customer’s part may damage the increasing popularity of cloud computing.
In a situation like this, Heroku’s Rap Genius dispute is important. Rap Genius about two years ago, observed performance issues with its Ruby applications running on Heroku platform. Even with high traffic on the network, the performance still lacked. Though Heroku’s log files and dashboard didn’t show the mess, Rap Genius faced unsatisfactory service queries from its clients.
It seems Heroku fiddled the readings of the site’s router raising a problem in ideal job execution and instead of “intelligent load distribution”, it was shifted to “random load distribution”. The point to be concerned is that Heroku didn’t file those changes and revealed the problem to customers alone, without taking Rap Genius on-board. Heroku is a PaaS service provider purchased by Salesforce.com in 2010.
In a blog post by Rap-Genius officials posted on February 13, the company stated:
“A Rails dyno isn’t what it used to be. In mid-2010, Heroku quietly redesigned its routing system, and the change — nowhere documented, nowhere instrumented — radically degraded throughput on the platform. Dollar for dollar a dyno became worth a fraction of its former self.”
The Rap Genius blog post has highlighted the latest hacking news and possibly endorsed Heroku’s apology published by TechCrunch.
In a recent phone interview, Tom Lehman, Rap Genius co-initiator explained they had 90 dynos operating on $20,000 per month, which were considered adequate based on the data figures the company was receiving from Heroku. However, in actual, 90 dynos were not enough. After getting informed about the problem, Rap Genius increased to 300 dynos for $40,000 per month, but still the performance didn’t boost, and the company was unable to pay $40,000 per month.
A detailed apology by Heroku was published on February 16, summarizing the action plan for Heroku that includes:
- Enhancing the documentation standards so they depict the real situation of the services both for Bamboo and Cedar stacks.
- Eliminating the erroneous and ambiguous metrics stated by Heroku or associate services like New Relic.
- To help customers identify and verify the impact of queuing on application response time, upgrading the metrics.
- To enhance Heroku’s latency and queuing metrics, offering added tools for the developers.
- Working to improve assistance of simultaneous-request Rails apps on Cedar.
Lehman further added that if nothing changes, they’ll shift to a new platform as the company can’t bear $40,000 per month. The expected destination for Rap Genius is Amazon Web Services, but it won’t be easy because Heroku was doing more than AWS. Though most of the Rap-Genius third-party providers are on AWS, Lemans said, “I still have love for Heroku. Without it we couldn’t get to where we are today but they have not been 100 percent upfront with customers.”
The Heroku’s dishonesty has raised an important concern for all the companies that still believe the transition to cloud leads to loss of control over IT, What if they start thinking that metrics given to them by providers aren’t real, but fabricated?