How CloudForecast’s Slack Reports Have Helped Bonsai Track Spending Costs
“I would absolutely recommend this product because I think the value that you get for the dollars that you spend is 100% there.”
VP of Engineering
About OMC / Bonsai
One More Cloud (OMC) is a privately-held company that provides hosted search engine services. Specifically, their two products Websolr and Bonsai provide “search as a service” for ecommerce applications, social media sites, magazine archives, classifieds search, job boards, CRMs, CMSs, and more. With over 10 years of experience in implementing and optimizing search, its customers include well-known names such as GitHub, Heroku, Pinterest, Best Buy, Fairfax Media, wikiHow, and many others.
Dru Sellers, the VP of Engineering, is responsible for managing One More Cloud’s AWS spend amongst many other tasks.
OMC powers their search technology with many hundred instances of Amazon’s Elastic Compute (EC2) cloud, with various instance types spread across multiple accounts in multiple regions. This infrastructure is used to run their two services: Websolr is managed Apache Solr as a service and Bonsai is managed Elasticsearch.
Prior to using CloudForecast, Dru wrote his own program to scour AWS APIs to get a sense of how his reserved instances across the different accounts and regions were utilized and managing AWS reserved instance purchase to meet demand.
As a custom program he wrote only to aid him, Dru spent a lot of time absorbing the data and then creating an email with the key numbers to share with his management team. Of course, he had to remember to run it frequently, and set aside time to “digest the data and then send it out to the rest of the team to let them know what was going on”. This took about three hours each week, which, as manager of a “90% developer shop”, he feels is time that he could have spent more profitably helping his team resolve technical challenges.
The Screening Process
In his search for a tool that could relieve him of the time and effort to gain visibility on AWS reserved instance spend, Dru found that the available tools were about managing reserved instance for a percentage of the reserved instance spend. Moreover, these reserved instance-management companies had complex contracts on how reserved instance could change. The savings such solutions offered were not significant, and it seemed easier for OMC to spend some development resources to automate their own reserved instance management solution rather than outsource it.
Of course, the AWS Cost Explorer remains another tool available as a part of the cloud infrastructure, but access to it is minimized to a few essential people for security and compliance reasons. Thus, again, without one of these authorized persons devoting time to use it and create a report based on the data, AWS Cost Explorer remains under-utilized.
The Benefits Of Using CloudForecast
Since using CloudForecast, Dru has been able to share with his team CloudForecast’s visual illustrations of spending at different levels of granularity per service, historical spending data and trends. “Where before I might’ve had to spend three hours, now I’m spending a couple of minutes clicking through the reports to figure out what it is I need to see,” Dru observes.
CloudForecast’s reports are “…more reliable, more robust, and it’s visual already. I don’t have to build visualizations.” He uses the weekly email overview as a springboard for a deeper dive to explore something in their AWS usage. As he says, “When I see your report, I think I need to dig in and really understand where I’m at. Do I need to get in there and fix something?”
One particularly useful feature is CloudForecast’s integration with Slack, their company-wide collaboration tool. Now, cost and spending are visible across teams, often leading to serendipitous insights into wasted spending. For instance, an engineer casually playing with the tool discovered a sudden spike in S3 spending because a new (unadvertised) feature was turned on. The additional spend revealed that this was more than the anticipated revenue for this feature, which was promptly turned off.
Dru notes that team members now “get in and poke around” the CloudForecast reports and question things they don’t understand, giving Dru another reason to probe and “if I don’t already know the answer, then I’ll want to know and so I’ll get in there and dig into it.” This team visibility “alone has been worth it” because they don’t have to give their teams unnecessary access to AWS to be able to analyze their spending and find unexpected cost spikes.
Another valuable feature is the notification received when RI contracts are about to expire. This helps Dru plan for when to buy a savings plan. “There’s been a couple of times when you guys [CloudForecast] have definitely saved me thousands of dollars by not missing a RI drop”, he says. Also, with the spending posted daily, he is able to gauge when to buy more RIs to meet growing demand.
“Where before I might’ve had to spend three hours, now I’m spending a couple of minutes clicking through the reports to figure out what it is I need to see…”Dru Sellers, VP of Engineering
Dru feels that visibility into spending is essential to effect really useful changes — such as happened with their unexpected S3 spend on a feature without enough business value. He says that other businesses that face the same challenges as OMC did prior to signing for CloudForecast need to “get a pulse of what your spend is and where it’s going”.
In parting, Dru notes: “I would absolutely recommend this product because I think the value that you get for the dollars that you spend is 100% there.”
“I would absolutely recommend this product because I think the value that you get for the dollars that you spend is 100% there.”Dru Sellers, VP of Engineering
Michael’s responsibility includes identifying and implementing the most efficient use of cloud resources. To that end, his.