AWS Optimization Calls in a Win for ExecVision
Get the Job Done
Fuzzy Logic was brought in by ExecVision to maintain, document, and improve the company’s call transcription, indexing and search product, which required consistent uptime and fast response.
Evaluating the environment, principal Lee Whalen quickly recognized two opportunities: lower server costs and reduce customer-facing website outages. He identified three primary issues contributing to both outages and high server costs:
- Long provisioning times
- Lack of operational visibility
- Underperforming cloud services
To address these issues, Lee/Fuzzy Logic:
- worked with the team to standardize the system configuration using familiar, readily-available tools
- introduced new instrumentation for monitoring and metrics
- optimized AWS spend by right-sizing VMs and deploying Reserved Instances where appropriate
- migrated costly cloud services to self-hosted solutions
In Search of Better Provisioning
There are times when more complex changes are beyond a contractor’s scope. In these cases, flexibility, deep knowledge of multiple toolchains, and an understanding of the client’s business goals and KPIs are required for the best possible outcome.
When Fuzzy Logic came on board, the ExecVision team was using Ansible to perform configuration management. A disorganized provisioning environment, lack of documentation, and no access to the previous contractor created additional issues.
After his recommendation to adopt a different configuration management system met with resistance, Lee got to work making ExecVision’s Ansible provisioning the best it could be.
To make that happen, he undertook two major tasks:
- Documenting the existing system
- Refactoring the provisioning environment to better match Ansible’s recommended practices
To begin, Lee explored the environment, host by host and task by task, documenting from first principles. To ensure that future team members would be able to follow his work, he recorded his steps in the previously-barren company wiki.
Having deciphered the system, he was able to streamline Ansible provisioning for greater deploy reliability, code readability, and speed. In the end, a process that previously demanded an average of six runs and two hours per VM required only two runs and twenty minutes.
The Case of the Missing Website
Sometimes, you don’t know what you don’t know. When Fuzzy Logic entered the picture, ExecVision suffered from poor uptime and frequent application errors that came out of nowhere: the website would be running well enough when a concerned customer would call in, saying, “Where are my transcriptions?“ or “I can’t log in!”
An impromptu all-hands meeting would be called to discover the issue. Signs pointed in all directions—
Looking at the cost of cloud services, Lee realized that right-sizing virtual machines would be an easy way to reduce costs without compromising customer experience.
When a company uses cloud services like Amazon Web Services (AWS) for a small project, it’s easy to make each VM larger than is ultimately needed. Initially, the price isn’t so high—but when the project grows to the size of ExecVision’s call transcription product, the cost of these roomy VMs begins to mount.
Assembling the size and usage data for each node, Lee demonstrated how sizing them more realistically would save the company money. He also showed how ExecVision could spend substantially less by trading their existing ‘on-demand’ hosts for ‘reserved-instance’ VMs. In addition, Lee mapped out a move from hosted services like GitHub and TravisCI to site-optimized self-hosted solutions using GitLab and GitLab-CI.
With trust established during the config-management and monitoring projects Fuzzy Logic worked with ExecVision to make additional optimizations to their infrastructure, substantially improving their bottom line.
Ultimately, ExecVision opted for a year-by-year contract for reserved-instance VMs that saved the company upwards of thirty-three percent off the previous cost of their cloud services and was able to further reduce OpEx by migrating from hosted to self-hosted services.
In a short six months, Fuzzy Logic evaluated and documented an existing system, laid out a case for making multiple changes, and implemented those changes in a way that saved ExecVision time and money, and reduced customer-facing outages.
Reuben Firmin, CTO of ExecVision, says this:
_ Lee brings stability and professional sysadmin expertise to the table. His manner is always calm and methodical, and you can count on him to get the job done efficiently._
Key contributions included:
Based in Seattle, Washington, Fuzzy Logic serves clients worldwide, assisting development and operations teams in leveraging the power of the latest technologies in virtualization, server administration, communication, and deployment to deliver solid products that are efficient, faster to market, and customer-friendly.
Fuzzy Logic’s clients operate in diverse fields such as digital media, data analytics, healthcare, social networking, and banking.
ExecVision is changing the way executives understand their business. ExecVision’s Conversation Intelligence™ platform is the next frontier of data gathering, transforming call recordings into data, and then turning that into actionable intelligence to improve sales conversation quality.