The Cost of Manual Infrastructure Work
Manual infrastructure tasks have a compound cost: the time to perform them, the time to document them, and the incidents caused when someone performs them slightly differently each time. Digicom Technology Solutions was running repetitive AWS operations multiple times a week, tasks that required an engineer's attention but didn't require engineering judgment.
John Newcomb's brief: automate the workflows, make them auditable, and make them runnable by a non-engineer.
Why PowerShell
PowerShell was the right choice for a specific reason: the Digicom team already lived in Windows environments and had existing PowerShell scripts. Introducing Terraform, Ansible, or a Python-based toolchain would have created a knowledge gap and an ongoing maintenance burden. We built on what the team already knew.
The Automation Suite
EC2 fleet management
Scripts for starting and stopping EC2 instances by tag, dev environments that should run during business hours, batch processing instances that should run only during scheduled windows. Scheduled execution via AWS Systems Manager State Manager eliminated the 8 AM "start the dev servers" ritual.
S3 lifecycle archival
Assets older than 90 days move automatically to S3 Glacier. A weekly summary report is generated and emailed via SES, showing what was archived, the storage savings, and any exceptions. The ops team no longer needs to remember to do this, it just happens.
RDS snapshot management
Daily automated snapshots with a 30-day retention policy. Snapshots older than 30 days are deleted automatically. Cross-region copy to a second region for disaster recovery. All of this previously required manual execution by whoever remembered to do it that week.
Cost anomaly alerting
CloudWatch alarms trigger when daily spend exceeds defined thresholds. Alerts route to a Slack webhook, the ops team sees the notification in their existing workflow rather than having to log into the AWS console to check billing.
Scheduling via AWS Systems Manager
We used AWS Systems Manager (SSM) to schedule and execute the PowerShell automations rather than running scripts from a cron server. SSM provides a full execution log, automatic retry on failure, and IAM-controlled access, no credentials stored on local machines, no single point of failure if a server is down.
Making It Non-Engineer-Friendly
A key requirement: a non-engineer should be able to trigger the common workflows. We wrapped the most-used scripts in a minimal web interface, a Lambda-backed HTML page protected by Cognito, where ops team members can run a predefined automation with a button click and see the output inline. No AWS console access required.
Result
The 4–6 hours of weekly manual operations are now fully automated. The suite has been running without modification for over a year, a sign that the initial design was solid. John Newcomb's summary: "Khalil was superb. He was responsive, knowledgeable, friendly, and a real professional. Hire him, you won't be disappointed."
Key Takeaway
Automation ROI compounds. The ops hours you save this week are also saved next week, and every week after that. For recurring infrastructure tasks, the question isn't whether to automate, it's how quickly you can recover the automation investment from the saved time. For Digicom, the payback period was under two months.