Nimbus Logistics: -63% in cloud infrastructure costs

The challenge

Nimbus Logistics had a problem that sounds familiar to any startup: AWS bill kept growing, nobody knew why, and the small engineering team was spending more time on infrastructure than on product.

The situation:

• $40k/month AWS bill (way too high for a Series A startup at their stage)
• Single AZ deployment (downtime risk, not enterprise-grade)
• Oversized EC2 instances (running at 20% utilization)
• Untagged resources accumulating for 18 months (orphaned EBS volumes, unused load balancers)
• No auto-scaling (manually scaling for traffic spikes)
• No observability (no idea what was actually being used)
• 2 engineers spending 50% of their time on infrastructure

They needed:

• 50-70% cost reduction without sacrificing reliability
• Multi-AZ deployment for enterprise SLAs
• Auto-scaling for traffic spikes
• Observability (know what’s running, what’s expensive, what’s broken)
• Self-service infrastructure for their small team

What we built

Cost audit and right-sizing

We started with a 2-week deep audit:

• Tagged all untagged resources (200+ items found)
• Right-sized EC2 instances (many oversized by 2-4x)
• Identified unused resources (50+ orphaned EBS volumes, unused EIPs, unused snapshots)
• Analyzed traffic patterns to inform auto-scaling config

First-month savings: $8k (just from right-sizing and cleanup, no architecture changes).

Reserved instances + savings plans

For predictable workloads (databases, core API servers):

• 1-year reserved instances for baseline capacity
• 3-year savings plans for additional predictable usage
• On-demand for everything else (auto-scaling peaks)

Additional savings: $12k/month.

Multi-AZ + auto-scaling

Moved to multi-AZ deployment for HA:

• RDS Multi-AZ for databases (automatic failover)
• EC2 Auto Scaling Groups for app servers
• Application Load Balancer for traffic distribution
• CloudFront CDN for static assets

Reliability improvement: from 99.5% (single AZ) to 99.95% (multi-AZ) measured uptime.

Terraform for Infrastructure as Code

Converted all manual infrastructure to Terraform:

• Version-controlled infrastructure
• Reproducible environments (dev, staging, prod identical)
• Easy to onboard new engineers
• Audit trail via Git history

Operational improvement: 1 engineer can manage the entire infrastructure (down from 2 engineers at 50% time each).

S3 lifecycle policies

Implemented automated S3 lifecycle policies:

• Frequent access: S3 Standard (first 30 days)
• Infrequent: S3 Standard-IA (30-90 days)
• Archive: S3 Glacier (90+ days)
• Delete: 365 days for non-critical logs

Additional savings: $3k/month.

Datadog observability

Implemented comprehensive observability:

• Infrastructure metrics (CPU, memory, disk, network)
• Application metrics (request rate, latency, error rate)
• Log aggregation (all logs in one place, searchable)
• Alerts on PagerDuty for critical issues
• Cost monitoring (track AWS spend by service, by tag)

Results

After 4-week migration:

• 63% cost reduction: $40k/month → $15k/month
• Improved uptime: 99.5% → 99.95%
• Operational efficiency: 2 engineers at 50% time → 1 engineer at 20% time
• Zero downtime during migration (rolling deploys, blue/green DB switchover)
• Better observability: alerts on real issues, not noise
• Self-service infrastructure: team can deploy without waiting

The business impact: Nimbus closed 8 enterprise clients in year 1 (each requiring enterprise-grade SLA). The infrastructure cost savings ($300k/year) directly funded an additional engineer hire.

What we learned

1. Tag everything from day 1. Untagged resources are the silent killer of cloud budgets.
2. Right-size first, optimize second. 80% of cost wins come from removing waste, not from better pricing.
3. Reserved instances are a finance decision, not a tech decision. Once you know your baseline, commit.
4. Observability pays for itself. Datadog costs $500/month, but catches issues that would have cost $50k in downtime.
5. Migration without downtime is a science, not an art. Tested repeatedly, deployed in stages, rolled back instantly when needed.

What’s next

Nimbus Logistics is now expanding to Europe. We’re helping them set up multi-region infrastructure (US + EU) with data residency compliance.

Want similar results?

If your AWS bill is too high and your team is spending too much time on infrastructure, book a free intro call. We’ll do a 1-week cost audit and tell you honestly what we can save.