Cloud & Deployment3 min read
Serverless or servers: cold starts, cost curves and connection limits
Serverless is excellent for spiky, stateless, short work. The trouble starts when your workload is none of those things.
Cloud spend is usually dominated by a handful of items: over-provisioned always-on compute, data transfer out, oversized managed databases, log and metric retention, and forgotten resources. Start by attributing cost with tags, then rightsize, delete the unused, move storage to appropriate tiers, and commit to reserved capacity only for genuinely steady workloads.
| Category | Typical share | Common waste |
|---|---|---|
| Compute | 40–60% | Over-provisioned, idle at night, dev environments always on |
| Data transfer | 10–25% | Cross-region chatter, un-cached assets, cross-AZ traffic |
| Managed databases | 10–25% | Sized for a peak that happens twice a year |
| Storage | 5–15% | Old snapshots, no lifecycle rules, wrong tier |
| Observability | 5–15% | Debug logs retained for a year, high-cardinality metrics |
Before optimising anything, tag resources by environment, service and owner, and enable cost allocation. Optimising an unattributed bill is guesswork.
Providers charge little to receive data and meaningfully more to send it out. Bills grow from patterns nobody costed: images served from the origin instead of a CDN, a service in one region querying a database in another, backups replicated across regions, or chatty microservices crossing availability zones.
No. Engineering time is usually more expensive than infrastructure. A managed service costing more but saving days of work per month is often the cheaper decision overall.
For interruptible workloads — batch processing, CI runners, stateless workers with graceful shutdown — yes, the savings are large. Not for anything that must not be interrupted.
In our audits, 20–40% without changing architecture, mostly from rightsizing, non-production schedules, storage lifecycle rules and log retention.
The practice of making cloud cost a shared engineering concern with visibility, accountability and regular review — rather than a bill finance receives after the fact.
ROVQIX Engineering
Engineering team, ROVQIX
The ROVQIX engineering team builds and maintains web platforms, APIs and infrastructure for clients across SaaS, ecommerce and enterprise. These notes come out of real production work — deploys, incidents, migrations and audits.
ROVQIXdesigns and builds production web platforms — Next.js front ends, Node.js APIs and the infrastructure behind them. Tell us what you're building and we'll scope it with you.
Serverless is excellent for spiky, stateless, short work. The trouble starts when your workload is none of those things.
Putting a CDN in front of a site does nothing by itself. The cache headers you send decide whether it helps or just adds a hop.
Every host looks fine on day one. The differences show up when traffic triples, a region goes down, or you need to leave.
No spam. Just the occasional case study and craft breakdown.