Multi-cloud migration, database optimisation delivering 59,000x query speedups, and £240K per year in savings. This is a capability we deploy as part of our operator engagements — not a standalone service.
A major UK e-commerce retailer processing 2 million daily requests across three Aurora MySQL databases totalling 465GB. Their entire infrastructure ran on AWS at premium pricing. Nobody had questioned it in years.
Two problems compounding each other: premium cloud pricing on workloads that did not need it, and database performance issues hiding behind generous hardware provisioning. The databases had accumulated dead indexes, audit logs, and migration artefacts that slowed queries and inflated storage bills. A production database that was performant at 50GB behaves very differently at 465GB.
Cloud cost and database performance are separate problems that compound each other. We attacked both simultaneously.
We mapped every service, its actual utilisation, and whether it genuinely needed premium-tier hosting. Found 30-60% of workloads could run on cost-effective alternatives without any performance trade-off.
Identified which workloads move (non-critical, stateless, batch processing) and which stay (latency-sensitive, compliance-bound, deeply integrated). The goal is cost reduction, not provider ideology.
Change Data Capture pipelines ensured data consistency across environments during and after migration. Zero downtime, zero data loss. The business never noticed the transition.
Independently audited every database for missing indexes, dead tables, query anti-patterns, and storage waste. This delivered bigger wins than the cloud migration itself.
The tech stack included Aurora MySQL databases, Varnish caching layers, Cloudflare WAF/CDN, Algolia search, and 20+ integrated SaaS platforms. We moved non-critical workloads — batch processing, staging environments, internal tools, data warehousing, monitoring — to dedicated servers on Hetzner, where equivalent compute costs 50-80% less than AWS EC2.
Mission-critical services stayed on AWS where the managed ecosystem genuinely added value. CDC pipelines maintained data consistency across environments throughout.
The economics are straightforward: a Hetzner dedicated server with 128GB RAM and 2TB NVMe costs roughly £80/month. The equivalent EC2 instance is £500+/month. Multiply across dozens of workloads and the savings are transformational.
Result: hosting costs dropped from £38K/month to £15K/month within three months — a 60% reduction, saving £276K per year. Zero downtime. The site continued processing 2 million daily requests at 780ms average response time with a 0% error rate throughout.
We audited 465GB across five production databases and found critical issues hiding in plain sight.
| Fix | Before | After | Improvement |
|---|---|---|---|
| Sales order email query | 35.6 seconds | 0.0006 seconds | 59,000x faster |
| Return item lookup | 0.24 seconds | 0.0002 seconds | 1,200x faster |
| Credit memo invoice query | 0.87 seconds | 0.0001 seconds | 8,700x faster |
| Recoverable storage | 465 GB total | 257 GB reclaimable | 55% recoverable |
| CPU time saved | -- | 12.2 days per cycle | Massive efficiency gain |
Six missing index additions eliminated 762 billion full table scans. A 379-million-row search index table, no longer used but still accumulating via active database triggers, was identified and truncated. Email notification cron jobs scanning 3.7 million rows every execution were optimised to sub-millisecond.
Cache hit rates improved from 44% toward the 70-90% target range by fixing broken catalogue indexers that were forcing constant cache invalidation. Every one of these issues was invisible at the application level — the site still worked, it was just working harder than it needed to.
Our DevOps agent now monitors database health continuously, flagging performance regressions before they become production issues.
Two optimisation tracks delivering compound savings. The cloud migration cut the hosting bill in half. The database audit made the remaining infrastructure dramatically more efficient.
Cloud cost reduction and database optimisation were two capabilities we deployed within a broader engagement that created £6.4M in annualised value. The infrastructure savings freed budget for reinvestment into revenue-generating work — AI pricing, customer data platforms, and marketing automation.
This is the pattern we see in every engagement: infrastructure waste is not just a cost problem, it is an opportunity cost problem. Every pound spent on over-provisioned cloud is a pound not spent on margin improvement. We find the waste, fix it, and redirect the savings.
Less risky than staying on a single provider. We manage migration risk through CDC pipelines that maintain data consistency, blue-green deployment patterns, and gradual traffic shifting.
The bigger risk is vendor lock-in — being dependent on a single provider for pricing, availability, and feature roadmap. When your only option is to pay whatever AWS charges, that is a strategic vulnerability, not stability.
Latency-sensitive production databases, services with deep ecosystem integration, compliance-bound workloads requiring specific certifications, and anything where the managed service genuinely reduces operational burden more than the cost premium.
The key word is "genuinely." Most businesses assume their workloads need premium hosting because they have never tested the alternative. We audit first, then recommend.
Depends on complexity. Non-critical workloads can move in 4-6 weeks. Full multi-cloud migration with CDC pipelines typically takes 3-6 months, phased by workload criticality.
Database optimisation is faster — most audit-and-fix cycles complete in 2-4 weeks. We run both tracks in parallel so both deliver value simultaneously.
Yes. We are operators. After migration, our DevOps agent monitors infrastructure health continuously, and our team handles ongoing optimisation. We do not hand you a migration plan and walk away — we run the infrastructure as part of your engagement.
Yes. The methodology applies to any cloud provider — Azure, GCP, or existing on-premise infrastructure. The principle is the same: audit utilisation, identify what is over-provisioned or over-priced, and move workloads to the most cost-effective option.
We have seen the same 30-50% savings potential across every major provider. The specific services differ, but the pattern of over-provisioning and defaulting to premium tiers is universal.
Let us look at your infrastructure. We will tell you what can move, what should stay, and how much you will save. Then we will do it.