Zero-downtime cloud migration

We migrate the systems that can't go down.

For UK teams running production they cannot take offline. We move infrastructure to the cloud — planned, rehearsed, zero downtime — then build and operate what runs on it.

0ms·cutover·infra as code

cutover · control plane · eu-west-2live
legacygreen
traffic 12%/88% green
legacy 12%green 88%
LEGACYdrained
CONTAINERISEimages ok
ORCHESTRATE4/4 ready
PRODUCTIONlive
Parallel runnight compare OK
Rollbackready
Trafficweighted · green

RPO

0

RTO

<30s

Health

4/4

Downtime

0ms

[09:41:02] drain legacy traffic ··············· ok

[09:41:07] promote green environment ·········· ok

[09:41:09] health checks 4/4 ·················· ok

[09:41:09] cutover complete — downtime 0ms

Cloud migrationPlatform & DevOps

02Working rules

systems run in parallel before any cutover
nightly compare · rollback ready
0ms
planned downtime target on cutover day
rehearsed · reversible traffic shift
1 wk
from kickoff to a written risk register
assessment before architecture
IaC
infrastructure versioned, reviewed, reproducible
terraform · no snowflake servers

03The playbook

Forty-five minutes of deploy risk, cut to five.

  1. Starting point

    A typical estate: a fifteen-year-old monolith on bare metal. Deployments take 45 minutes, happen out of hours, and fail often enough that every release needs a rollback engineer on standby.

  2. The migration

    We containerise service by service and stand up managed Kubernetes alongside the legacy estate — then run both in parallel. Same traffic, same data, compared nightly.

  3. Cutover day

    Cutover happens on a weekday morning, inside business hours — a cutover you can only attempt at 2am is a cutover you don't control. Traffic shifts gradually, and every step is reversible.

  4. What changes

    Releases go from one a week to twelve a day. The out-of-hours deploy window disappears. Infrastructure cost typically falls by a third.

Fig. 01 — release frequency, before and afterdrawing

BEFORE1 release / week · 45 min · manual rollbackAFTER12 releases / day · 5 min · automatic rollbackMONTUEWEDTHUFRISATSUN

45→5

minutes per deploy

12/day

releases

0ms

cutover downtime

See how we run a migration →

04Method

  1. 00

    Assess

    infrastructure audit · dependency graph · risk register

  2. 01

    Plan

    cutover design · rollback rehearsal · stakeholder comms

  3. 02

    Migrate

    staged cutover · dual-run validation · live monitoring

  4. 03

    Verify

    load & failover tests · security review · sign-off

  5. 04

    Optimise

    cost tuning · runbooks · team handover

Every phase ends with a decision point. You can stop at any of them.

05Capabilities

Practice 01

Cloud migration

We take production systems off ageing infrastructure and onto AWS, GCP, or Azure without stopping them — every cutover planned, rehearsed, and reversible.

  • Audit and risk register in week one
  • Rehearsed, reversible cutover — zero downtime
  • Ninety days of post-migration optimisation
Cloud migration in detail →

Practice 02

Full-stack development

Product work across MERN, Angular, and Laravel — typed, tested code shipped through the same pipelines we run for infrastructure.

  • Working software every two weeks, not a slide deck
  • Typed, tested code with CI/CD from the first commit
  • Built to run on the infrastructure we already operate
Full-stack development in detail →
03

Platform & DevOps

Kubernetes operations, CI/CD, observability, and cost control as an ongoing engagement.

07Tools

Something useful from StackGB

Digital business card

One QR. A live page. Call, WhatsApp, email, website, and save-to-contacts — then change any information without reprinting. Same craft we use for client systems.

Prefer to see one first? Open a live example

  • Live page, not a dead QR

    Print once. Change any information later — scanners always hit the current card.

  • Every channel

    Call, WhatsApp, email, website, and save contact. Visitors pick how they prefer to reach you.

  • Save to contacts

    Real vCard for Android and iPhone — numbers, WhatsApp link, email, and website included.

  • Branded QR designs

    Colour presets, module shapes, centre brand mark, print caption, and PNG download.

  • Up to 3 cards

    Personal, events, and a side brand without a subscription wall on day one.

  • Sign in your way

    Username and password to manage your cards.

Also from StackGB

Quick POS

Take a payment on the spot — a custom amount or a product from your own list. No card reader to set up, and every charge is logged for you to look back on.

  • Custom amount, fast

    A numeric keypad and an optional note — type an amount and you are done.

  • Your own product list

    Add products with a name and price once, then search and tap to add them to a cart.

  • Works on any phone

    No dedicated hardware. Open SuperDash on your phone or tablet and start taking payments.

  • Charges, logged

    Every charge is recorded so you can pull up sales reports whenever you need them.

  • Enabled by StackGB

    Quick POS is switched on for your account by us — reach out and we will turn it on.

  • Sign in your way

    Username and password to get in and start charging.

07Contact

Tell us what you're running.

One conversation. An honest read on what your migration or product build would take: the effort, the risk, and what it would save. Assessment takes a few minutes. No deck. No pressure.

Dartford, United Kingdom