Cloud & Infrastructure Consulting in New Zealand

The monthly number is the real quote

Infrastructure is the one category where the price on the proposal is the smaller figure. A $30,000 migration that leaves you paying $4,000 a month is a $174,000 decision over three years; a $45,000 migration that leaves you on $900 a month is cheaper by a wide margin.

Ask every supplier for the projected monthly bill after the work is done, based on your actual usage rather than a tier chart. Ask what drives it up — more users, more data, more environments — and what it looks like at twice your current load. A supplier who cannot produce that number has not sized anything.

Choose the smallest platform that does the job

AWS and Azure are excellent and there are good reasons to pick them: breadth of services, an existing organisational standard, a compliance requirement, or a scale that genuinely needs them.

There are also bad reasons, and the most common is that they sound more serious. For a single business application, a simpler managed platform is cheaper to run and much cheaper to operate, because it needs less specialist attention. The operating cost of infrastructure is mostly human.

Ask why the proposed platform was chosen. "It is what we use" is honest and sometimes fine. "It is what your workload needs, because…" is better.

Backups you have never restored are not backups

Almost every business has backups. A striking number have backups that stopped running months ago, or that would take three days to restore, or that cover the database but not the files it references.

You find out on the worst possible day. Ask any supplier to test a restore as part of the engagement, and to tell you how long it took and what was missing. That single exercise is worth more than a page of architecture diagrams, and it is the question most likely to change how you feel about your current setup.

Make deployment boring

If shipping a change requires a specific person, a written checklist and a quiet Saturday, the cost is not the Saturday. It is that changes get batched up, batches get risky, and eventually nobody wants to touch anything.

Automated builds, tests and deployments turn releases into a routine event. The investment is real — expect it to be a meaningful slice of any infrastructure project — and it pays back in every change after it, including the urgent security patch you will need one day at short notice.

Plan for the bad day before it arrives

Ask what happens if the main region has an outage, or if someone deletes the production database at 4pm.

You are not necessarily buying protection against every scenario — full redundancy costs real money and most New Zealand businesses do not need it. You are buying a documented, agreed answer: how much data you could lose, how long you would be down, and who does what. Deciding that consciously is the difference between an incident and a crisis.

Know where the lock-in is

Every platform ties you in somewhere, and some of that is a fair trade for convenience. The problem is not lock-in; it is unexamined lock-in.

Ask which parts of the setup are portable, which are specific to the provider, and roughly what moving would cost in a few years. A supplier who says none of it matters has never moved a live system.

Security basics, before anything clever

Most incidents that affect New Zealand businesses do not involve sophisticated attackers. They involve a credential that was reused, an admin account without two-factor authentication, a database exposed to the internet by a temporary rule that was never removed, or software three years behind on patches.

Ask what the supplier's baseline is: who has access, how access is granted and revoked, how secrets are stored, and how patching happens. It is a short conversation, and the answers are revealing.

What to compare across quotes

• Monthly cost after the project, at real usage.
• A tested restore. Included, with a measured time.
• Deployment. How a change reaches production, and who can do it.
• The outage plan. Data loss and downtime, quantified.
• Handover. Documentation and access so you are not dependent on one supplier.
• Lock-in, named honestly.

Describe the system and the fear

You do not need to name a platform. Describe what you run, what keeps you up at night — the bill, the single server, the person who is the only one who knows — and what has already gone wrong. Suppliers who do this work weekly will tell you which risk is worth spending on first, and the good ones will tell you when the answer is cheaper than you feared.