Cloud management after the migration: where enterprise costs and risks build up
Most enterprises have already moved a large share of their workloads to AWS, Azure or Google Cloud. The migration project ended, the team moved on, and the environment kept growing. A year later the monthly bill is higher than planned, no one is sure which accounts are still in use, and a configuration change in one team’s subscription has opened a gap that security finds in an audit.
Most of this comes from how the environment is managed after go-live, and it can be fixed.
Where it goes wrong
The same four issues come up again and again.
Nobody owns the bill. Resources are created by many teams, but the invoice lands with finance or central IT. Idle test environments, oversized databases and storage nobody reads keep running because no one is accountable for them.
Each team builds its own way. Without an agreed landing zone, every account has slightly different network rules, logging and access settings. That makes security reviews slow and makes it hard to move people between teams.
Problems are found by users. Monitoring was set up for the migration and never extended. New services go live without alerts, so the first sign of trouble is a ticket from the business.
The skills sit with a few people. Two or three engineers know how the environment really works. When they are on leave or leave the company, changes stop or go wrong.
Five things that bring it back under control
1. Give every workload an owner and a cost centre
Tag every resource with an application, an owner and a cost centre, and refuse to deploy anything without those tags. Then send each owner a monthly report of what their workloads cost. Owners who can see their own costs usually act on them.
2. Agree one standard build
Define a landing zone: how accounts or subscriptions are created, the network layout, identity and access rules, logging and encryption settings. Build it as code so new environments start compliant. Existing accounts are brought into line one at a time, starting with the ones that hold customer or financial data.
3. Make security part of daily work
The basics catch most problems:
- single sign-on and least-privilege access, with no shared admin accounts
- encryption at rest and in transit by default
- continuous checks for misconfigurations such as public storage or open ports
- a named person who reviews the findings every week
Findings only reduce risk when someone works through them, so the weekly review matters more than the choice of tool.
4. Monitor what users experience
Set alerts on the things the business notices: response times, failed transactions, queue backlogs and error rates, alongside CPU and memory. Every production service should have a dashboard and an on-call owner before it goes live.
5. Review cost and capacity every month
Treat cost control as a monthly routine. The review should cover:
- the biggest changes in spend and why
- idle or oversized resources to remove or resize
- commitments such as reserved instances or savings plans for steady workloads
- environments that can shut down outside working hours
Run it yourself or use a managed service?
If you have a cloud team with time for governance, cost reviews and on-call cover, run it yourself and use the five points above as a checklist. If the same few engineers are doing all of that on top of project work, a managed service can take on monitoring, patching, cost reviews and incident response, while your team keeps the decisions on architecture and spend.
Either way, the work has to be visible. You should get a monthly report on cost, incidents and open security findings, and you should be able to see who is responsible for each one.
What to do next
Start with one question: for each of your ten most expensive cloud services, can you name the person who owns it? If not, that is the first fix.
If you want help putting these controls in place on AWS, Azure or Google Cloud, see our cloud modernization service.
