For the estate after it is already live
Cloud Operations
A cloud estate does not stay reliable on its own once the migration project ends and the pipeline is shipping steadily. This practice takes ongoing ownership of monitoring, incident response, patching and capacity, so reliability is a maintained condition rather than a memory of launch day.
How day-two operations actually runs
Monitoring, on-call structure and capacity planning, walked through as one operating system.
90 secSample frame · video pending01Before adding more alerts
Detection gap or response gap: which is it?
Adding alerts to a broken response process just produces louder noise. Four questions settle which side of the operations problem is real.
Do incidents get noticed before a customer complains?
If customers report problems first, the gap is detection: instrumentation is not covering a real failure mode yet.
When an alert fires, does anyone act on it quickly?
If alerts exist but sit unanswered, the gap is response structure and on-call ownership, not more monitoring.
Does capacity get planned ahead, or added after strain?
Reactive scaling after a slowdown points at a capacity planning gap, separate from both detection and response.
Are patches applied on a schedule or after a warning?
Alarm-driven patching signals a missing cadence, which is a planning fix rather than a monitoring or response one.
| Detection gap | Response gap | |
|---|---|---|
| Symptom | Customers report it first | Alerts fire and go unanswered |
| Fix | Better instrumentation | On-call structure and ownership |
| Effort | Moderate, tooling-focused | Process and staffing focused |
| Best fit | Blind spots in coverage | Alerts nobody is accountable for |
02Agreed before monitoring goes live
The ledger that keeps an operations handoff honest
Operations engagements drift when nobody can say who is accountable for an alert at three in the morning. This sets the terms plainly: what you bring, what gets built, what comes back, and which approvals never move without you.
- Your side
You bring
- Access to the current production environment
- Any recent incident history worth reviewing
- Constraints on how monitoring can be added
- Our side
We build
- Monitoring and alerting built against real failure modes
- An on-call rotation and escalation path
- A reliability baseline your team can measure against
- Your side
You receive
- A monitored estate with alerts tied to real impact
- A documented on-call process ready to run
- A reliability report your leadership can read
Never leaves you
- The production environment and every workload in it
- Final say on alert thresholds and escalation policy
- Ownership of the reliability programme after handover
Where the line sitsWe build the monitoring and the on-call structure. The environment and every decision it produces stay yours throughout and after.
03From baseline to a reliable operating rhythm
Four stages, each closing on an artefact
An estate that stays unreliable usually skipped one of these stages. This system closes each one with something your team can verify before the next begins.
Baseline the current reliability
Current incident history, alert coverage and response times get measured honestly before any change gets proposed.
A reliability baseline measured honestlyBuild monitoring against failure modes
Instrumentation targets the failure modes that actually hurt customers, not every metric a dashboard could technically show.
Alerts tied to real customer impactStand up the on-call structure
Escalation paths and response runbooks get written and rehearsed, so an incident at any hour has a clear first responder.
A response process that survives pressureHand over the operating rhythm
Patching cadence, capacity review and the on-call rotation move to your team, with the reliability report they can keep reading.
A rhythm your team keeps running
04What actually drives the number
Four drivers, priced from the baseline
Operations work is priced from the reliability baseline, not a flat retainer, because these four factors change scope by multiples of the base case.
Size and complexity of the estate under coverage
A handful of services monitors far more cheaply than a large estate spanning several teams and environments.
Current gap between detection and real impact
An estate with almost no instrumentation costs more to bring to a reliable baseline than one with partial coverage already.
Response coverage required
Business-hours-only response costs less to staff than a rotation covering every hour across multiple time zones.
Growth rate driving capacity planning
A slowly growing estate needs lighter capacity review than one scaling fast enough to strain limits every quarter.
05Evidence and proof
Estates that stayed reliable after launch day faded.
Operations proof is a measured drop in customer-reported incidents, not a dashboard screenshot. Every record here waits for written client approval before production.
Explore case studies
Incidents stopped reaching customers before the team knew.
Koru Health's on-call team learned about problems from support tickets more often than from an internal alert. New instrumentation closed that gap.
- Customer-reported incidents
- Sharply reduced
- Detection lead time
- Improved
Patching moved from alarm-driven to scheduled.
Nexa Foundry patched systems only after a vendor warning, with no planned cadence. A scheduled patch and capacity process replaced the reactive pattern.
- Patch cadence
- Fully scheduled
- Capacity incidents
- None reported
We finally know about a problem before a customer has to tell us.
Sample · pending client approval
06Read this first
When an operations engagement is the wrong spend
Commissioning ongoing operations for an estate that has not migrated yet, or a pipeline that is not shipping reliably, spends the budget too early. The exits are marked before the pitch.
- Cloud Migration
The estate has not migrated yet
Day-two operations only makes sense once workloads are actually running in the cloud. The move itself is a separate practice.
- Cloud DevOps
The real gap is in the release pipeline
If releases themselves are slow or risky, pipeline work closes that gap faster than adding monitoring around it.
- Hire Cloud DevOps Engineers
Your team should carry on-call itself
If your engineers already run a solid on-call rotation and just need capacity, hiring in fits better than a delivery layer.
| Shape | Best when | You direct | We carry |
|---|---|---|---|
| Scoped baseline | You need a reliability picture first | Which systems get baselined first | Measurement and the written report |
| Ongoing retainer | The estate needs continuous coverage | Priorities and escalation policy | Monitoring, on-call and patching |
| Team extension | Your leads already run operations well | Standards and architecture | Named engineers and continuity |
07Straight answers
Asked before an on-call handoff.
01Do you take over our existing on-call rotation?
We help design and staff the structure alongside your team rather than replacing it outright. Ownership of the rotation moves back to you once it is proven.
02What monitoring tools do you use?
Whichever fit your existing stack and budget, chosen against the baseline rather than a default. Coverage against real failure modes matters more than the specific tool.
03Will you be the ones talking to our customers during an incident?
No. Detection and response structure are ours to build; customer communication during an incident stays with your team throughout.
04How is this different from cloud DevOps?
DevOps covers the delivery pipeline that ships code. Operations covers running the live estate afterward: monitoring, incident response, patching and capacity.
05Can you help with capacity planning before a big growth event?
Yes, capacity planning against real growth data is part of the practice, timed ahead of an expected surge rather than reacting after strain appears.
06How quickly can monitoring coverage improve?
A baseline and initial instrumentation can close within a few weeks. Full on-call structure and patch cadence typically take longer to prove reliable.
07What does an ongoing operations engagement cost?
Pricing follows estate size, current coverage gaps and required response hours, set during the baseline review rather than a flat retainer figure.
08Do you provide the actual on-call engineers?
Yes, where a retainer requires it, alongside the process and instrumentation. The team can also be sized to extend yours rather than replace it.