Cloud Architecture Practice
- May 29, 2026
- 9 min read
Why Most GCP Programs Stall Before Anyone Opens Kubernetes
Most cloud migrations fail at the foundation—not the technology. Four questions Indonesian enterprises should answer before production, and how PT CPI helps regulated teams get it right.
Executive summary: In architecture reviews across banking, FinTech, and large enterprises in Indonesia, we rarely hear “our GKE cluster is the problem.” We hear a quieter, more expensive story: migrations that slip quarter after quarter…
In architecture reviews across banking, FinTech, and large enterprises in Indonesia, we rarely hear “our GKE cluster is the problem.” We hear a quieter, more expensive story: migrations that slip quarter after quarter, audits that freeze releases, and cloud bills that leadership cannot map to products. The technology stack is often capable. The foundation usually is not.
Who should read this
| Role | Why read this |
|---|---|
| CEO / COO | Cloud investment decisions and evidence of delivery discipline |
| CTO / Head of Engineering | Architecture patterns, pipelines, and adoptable quality gates |
| Engineering & Platform | Technical detail, trade-offs, and operational practice |
The symptom leaders describe
Google Cloud is not a single switch you flip. It is an operating model—who may create resources, how environments are separated, how logs prove who changed what, how backups restore within agreed RTO/RPO, and how spend is attributed when FinTech platforms scale on market hours. When those agreements are vague, every squad optimizes locally. Engineering optimizes for speed. Security optimizes for control. Finance optimizes for forecasts. None of them are wrong; they are simply not aligned.
Cloud as an operating model
We call the aligned starting point a landing zone. It is not a slide deck template downloaded from the internet. It is a set of decisions recorded where product, platform, security, and compliance can point and say: this is how we run GCP here. PT Cloud Platform Indonesia (PT CPI) builds landing zones as a mitra Google Cloud for organizations that cannot treat cloud as an experiment—they need evidence, predictability, and room to innovate.
What a landing zone really is
Before anyone provisions a cluster, four questions predict whether the program will accelerate or stall. First: which workloads move in the first wave, and which dependencies must move with them? Second: which data classes require residency, segregation, or additional logging—and who approves exceptions? Third: who owns day-2 operations after cutover—platform SRE, application teams, or a partner—and what does escalation look like at 2 a.m.? Fourth: what audit evidence must exist before production traffic is allowed—scan results, approval records, architecture diagrams, recovery tests?
Four questions before any cluster
When these questions lack crisp answers, teams still ship—but they ship into fog. A payment service lands in a project with permissive IAM because a deadline loomed. A trading feed shares a VPC with internal tools because networking was “temporary.” A staging environment holds production-like data because segregation was deferred. Each shortcut is rational in the moment. Together they become the audit finding that stops the next release.
Shipping into fog
The opposite pattern is slower at the start and faster for years afterward. Folder hierarchy mirrors how the business thinks about risk. IAM roles follow least privilege with break-glass documented. Shared VPC or hub-spoke networking is chosen for a reason, not inherited by accident. Centralized logging and retention policies mean security teams review posture before go-live, not after an incident. Budgets and labels connect spend to products so FinOps conversations use the same facts as engineering stand-ups.
The slower start, faster years pattern
Security and compliance teams in regulated industries are not obstacles to cloud—they are stakeholders in a shared definition of evidence. PT CPI integrates controls early: Cloud Logging exports, configuration baselines, pipeline gates with DevSecOps tooling (Snyk, GitLab, GitHub), and cloud security posture with partners such as Wiz where CNAPP visibility is required. The goal is not more tools; it is one narrative that auditors and institutional counterparties can follow.
Security as a stakeholder
FinTech and institutional programs add another layer. Partner onboarding, vendor due diligence, and exchange or custody integrations often require architecture packs—not slides assembled the week before a deadline. A mature landing zone produces those artifacts continuously: data-flow diagrams, environment boundaries, recovery assumptions, and change traceability from commit to production. That is why we plan modernization—GKE, Cloud Run, data platforms, low-latency pipelines—as extensions of the foundation, not parallel projects that ignore it.
FinTech and institutional layers
Leaders sometimes ask whether they should pause product work to “fix the foundation.” Our experience suggests a phased answer: freeze net-new production patterns that violate agreed guardrails, run a focused landing zone sprint in parallel with a bounded pilot workload, and measure success by audit readiness and deployment frequency—not by how many services were clicked in the console. A four-to-eight-week foundation sprint with executive sponsorship often unlocks more delivery in the following year than twelve months of heroic engineering without guardrails.
Fix the foundation without stopping product
PT CPI’s role is to make those sprints concrete. We facilitate workshops that surface the four questions, translate answers into architecture decision records, implement bootstrap automation, and hand over runbooks so internal teams own the platform after go-live. We stay when needed for FinOps reviews, DevSecOps enablement, and FinTech platform engineering—but the foundation should belong to your organization, not to a permanent external dependency.
How PT CPI makes sprints concrete
If your GCP program feels stuck—if Kubernetes is “next quarter” for the third quarter in a row—look backward to the foundation. Agree the four questions in writing. Then the technology choices become easier, audits become shorter, and leadership sees cloud as an investment with a ledger, not a line item that grows mysteriously.
Conclusion and next steps
Ready for a structured conversation? PT CPI offers architecture reviews and GCP Cloud services from assessment through managed operations. Visit ptcpi.cloud or contact hello@ptcpi.cloud to schedule a session with our team.
