ABK Consultancy · Platform engineering for regulated enterprises
Arshad Badar Khan
.NET and Kubernetes on Azure and OpenShift — legacy modernisation, zero-trust service mesh and platform engineering for enterprises regulated under DORA, NIS2 and PCI DSS. 18+ years, previously at the IAEA in Vienna and CheckMyBus in Nuremberg.
LinkedIn profile ↗ EmailKey skills & topics
46 topics across the series. Pick one or more to filter the articles.
Platform
Kubernetes 12AKS 2Azure 5Docker & containers 1Kubernetes operators 5GitOps 6Multi-cluster & residency 5Disaster recovery 1Upgrades & patching 1FinOps & cost 1
Service mesh
Istio 18Service mesh 6Ambient mesh 10mTLS 19SPIFFE 12AuthorizationPolicy 17Network policy 4Egress control 3
Identity
Microsoft Entra ID 22OAuth 2.0 & OIDC 4JWT validation 5Workload identity 8Passwordless 4Conditional Access 1Scopes & app roles 4On-behalf-of flow 1Secrets & certificates 4
Observability
OpenTelemetry 2LGTM stack 10Distributed tracing 5Audit logging 5Compliance evidence 26Drift & posture 7Incident reporting 1
Code
The Mesh Series
45 articles on Kubernetes, Istio, .NET and Entra ID — built for, and defended in, an audit.
-
What Regulated Enterprises Can't Ignore
You're moving your web apps to Kubernetes. The kind of apps that ran happily on Windows servers for years. Back there, authentication was almost free. They sat inside the domain, Windows-integrated auth did the quiet work, LDAP…
-
Kubernetes in Five Boxes
You've decided to move to Kubernetes. Good. Then you open the documentation and meet forty resource types, three networking layers, and a YAML file that wants to know your pod's DNS policy. This is the moment most people quietly…
-
Your First .NET 8 Service on Kubernetes
You have the mental model now. Article 2 gave you a .NET 8 minimal API and the Deployment plus Service YAML that fronts it. Both sit on your disk, doing nothing. A YAML file is not a running service. A cluster does not care about…
-
What a Service Mesh Actually Does (and What It Doesn't)
A service mesh is a layer of proxies. That's the whole idea, stripped of the conference-stage mystique. Next to every pod, you run a second container — an Envoy proxy, the "sidecar" — and you route all of that pod's traffic…
-
Istio Without the Overwhelm
Istio has a reputation problem. You open the docs and drown: control planes, sidecars, gateways, virtual services, destination rules, telemetry pipelines, a CRD zoo that reads like someone was paid per resource kind. Most…
-
Identity 101 — OAuth2, OIDC, and Tokens
Ask five engineers on your team which OAuth flow their service uses. You will get three shrugs, one confident wrong answer, and one person who actually knows. That last person is the one who has read a token expiry error at 2am.…
OAuth 2.0 & OIDCJWT validation.NET & C#Microsoft Entra IDScopes & app roles
-
Microsoft Entra ID in plain language
You moved your web apps to Kubernetes and the authentication story fell apart on the way in.
Microsoft Entra ID.NET & C#Scopes & app rolesAzureKubernetes
-
Workload identity — how machines prove who they are
A user proves who they are with something they know, something they have, or something they are — a password, a passkey, a fingerprint. A service has none of those. It has no fingers, no phone, no memory between restarts. Yet the…
-
Going passwordless — for users and workloads
"Just add an OTP" is now the compliance answer that fails the audit.
-
The integration thesis — Entra for users, Istio for workloads
The most common identity mistake in a Kubernetes estate isn't a weak policy. It's a merged one.
-
Terminating user identity at the mesh edge
A bug that ships to production quietly, in regulated shops, all the time:
Microsoft Entra IDJWT validationOAuth 2.0 & OIDCAuthorizationPolicy.NET & C#
-
Validating Entra JWTs in C#
An unverified token is just a string. A HALF-verified one is worse — it looks trusted, gets waved through, and carries an authority it never earned.
-
Defense in depth — validate at the mesh and in the code
If the gateway already validated the token, why validate it again in the service? Because no single control is the control — and the question that decides your architecture is which direction your mistakes fail.
-
mTLS and SPIFFE — identity on the wire
Everyone claims they have mTLS. Almost nobody can prove it to an auditor. Those are two different jobs.
-
Entra Workload ID Federation on AKS
A workload can present a SPIFFE cert to its neighbours. It can't present one to Key Vault. So how does it prove itself to Azure — still with no secret in config?
Microsoft Entra IDWorkload identityAzureSPIFFESecrets & certificates
-
Default-deny — the AuthorizationPolicy that fails closed
In default-allow, a forgotten rule leaves a workload open — and you find out when infosec does, or an attacker does. In default-deny, a forgotten rule breaks a deployment in testing, in your face, where you fix it.
AuthorizationPolicyNetwork policyIstioMicrosoft Entra IDSPIFFE
-
Identity propagation across hops — keeping the user alive
Your service needs to call the next service on a user's behalf. The instinct — forward the user's token — is the confused-deputy bug.
On-behalf-of flowMicrosoft Entra IDScopes & app rolesWorkload identitymTLS
-
Scopes, roles, and app roles — from claim to decision
A valid token is not a permission. Authentication proves who you are; it doesn't say you're allowed to do THIS. Two Entra claims decide that — and confusing them is how a valid token does the forbidden thing.
-
Secrets and certificates — for the ones you couldn't eliminate
Eighteen articles deleting secrets. But "fewer" isn't "none." Some secrets must exist — a third-party API key, a legacy DB password, a signing key you own. For those, and for certificates, there's an order of preference most teams…
Secrets & certificatesKubernetesWorkload identityGitOpsNetwork policy
-
Network policy and mesh policy — belt and suspenders
Your mesh AuthorizationPolicy has a blind spot: a pod without a sidecar. A DaemonSet, a node agent, a debug pod — it can open a raw socket to your ledger and the mesh never gets a vote, because the request never hit a proxy.
Network policyAuthorizationPolicyPolicy as codeSPIFFECompliance evidence
-
The policy you wrote isn’t the policy you enforce
You wrote a default-deny policy. You reviewed it. You merged it. And nothing in your cluster stops someone from deploying a workload that ignores it.
Policy as codeKubernetesAuthorizationPolicyNetwork policyCompliance evidence
-
Egress control and the locked-down cluster
You locked the front door of your Kubernetes cluster. Ingress gateway, token validation, default-deny — the works.
-
Threat-modelling a meshed .NET API
You've spent a dozen articles building controls. Now ask the question nobody budgets for: what does each one do when it BREAKS?
Threat modellingAuthorizationPolicyCompliance evidenceJWT validationCERT-In & DPDP
-
If you can’t prove it, it didn’t happen — audit logging in the mesh
Your service mesh is already logging every request. Method, path, status code — for free.
-
The LGTM stack for compliance evidence
Loki, Grafana, Tempo, Mimir. Four letters everyone treats as one observability blob. To a regulator they answer four different questions — and putting the wrong one in the wrong store is how evidence quietly disappears.
LGTM stackCompliance evidenceDistributed tracingMulti-cluster & residencyOpenTelemetry
-
Tracing identity through a request
You turned on distributed tracing. Grafana draws the pretty waterfall. You feel covered.
Distributed tracingLGTM stackCompliance evidenceIstioOpenTelemetry
-
Turning telemetry into signed evidence
You have the logs. That is not the same as having evidence.
Compliance evidenceSupply-chain signingLGTM stackDrift & postureMulti-cluster & residency
-
Continuous posture and drift — proving the control is still holding
Your admission controller said yes to the policy six months ago. Is the policy still there?
Drift & posturePolicy as codeCompliance evidenceGitOpsLGTM stack
-
Mapping controls to frameworks — PCI-DSS 4.0 on the mesh
"Our service mesh makes us PCI compliant." If a vendor says that, end the meeting.
PCI DSSAuthorizationPolicyCompliance evidenceDrift & posturemTLS
-
Regional frameworks — one control surface, four rulebooks
If you run card workloads across Mumbai, Riyadh, and Dubai, you answer to RBI, SAMA, the UAE regulators, and ISO 27001 at once. The instinct is four compliance programs. That's expensive and wrong.
SAMACompliance evidenceCERT-In & DPDPRBIMulti-cluster & residency
-
GitOps for the mesh — when the pull request becomes the change record
In a regulated shop, a merged pull request IS the change-management record. Author, reviewer, diff, timestamp — maker-checker, enforced by branch protection, mapped straight onto PCI 6.5 and ISO A.8.32.
GitOpsMicrosoft Entra IDCompliance evidenceAuthorizationPolicyDrift & posture
-
Sidecar to ambient — what gets cheaper, what just moves house
"Sidecarless mesh, zero downtime" is a half-truth — and the missing half is exactly the one a regulated cluster cares about.
Ambient meshService meshMicrosoft Entra IDmTLSAuthorizationPolicy
-
The reference architecture — thirty-two pieces, one system
33 articles ago this series opened on a distributed monolith held together with auth tokens and prayer. This is where the pieces become one picture.
Ambient meshCompliance evidenceMicrosoft Entra IDAuthorizationPolicyPasswordless
-
DORA for platform engineers — what the Act actually asks of your cluster
You have spent thirty‑three articles building one system: a regulated .NET platform on AKS, Istio and Entra, where every request is authenticated, authorised, encrypted and logged, and where the logs are signed and the posture is…
-
NIS2 and the Austrian NISG 2026 — when the platform is the compliance measure and the board is on the hook
Last time you mapped DORA onto the cluster and drew one box with a dashed border: governance, owned by the management body, living in GRC rather than YAML. You treated that box as somebody else's problem.
-
DORA third-party risk and the Register of Information — observed, not declared
In Article 22 you switched the mesh's outbound traffic policy to REGISTRY_ONLY and wrote a ServiceEntry for every external host a workload is allowed to reach. You called it egress control. It was also something else: the only…
Egress controlThird-party riskDORARed teaming & TLPTCompliance evidence
-
Threat-led penetration testing under DORA — what the red team does once the mesh says yes
The letter from the supervisor is two paragraphs long. Your institution has been identified for threat-led penetration testing. A white team is to be formed. Scope will cover critical or important functions. Testing will run…
-
One detection pipeline, four confession clocks — incident reporting across DORA, NIS2, CERT-In and DPDP
In Article 37 a red team walked through an allow rule that was too wide, and the mesh logged every step of it. The debrief is going well. Then someone from legal asks the question that ends the good mood: when did we become aware?
-
Multi-cluster mesh for data residency — residency is a topology decision, not a config flag
Article 38 ended on a problem that sounds administrative and is not. Your Indian arm's CERT‑In report is due in six hours. The evidence that proves it — the access‑log lines, the signed aware‑at record, the Loki result set — must…
Multi-cluster & residencyCompliance evidenceSPIFFEAmbient meshIstio
-
Disaster recovery for a regulated mesh — recovering the proof, not just the app
A regional outage takes your primary cluster offline at 03:10. By 03:40 the platform team has a cluster in the paired region, Argo CD has reconciled the application manifests, and the dashboards are green. At 09:00 the ICT risk…
Compliance evidenceIstioDisaster recoveryGitOpsWorkload identity
-
Upgrading Istio without a control‑failure event — the traffic shift that is not a mutation
Istio 1.31.1 fixes GHSA‑qm8v‑g4f9‑qhjx: a BackendTLSPolicy that quietly reverts to plaintext when its CA reference fails to resolve. The control that encrypts was, under one specific failure, not encrypting. You read the advisory…
-
What the mesh actually costs — two curves on two axes, and the one the CFO sees
The platform lead has one slide. It says the sidecar costs about 100 millicores and 128 mebibytes per pod, multiply by your pod count, and ambient makes most of that go away. Finance nods. The slide is correct. It is also the…
Ambient meshService meshFinOps & costCompliance evidenceMulti-cluster & residency
-
Designing the ComplianceProfile operator — compliance is a reconcile loop with a different noun
You have a PeerAuthentication that says STRICT. You have a Gatekeeper constraint that says every namespace must carry a default‑deny AuthorizationPolicy . You have Envoy access logs with SPIFFE identities, signed evidence packs…
Kubernetes operatorsCompliance evidencePolicy as code.NET & C#Supply-chain signing
-
Building a Kubernetes operator in C# — Kubernetes has no opinion about Go
The platform team reviews the Article 43 design and says, kindly, that operators are written in Go. Kubebuilder, controller‑runtime, the Operator SDK, the language the API server itself is written in. They are not wrong about the…
-
Shipping continuous evidence as a product — a report is something you send; evidence is something an auditor comes and takes
The ComplianceProfile status says Satisfied. It is signed. It is in the cluster. An external auditor from the FMA's appointed firm arrives, and the first thing she does is decline the dashboard, decline the kubectl session, and…
Compliance evidenceSupply-chain signing.NET & C#Kubernetes operatorsMicrosoft Entra ID
No article matches those filters. Show all 45.