Skip to content
Kamiti Labs
The Playbook

Volume 3 · Chapter 3Enterprise Reference Architecture, New Relic Observability & Monitoring Platform

The Technical Blueprint

The target-state architecture for enterprise observability — FMCG reference architecture, the full New Relic and OpenTelemetry strategy, Kubernetes and database monitoring, dashboards, alerting, and AI-assisted operations.

01

Executive Summary

The enterprise monitoring vision

A modern FMCG enterprise operates hundreds of interconnected technology components. Every customer order, warehouse movement, production schedule, dealer transaction, and financial posting depends on applications performing reliably. Monitoring only servers or databases is no longer enough.

Kamiti Labs proposes a unified observability platform where infrastructure, applications, cloud services, Kubernetes, APIs, databases, business transactions, and user experience are monitored through a single operational framework.

The goal is one source of operational truth for both technical teams and business stakeholders.

Business objectives

  • Achieve end-to-end visibility across the digital landscape.
  • Detect incidents before users report them.
  • Reduce Mean Time to Detect (MTTD) and Mean Time to Resolve (MTTR).
  • Improve application availability and customer experience.
  • Enable proactive capacity planning and performance optimization.
  • Support business continuity and executive reporting.

Enterprise observability vision

  1. 1

    Business users

  2. 2

    Digital applications

  3. 3

    Infrastructure & cloud

  4. 4

    Unified observability platform

  5. 5

    AI-assisted operations

  6. 6

    Reliable business services

02

Customer Perspective

The typical FMCG technology landscape

Understanding the customer’s ecosystem is essential before onboarding monitoring. A large FMCG organization typically operates five clusters of business systems.

Customer & sales

  • Consumer mobile app
  • Dealer portal
  • Distributor portal
  • Retail ordering portal
  • E-commerce platform
  • CRM

Manufacturing

  • Manufacturing Execution System (MES)
  • Production planning
  • Quality control
  • IoT sensors
  • SCADA interfaces

Supply chain

  • Warehouse Management System (WMS)
  • Transportation Management System (TMS)
  • Inventory management
  • Barcode systems
  • Vendor portal

Enterprise systems

  • SAP ERP / Oracle ERP
  • HRMS
  • Finance
  • Procurement
  • Payroll

Analytics

  • Data warehouse
  • Business intelligence
  • AI/ML
  • Forecasting
  • Executive dashboards

Reference architecture

  1. 1

    Consumers, dealers, retailers & employees

  2. 2

    Internet

  3. 3

    CDN + WAF

  4. 4

    API Gateway

  5. 5

    Load balancer

  6. 6

    Kubernetes platform

  7. 7

    Microservices

  8. 8

    Business APIs

  9. 9

    ERP, CRM, WMS, Finance

  10. 10

    Databases

  11. 11

    Cloud infrastructure

03

Architecture

New Relic reference architecture

New Relic provides a unified observability platform that combines infrastructure monitoring, APM, distributed tracing, browser and mobile monitoring, Kubernetes monitoring, synthetic monitoring, logs, dashboards, and alert intelligence. Instead of stitching together disconnected tools, New Relic becomes the single operational platform.

Logical architecture

  1. 1

    Applications, VMs, containers, Kubernetes, databases & cloud services

    The full estate under management.

  2. 2

    OpenTelemetry

    Vendor-neutral instrumentation layer.

  3. 3

    New Relic agents

    Language and infrastructure agents collect telemetry.

  4. 4

    Telemetry pipeline

    Metrics, logs, and traces are ingested centrally.

  5. 5

    New Relic platform

    Data is stored, correlated, and made queryable.

  6. 6

    Dashboards, alerts, NRQL, AI & incident intelligence

    The operational surface teams work from daily.

04

Standards & Best Practices

Monitoring layers

Monitoring is implemented in five layers, from the infrastructure up to the business outcomes it exists to protect.

Layer 1 — Infrastructure

  • CPU
  • Memory
  • Disk
  • Filesystems
  • Network
  • Processes
  • OS health

Layer 2 — Platform

  • Kubernetes cluster
  • Nodes
  • Pods
  • Services
  • Ingress
  • Namespaces
  • Autoscaling
  • Persistent volumes

Layer 3 — Applications

  • Response time
  • Throughput
  • Error rate
  • JVM/.NET runtime
  • API performance
  • Exceptions

Layer 4 — Databases

  • Connections
  • Slow queries
  • Locks
  • Replication
  • Buffer cache
  • Storage growth

Layer 5 — Business

  • Orders processed
  • Dealer transactions
  • Consumer logins
  • Payment success rate
  • Inventory synchronization
05

Implementation Guidance

OpenTelemetry strategy

OpenTelemetry provides a vendor-neutral way to collect telemetry. Kamiti Labs recommends it as the standard instrumentation framework across Java, .NET, Node.js, Python, and Go — for standardized telemetry, future portability, consistent traces across microservices, and easier troubleshooting.

Telemetry flow

  1. 1

    Application

  2. 2

    OpenTelemetry SDK

  3. 3

    Metrics, logs & traces

  4. 4

    Collector

  5. 5

    New Relic

  6. 6

    Dashboards & alerts

06

Architecture

Kubernetes monitoring

Modern FMCG applications increasingly run on Kubernetes. Coverage spans cluster health, node health, workloads, networking, and autoscaling.

Cluster health

  • API server
  • Scheduler
  • etcd
  • Controller manager

Node health

  • CPU
  • Memory
  • Disk
  • Node availability

Workloads

  • Pods
  • Deployments
  • ReplicaSets
  • StatefulSets
  • DaemonSets

Networking

  • Ingress
  • Services
  • DNS
  • Load balancers

Autoscaling

  • HPA
  • Cluster autoscaler
  • Node pools

Monitoring objectives

  • Detect unhealthy nodes.
  • Identify pod crashes.
  • Monitor restart frequency.
  • Track resource utilization.
  • Prevent capacity bottlenecks.
07

Standards & Best Practices

Database observability

Every production database should have a monitoring baseline appropriate to its engine.

Oracle

  • Sessions
  • Tablespaces
  • ASM
  • Wait events
  • AWR indicators

PostgreSQL

  • Active connections
  • Vacuum health
  • Replication lag
  • Index usage
  • Slow queries

SQL Server

  • Blocking
  • Deadlocks
  • TempDB
  • Buffer cache

MySQL

  • Replication
  • Query performance
  • InnoDB health
08

KPI & SLA Examples

Business transaction monitoring

Technical metrics alone do not tell the full story. Every critical business transaction needs to be watched end to end, not just the infrastructure underneath it.

Transactions to monitor

  • Dealer login
  • Product search
  • Order creation
  • Payment processing
  • Invoice generation
  • Warehouse dispatch
  • Shipment confirmation

Tracked for every transaction

  • Availability
  • Response time
  • Error rate
  • Success percentage
  • User journey
09

Deliverables

Dashboard architecture

Different audiences need different dashboards — an executive and an on-call engineer should never be staring at the same screen.

Executive dashboard

  • Availability
  • SLA compliance
  • MTTR
  • Critical incidents
  • Business KPIs

Operations dashboard

  • Open alerts
  • Active incidents
  • Infrastructure health
  • Application health

SRE dashboard

  • SLO compliance
  • Error budget
  • Deployment frequency
  • Service reliability

Application dashboard

  • Response time
  • Throughput
  • Errors
  • JVM health
  • API latency
10

Common Challenges

Alert strategy

Alerting must be actionable. Every alert is classified by severity, and every alert definition carries the same set of fields — otherwise alert fatigue sets in and real incidents get lost in the noise.

Severity classification

  • Critical (P1)
  • High (P2)
  • Medium (P3)
  • Informational (P4)

Every alert defines

  • Trigger condition
  • Business impact
  • Notification targets
  • Escalation path
  • Runbook
  • Auto-remediation options

Avoid alert fatigue by tuning thresholds and suppressing duplicate events.

11

Kamiti Recommendations

AI-assisted operations

New Relic’s AI capabilities assist operations teams rather than replace their judgment.

  • Correlating related alerts.
  • Detecting anomalies.
  • Highlighting probable root causes.
  • Reducing duplicate incidents.
  • Prioritizing customer-impacting events.

Kamiti Labs treats AI recommendations as decision support, with production changes remaining subject to operational controls and approvals.

12

KPI & SLA Examples

Success metrics

Measurable targets are defined from day one, not retrofitted after go-live.

KPITarget
Platform availability≥99.9% (or customer-agreed SLA)
Critical alert acknowledgement≤15 minutes
P1 initial response≤15 minutes
Dashboard coverage100% of production applications
Instrumented services100% of in-scope microservices
Synthetic monitoringAll customer-facing applications
Monthly availability report100% delivery

Deliverables

What Volume 3 produces

By the end of the observability implementation, Kamiti Labs delivers architecture, New Relic configuration, and documentation as three distinct, handover-ready packages.

Architecture

  • Enterprise monitoring architecture
  • Application instrumentation standards
  • Kubernetes monitoring design
  • Database monitoring design
  • Business transaction monitoring design

New Relic

  • Agent deployment plan
  • Dashboard catalogue
  • Alert catalogue
  • NRQL query library
  • Service map
  • Entity inventory

Documentation

  • Monitoring standards
  • Instrumentation guide
  • Dashboard user guide
  • Alert response guide
  • Operations handover document

Consultant Tips

Consultant’s note

They aren’t buying a tool

Enterprise customers rarely buy a monitoring tool — they buy confidence that their business-critical services will remain available and that issues will be detected and resolved quickly. Position New Relic as the foundation of a broader observability strategy that combines technology, governance, and operational excellence.

The value lies not just in dashboards, but in the processes, people, and continuous improvement practices that surround them.