MTN has migrated its Enterprise Value Analytics (EVA) platform to Microsoft Azure, unveiling a cloud-native EVA 3.0 that the operator says now runs on Azure Databricks, is protected by Microsoft Defender, and processes roughly 22 billion records per day across more than 800 analytics workflows fed by over 1,700 data sources — a modernisation MTN and partners describe as the largest telco cloud implementation in the Middle East and Africa and a blueprint for Group-wide data modernisation.
Background / Overview
MTN’s move is the latest, high-profile example of telco digital transformation anchored on hyperscale cloud platforms and the lakehouse paradigm. Over the past several years MTN and Microsoft deepened a strategic partnership that positioned Azure as a primary public-cloud partner for network, OSS/BSS and analytics modernization. The EVA 3.0 announcement frames the migration as a re‑engineering — not merely a lift-and-shift — to achieve near-real-time analytics, improved operational agility, and a foundation for responsible AI. This article provides a rigorous technical and commercial analysis of EVA 3.0: what MTN says it built, why the chosen architecture is sensible for telco-scale analytics, the observable benefits, and the operational and governance risks that enterprises and telcos should weigh when adopting a similar blueprint.What MTN says EVA 3.0 delivers
MTN’s public statements and trade coverage consistently highlight a cluster of capabilities the cloud migration unlocks:- Massive ingestion and throughput — roughly 22 billion records per day, ingesting thousands of telemetry and business feeds and executing hundreds of analytics workflows.
- Faster time-to-insight — improved processing speed and near‑real‑time analytics to detect service incidents earlier and reduce mean time to remediation.
- Operational automation and closed‑loop processes — the ability to detect anomalies, correlate root cause across data domains, and trigger remediation or operational playbooks with minimal manual intervention.
- Customer experience and commercialisation — richer cross-domain analytics (network telemetry, CRM, OSS/BSS, usage) to create more relevant offers, better retention campaigns and potential analytics-as-a-service for enterprise customers.
- Security and responsible AI — platform protection using Microsoft Defender and Azure governance services, plus explicit emphasis on model registries, explainability, and staged rollouts to operationalise responsible AI.
Technical architecture: the lakehouse foundation
Core components (as reported and inferred)
MTN’s EVA 3.0 maps to a contemporary cloud-native lakehouse architecture. The public descriptions and technical inferences point to the following component stack:- Ingestion: high-throughput collectors for streaming telemetry (network probes, packet/flow, CDRs), OSS/BSS events, and application logs using Kafka/Event Hubs or managed ingest services.
- Storage: Azure Data Lake Storage (ADLS Gen2) holding Delta Lake-formatted tables for ACID-like semantics, time travel and schema evolution.
- Compute: Azure Databricks providing Spark-based scaling for streaming and batch ETL, feature engineering, ML training and inferencing.
- Orchestration: Databricks Jobs/Workflows, optionally paired with Azure Data Factory or Synapse pipelines to coordinate complex dependencies across hundreds of jobs.
- Security & governance: Azure Active Directory/Entra for identity, Microsoft Defender for workload and data-plane protection, and Unity Catalog or equivalent metadata tooling for fine-grained access and lineage.
Why Databricks + Delta Lake makes sense for telcos
Telco analytics requires both high velocity (streams of telemetry) and heavy cross-domain correlations (billing, CRM, inventory). The Databricks lakehouse approach addresses these needs by:- Providing autoscaling Spark clusters for bursty workloads (e.g., network events).
- Offering Delta Lake semantics for consistent incremental processing and schema evolution.
- Enabling a shared, governed data surface for analytics, ML and downstream operational systems.
Scale & performance: parsing the headline numbers
MTN reports EVA 3.0 handles ~22 billion records per day, 800+ analytics workflows, and ~1,700 data feeds. Those figures are striking and, if steady-state, indicate a highly parallelised architecture with mature platform engineering practices. The practical implications include:- Need for aggressive autoscaling and partitioning strategies to avoid small-file and compaction problems.
- Sophisticated back-pressure handling and dead-letter strategies to prevent pipeline collapse.
- Deep investment in observability: SLAs, latency percentiles, job failure metrics, and cost-per-event tracking.
- A strong testing and runbook culture to manage interdependent workflows and retries.
What EVA 3.0 enables for operations and CX
The business and operational outcomes MTN emphasises are plausible outcomes of modernising to a lakehouse on Azure:- Faster incident detection and reduced MTTR through near-real-time streaming analytics and automated root-cause correlation.
- Better personalization via cross-domain feature engineering (combining network QoS, usage and CRM signals).
- Potential new revenue streams by packaging anonymised network analytics for enterprise customers or by exposing analytics capabilities internally as a product.
- A repeatable deployment blueprint that reduces duplicate engineering work across MTN’s 19 markets if templatized correctly.
Security, governance and regulatory considerations
Protecting telco telemetry and customer data at this scale requires a layered, auditable security posture. Public reporting states the platform is “protected by Microsoft Defender” and integrated with Azure identity and governance services — a solid baseline but not a silver bullet. Key expectations for a telco lakehouse include:- Zero Trust identity controls (MFA, conditional access, tightly scoped roles).
- Data classification, tokenisation/pseudonymisation for PII and sensitive location telemetry.
- Network isolation for highly sensitive flows (ExpressRoute/private peering), strict egress policy controls, and auditable cross-border transfer governance.
- Integration of Defender alerts into a SIEM (for example, Microsoft Sentinel) and mature incident response playbooks with runbook automation.
- Responsible AI controls: model registries, explainability tooling, bias testing, and staged deployment with human-in-the-loop for high-risk actions.
Strengths: what MTN (and similar telcos) stand to gain
- Operational velocity: centralized analytics and near-real-time visibility compress detection-to-remediation cycles and materially improve NOC responsiveness.
- Repeatability: a templatized EVA 3.0 can accelerate rollouts across markets and lower duplicate engineering effort.
- Platform economics and innovation: access to hyperscaler-managed services reduces custom stack maintenance and accelerates adoption of new Azure/Databricks features for AI and analytics.
- Improved governance potential: a single cloud-provider stack can make unified policy enforcement and auditing more straightforward when implemented rigorously.
Risks and limitations — what to watch for
- Vendor lock-in and architectural coupling
- Deep coupling to Azure Databricks and Microsoft security tooling simplifies operations but increases migration costs if future strategic direction changes. Design modular data access layers and exportable data contracts to mitigate lock-in.
- Cost governance
- At the scale of billions of records daily, unchecked autoscaling and inefficient job patterns can drive exponential monthly cloud bills. Strong cost telemetry, quotas, and job cost attribution are essential.
- Operational complexity
- Running hundreds of interdependent workflows at telco scale requires mature SRE practices, reliable orchestration, and runbook automation. Expect a multi-year engineering investment beyond the initial migration.
- Data sovereignty and regulatory exposure
- Cross-border processing, lawful interception obligations and PII handling must be contractually and operationally addressed for each jurisdiction where the platform will be applied. Certifications alone don’t guarantee compliance.
- Claims verification
- Public numerical claims (22B records/day, 800 workflows, 1,700 feeds) are plausible but company-reported. Third-party benchmarks or audit logs would strengthen procurement and partner evaluation.
Replicability across MTN Group markets: practical realities
MTN positions EVA 3.0 as a group-wide blueprint. Reuse is achievable — but it requires:- Strong Cloud Centre of Excellence (CCoE) templates for infrastructure-as-code, data models and security baselines.
- Localised data residency and regulatory mappings for each market.
- Targeted upskilling and staff rotations to transfer operational knowledge.
- A staged rollout cadence that validates performance and governance in one market before replication.
Practical recommendations for telcos and enterprise IT teams
- Treat headline throughput numbers as vendor-provided until validated by an independent technical case study or third-party audit. Demand measurable SLAs, cost metrics, and proof-of-concept results at scale.
- Design for graceful degradation: maintain isolated failover paths for critical NOC tooling and ensure identity/control-plane contingencies exist for portal or authorization outages.
- Bake cost governance into pipelines from day one: assign job-level cost tags, create per-team budgets and run monthly cost reviews tied to business outcomes.
- Invest in governance automation: policy-as-code, automated data classification, and centralised metrics for data access and model performance.
- Prioritise responsible AI controls for customer-impacting models: model registries, bias checks, post-deployment monitoring and human-in-the-loop approvals for high-impact actions.
Verification summary and journalistic caution
Independent trade reporting (CIO Africa, ITWeb Africa, TheFastMode and others) corroborates the core narrative: MTN migrated EVA to Azure Databricks and emphasised the improved performance and security posture of EVA 3.0. However, several important technical metrics cited in MTN’s announcement — notably the exact daily record count, workflow count and feed tally — are company-reported and have not been independently audited in the public domain. Procurement teams, partners and enterprise customers should request detailed benchmarks, billing and telemetry extracts, or third-party verification during due diligence.Final assessment: meaningful step, but the hard work is ongoing
EVA 3.0 is a significant and credible step for MTN toward cloud-native, AI-ready telco operations. The combination of Azure Databricks, Delta Lake semantics and Microsoft security tooling is a pragmatic and well-supported choice for telco-scale analytics. The reported benefits — faster detection, improved personalization and a reusable blueprint — align with industry best practices for lakehouse deployments. At the same time, the real work begins after migration: controlling cost, operationalising security and compliance across multiple jurisdictions, validating throughput and reliability at scale, and transforming organisational processes so analytics become productised and operationally resilient. MTN’s investment in skilling and a Cloud Centre of Excellence addresses these non‑technical dimensions, but process, governance and independent verification will determine long-term success.EVA 3.0 is a template other telcos will study closely — not because it eliminates risk, but because it crystallises the pragmatic trade-offs that come with moving telco analytics into hyperscale clouds: faster insights and new revenue potential balanced against vendor coupling, regulatory complexity and the need for sustained platform engineering. The next visible milestones to watch will be published technical case studies, independent performance audits and MTN’s documentation of cross-market rollouts that demonstrate the blueprint’s reproducibility under varying regulatory constraints.
Conclusion
MTN’s migration of EVA to Azure marks a clear inflection point in telco cloud adoption in Africa and the Middle East. The architectural choices — Azure Databricks, Delta Lake, and integrated Microsoft security — are defensible and powerful when paired with disciplined engineering, cost governance and robust regulatory controls. The promise is real: earlier detection of faults, faster operational responses, and the ability to build personalised, AI-driven customer experiences at scale. The caveat is equally real: headline metrics are company-reported, and the sustained operational, security and compliance work required to make the platform both durable and affordable will be the decisive factor in whether EVA 3.0 becomes a long-term competitive advantage or a costly experiment.
Source: Developing Telecoms MTN moves customer analytics platform to Microsoft Azure cloud
