Compendium

Walkthrough: Design the BSS + OSS Stack for a Tier-2 Mobile Operator (5M Subscribers, 5G SA, TM Forum ODA)

24 min read

Walkthrough: Design the BSS + OSS Stack for a Tier-2 Mobile Operator (5M Subscribers, 5G SA, TM Forum ODA)

This walkthrough designs the Business Support System (BSS) and Operations Support System (OSS) stack for a tier-2 mobile network operator (MNO): ~5 million subscribers, hybrid prepaid + postpaid + B2B, 5G Standalone (SA) core with network slicing, multi-band spectrum, ~15,000 sites across an urban + suburban footprint, ~200-450M over a 3-5 year strangler-fig migration; recurring opex $30-60/subscriber/year — the industry-standard BSS+OSS opex benchmark per Analysys Mason + STL Partners 2023-2024 publications.

Reference operators and recent events shaping the design (2022-2026): Vodafone’s “Vodafone Spirit” stack consolidation (2022-2025); Telstra’s T22 / T25 strategy + Matrixx Software cloud-native BSS pilot (2023-2025); Optus’s Pega-based digital transformation; Telefónica’s Kernel + UNICA / UNICA Next cloud-native platform; AT&T’s Network Cloud + ONAP transformation; Verizon’s Project Mantra digital channels; Etisalat e& cloud-native BSS pilot with Optiva (2023); Reliance Jio’s in-house BSS at extreme scale (~450M subs on home-built); Rakuten Symphony’s fully containerized Symworld stack; Deutsche Telekom’s MagentaCloud + ONE.Identity; Vodafone-Three UK merger (approved Dec 2024) and the resulting BSS consolidation challenges; Synamedia’s video-OTT bolt-ons; CSP consolidations driven by 5G CapEx pressure; lawful intercept obligations expanded post-Salt Typhoon (Oct 2024) Chinese state intrusion into US telco infrastructure.

1. Platform spec

ParameterTargetNotes
Subscribers5M total (3M prepaid + 1.5M postpaid + 0.5M B2B)Plus MVNO wholesale tenancy capability
ARPU range30-120 postpaid; $50-1500 B2BDrives BSS revenue tier model
Daily CDR ingestion80-120M call detail records / day+ 1B+ data records, + 200M SMS records
Peak transaction TPS (charging)50K TPS authorize, 100K TPS rateOCS sized for 3-5× peak headroom
Active services per subscriber3-15 (voice, data, SMS, OTT bundles, slice access)Catalog complexity drives BSS
Network slices4-12 commercial slices (eMBB, URLLC, FWA, IoT) + privateEach with own SLA, charging
Sites~15,000 (macro 4G/5G NSA + 5G SA)+ ~150 core sites
ChannelsRetail (200 stores), Digital (app + web), Call center, USSD, B2B salesOmnichannel ordering
MTTR (BSS critical)<30 minOCS down = revenue stop
SLA charging availability99.999% (5.26 min/yr)5-9s — telco-grade
SLA OSS availability99.99% (52 min/yr)4-9s sufficient
ComplianceE911 / 112, CALEA / lawful intercept, GDPR/national privacy, anti-fraud, number portability, FCC + national regulator reportingAnnual compliance audits
Build CAPEX (BSS+OSS modernization)$200-450M over 3-5 yearsStrangler-fig from legacy
Annual OPEX (BSS+OSS)30-60/sub/yr)Including vendor support + cloud + ops

2. BSS vs OSS — the canonical split

The historical division between BSS and OSS has eroded in cloud-native designs, but the functional taxonomy still organizes the platform.

2.1 BSS — customer-facing, money-flow

FunctionResponsibility
CRMCustomer master, accounts, contacts, interactions, omnichannel touchpoints
Product catalogOffers, plans, services, prices, eligibility, dependencies
CPQ (Configure-Price-Quote)B2B custom quotes, MEF-based for enterprise data services
OrderingOrder capture, decomposition, fulfillment orchestration
Charging (OCS / PCRF / SCEF)Real-time rating + balance management for prepaid + bill-on-use
BillingPostpaid invoicing, aggregation, discounts, taxation
MediationNetwork event collection + normalization for charging + billing
Partner settlementRoaming, interconnect, MVNO settlement
Revenue assuranceDetect + recover revenue leakage (sub-1% target)
Collections + DunningLate-pay management, suspend / disconnect, restoration
Customer self-careApp, web portal, USSD, IVR

2.2 OSS — network-facing, service-flow

FunctionResponsibility
InventoryPhysical assets (sites, towers, equipment) + logical entities (cells, slices, VLANs, paths)
Service activation / provisioningTranslate ordered service into network commands
Configuration managementNetwork element parameters, change control
Fault managementAlarm collection, correlation, root cause
Performance managementKPIs (RTT, throughput, drop rate, RAN counters), trending
SLA monitoringPer-customer SLA tracking against contractual thresholds
Workforce managementField technician dispatch, work orders
Network operationsNOC dashboards, change calendar, incident management
Spectrum + radio planningCell planning, neighbor lists, frequency reuse

2.3 The blurred middle

Several functions cross BSS+OSS:

  • Service orchestration — order received in BSS, fulfillment touches both BSS records + OSS network provisioning
  • Charging + policy — OCS sits in core network (3GPP-defined as SGSN/PGW interface, now 5G SMF / UPF), but invoiced in BSS
  • Customer SLA — BSS owns the contract, OSS owns the measurement
  • Per-slice billing — completely new in 5G — requires deep BSS+OSS integration

3. TM Forum frameworks — the industry conventions

TM Forum (formerly TeleManagement Forum) publishes the de-facto industry standards for telco IT architecture. Every modern operator references at least four:

3.1 eTOM — Business Process Framework

eTOM (enhanced Telecom Operations Map) — top-down process taxonomy in three levels:

   Level 0: Strategy, Infrastructure & Product | Operations | Enterprise Management

   Level 1: ~70 process groupings under L0

   Level 2-7: Hierarchical decomposition to individual processes

   E.g., Operations / Customer Relationship Management / Order Handling /
         Determine Customer Order Feasibility

Maps to ITIL-style process modeling but telco-specific. Used to scope process re-engineering, vendor RFPs, and BPM tooling.

3.2 SID — Information Framework

SID (Shared Information / Data Model) — canonical object model for telco entities. Defines Customer, Product, Service, Resource, Party, Account, etc. Used to:

  • Design canonical data models across heterogeneous systems
  • Map vendor-proprietary models (Amdocs, Oracle, Ericsson) into common terminology
  • Architect data warehouse + master data management

3.3 TAM — Application Framework

TAM (Telecom Application Map) — catalog of ~700 telco application functions. Maps eTOM processes to logical applications. Used in vendor RFP scoring + architecture assessment.

3.4 Open APIs + ODA

TM Forum Open APIs — 70+ standardized REST APIs covering customer, product, service, resource, partner, payment, billing, troubleshooting domains. Versioned (v4 → v5 in 2024-2025). Examples:

  • TMF620 Product Catalog Management — read product hierarchy
  • TMF641 Service Ordering Management — submit orders
  • TMF652 Resource Order Management — provision resources
  • TMF678 Customer Bill Management — query bills
  • TMF666 Account Management — billing account hierarchy
  • TMF688 Event Management — event-driven integration
  • TMF635 Usage Management — usage data exchange

ODA (Open Digital Architecture, Release 24 finalized 2024) — reference architecture for cloud-native CSP IT. Five layers:

  1. Engagement — UX, omnichannel, conversational AI
  2. Domain capabilities — BSS + OSS functional building blocks (Open Digital Components, ODCs)
  3. Intelligence — data + analytics + AI
  4. Common services + integration — API gateway, event bus, identity
  5. Infrastructure + cloud — Kubernetes, CNF runtime, secret management

The vendor ecosystem ships ODA-conformant components — Matrixx Open Cloud, Optiva BSS, Amdocs Connected Service Hub, Ericsson Catalog Manager, Netcracker Digital Platform, IBM Cloud Pak for Networks, Microsoft Industry Cloud for Telecom, Salesforce Industries Communications Cloud (formerly Vlocity).

The TM Forum Conformance certification + Catalyst proof-of-concept program brings vendors + operators together — a critical procurement signal.

4. The strangler-fig modernization pattern

Greenfield BSS is rare — most operators run multi-decade legacy. The migration pattern is the strangler-fig (Fowler 2004):

   Year 0: Pure legacy
      [Amdocs CES Monolith]
      |
      v
      Customers, orders, billing all in single stack

   Year 1: Edge introduction
      [Legacy CES]                [New cloud-native BSS]
      |                                |
      |                                v
      |                       [API gateway / event bus]
      |                                ^
      |                                |
      |---- selected services routed --+

   Year 2-3: Capability migration
      [Legacy] -- shrinking ---->    [New stack] -- growing
      Product catalog ─────────────► moved
      Charging ────────────────────► moved (greenfield only)
      Postpaid billing ────────────► moved (cohort by cohort)
      CRM ─────────────────────────► moved

   Year 4-5: Legacy decom
      [Legacy minimal] ─────────────► retired
      [New cloud-native stack]

Risk: dual-running cost (paying two stacks for 2-4 years), data sync (changes in legacy must reflect in new + vice versa), rule divergence (charging rules implemented differently in two stacks = revenue leakage).

Mitigation patterns:

  • Customer-cohort migration — move new customers to new stack first; migrate legacy customer base in waves
  • Capability cohort — move one capability at a time (catalog first, then ordering, then charging, then billing)
  • Event-driven sync — single source of truth per entity; the other gets event-sourced view
  • Comprehensive reconciliation — daily comparison of derived state (balances, invoice amounts) between systems with strict tolerance

Reference programs: Telstra T22 (~5 years to substantial decom of legacy IT 2018-2023), Vodafone Spirit (2020-2024), Telefónica UNICA (2017-ongoing), Reliance Jio (greenfield 2016 — different — luxury of no legacy).

See microservices-patterns for the underlying migration architecture and industrial-organization for the competitive dynamics that drive these projects.

5. Data architecture — CDR ingestion + lake

The fundamental data engineering challenge: ingest 80-120M CDRs/day plus 1B+ data + 200M SMS records, normalize, route to charging in real-time, retain for billing dispute, archive for regulatory (often 5+ years).

5.1 Mediation layer

The mediation layer collects network events from heterogeneous sources, normalizes, deduplicates, routes:

   Network elements: MSC, PGW (4G), SMF/UPF (5G SA), IMS, OCS, AAA, MMSC,
                     IP probes, deep-packet inspection
        |
        | Diverse protocols: 3GPP Gy (Diameter), Gz (GTP'), syslog,
        |                    sFTP file pickup, RADIUS, streaming
        v
   [ Mediation ]
        |
        | Deduplication, normalization to canonical event schema,
        | enrichment (cust ID lookup, location), routing
        |
        +---> Real-time → OCS for prepaid + on-demand rating (10K-100K TPS)
        |
        +---> Near-real-time → Postpaid bill aggregator
        |
        +---> Batch → Data lake → analytics + revenue assurance + ML

Mediation throughput at 5M subs: 100M CDRs/day average ≈ 1,200 records/sec sustained; peaks 5-20× during evening peak hours.

Modern stack: Apache Kafka (or Confluent Cloud) + ksqlDB or Flink for streaming; legacy Comptel / DigitalRoute / WeDo / Subex still common.

5.2 OCS — Online Charging System

OCS sits between the network (3GPP Gy interface, Diameter protocol) and the BSS. Authorizes service usage in real-time against subscriber balance:

   Subscriber: starts data session
        |
        |  GTP-C session creation → SMF (5G) / PGW (4G)
        v
   SMF/PGW: needs charging decision
        |
        |  Diameter Credit Control Request (CCR-I) to OCS
        v
   OCS:
        |  Read subscriber balance, applicable rate
        |  Authorize initial granted units (e.g., 500 MB or 10 minutes)
        |  Diameter Credit Control Answer (CCA-I)
        v
   Session continues, consumes granted units
        |
        |  When units depleted, CCR-U for more
        |  When ended, CCR-T with final consumed
        v
   OCS: debit balance, return remaining

Performance requirements: <50 ms p99 response. Mature OCS achieve >10K TPS authorize on commodity hardware.

Vendors: Matrixx Software (cloud-native, BTML-based), Optiva BSS Platform, Ericsson Charging System, Huawei CBS, Nokia CBiO, Amdocs CES, in-house Mavenir.

5.3 PCF / SMF — policy + 5G charging

In 5G SA, charging architecture restructured:

  • PCF (Policy Control Function) — replaces 4G PCRF; manages QoS policies + URSP rules
  • SMF (Session Management Function) — owns charging trigger logic
  • CHF (Charging Function) — 5G OCS equivalent; standardized service-based architecture
  • NEF (Network Exposure Function) — exposes charging events to external

Per-slice charging (5G novelty): each network slice has independent CHF instance + charging rules; sub can have different rates on the same physical session across slices.

See design-5g-6g-network-buildout for the 5G core architecture; distributed-systems-fundamentals for the underlying ordering guarantees.

5.4 Data lake + warehouse

  • Streaming ingestion: Kafka → S3 (raw, parquet) → Apache Iceberg or Delta Lake (structured) → Snowflake / Databricks / BigQuery for analytics
  • Schemas: SID-aligned canonical entities + raw-format-preserved fact tables for CDRs
  • Retention: hot 90-180 days, warm 1-2 years, cold 5-7 years (regulatory)
  • Lineage: column-level lineage via DataHub, OpenLineage, Atlan, Collibra — required for revenue assurance audit + privacy DSRs

See database-internals and sql-nosql-design for the storage engine considerations.

6. Network slicing — BSS+OSS implications

5G network slicing is the single biggest architectural change for BSS+OSS in a decade. Each slice is a logically isolated end-to-end network with its own:

  • QoS profile (throughput, latency, jitter, reliability)
  • Authentication credentials (separate USIM profile or NSSAI)
  • Charging rules
  • SLA + uptime commitment
  • Lifecycle (instantiate, scale, terminate)

6.1 Slice taxonomy (3GPP)

  • SST=1 eMBB — enhanced mobile broadband (default, consumer)
  • SST=2 URLLC — ultra-reliable low-latency (industrial, V2X)
  • SST=3 mMTC — massive machine-type communications (IoT, sensors)
  • SST=4 V2X — vehicle-to-everything
  • Operator-defined custom SST values for slices like FWA, premium consumer, enterprise data

6.2 BSS impacts

  • Catalog: each slice = a sellable product with QoS attributes + pricing
  • CPQ: B2B customers negotiate slice-specific SLAs
  • Ordering: order triggers slice instantiation via OSS orchestration (NSMF / NSSMF)
  • Charging: per-slice usage records; can differ from underlying network charge
  • Billing: invoice line items per slice
  • Self-care: customer sees slice-level performance + usage

6.3 OSS impacts

  • Inventory: model slice as logical resource with sub-instances per geography + customer
  • Service activation: slice instantiation = NSI (Network Slice Instance) creation via E2E orchestrator
  • Performance: per-slice KPI monitoring + per-customer SLA monitoring
  • Assurance: slice SLA violations auto-credit + escalate

CRM platforms now ship “5G Slice Order to Activate” templates (Salesforce Industries Communications Cloud, Microsoft Industry Cloud for Telecom). Reality: most operators are still at slice proof-of-concept stage as of 2025-2026; commercial slice catalogs limited.

7. Customer journey orchestration + omnichannel

Telco CRM is no longer a database — it’s an orchestration layer across web, app, call center, retail store, IVR, USSD, and B2B portal.

7.1 CRM evolution

Three platform tiers compete:

TierExamplesStrength
Telco-specificAmdocs CES, Netcracker, Comverse / Mavenir, Optiva, Ericsson CatalogBuilt-in product catalog complexity, BSS-native
Cross-industry CRM with telco overlaySalesforce Industries Communications Cloud (Vlocity), Microsoft Dynamics 365 + Industry Cloud Telecom, Oracle CXStronger sales + service UX; weaker on telco-specific complexity
Process orchestrationPega Customer Decision Hub, Pega for CommunicationsStrongest decisioning + journey orchestration; bolt-on architecture

Hybrid is standard: Salesforce for B2B sales + agent UX + service; telco-native BSS for catalog + ordering + charging + billing; Pega for next-best-action + retention.

7.2 Journey orchestration

CDP-style customer journey orchestration platforms (Pega, Salesforce Interaction Studio, Adobe Journey Optimizer) track every customer interaction across channels + trigger contextual actions. Telco-specific use cases:

  • Churn prediction → retention campaign — ML scores subs daily; high-risk → next-best-offer
  • Upgrade nudge at contract renewal
  • Bill shock prevention — alert when monthly spend exceeds historical 95th percentile
  • Network event reactive — outage detected, proactive notification + credit
  • Cross-sell — broadband + mobile + content bundle propensity

7.3 Self-care

The app + web portal handle 60-80% of customer interactions in mature operators:

  • Account balance + recharge (prepaid)
  • Bill view + payment (postpaid)
  • Plan change + add-on purchase
  • Usage tracking + alerts
  • Troubleshooting + service ticket
  • eSIM activation (increasingly default for new devices)

Tech: React + React Native for web + app; backend microservices invoking BSS Open APIs (TMF629 Customer Management, TMF678 Customer Bill, TMF620 Product Catalog, TMF641 Service Ordering).

See auth-authz and design-saas-platform-launch for the multi-tenant patterns.

8. Revenue assurance + leakage

Revenue assurance (RA) is the discipline of detecting + recovering revenue lost between service usage and invoicing. Industry benchmark: 1-3% of revenue leaks; well-managed operators get below 1%.

8.1 Common leakage patterns

Leakage sourceSymptomDetection method
Unbilled usageCDRs in mediation never reach billingReconciliation: mediation count vs billed count
Missing CDRsNetwork event happened but no CDR generatedSwitch counter vs CDR count
Tariff misconfigWrong rate appliedSample audit of rated calls vs catalog
Discount over-applicationEligibility rules misappliedDaily discount audit
Provisioning without billingService active, customer not invoicedInventory ↔ billing reconciliation
Subscription expiry without renewalCustomer kept service post-expiry without chargeSub state ↔ charge audit
Interconnect under-chargeWholesale partner under-billedInterconnect CDR cross-check
FraudSubscription fraud, dealer fraud, SIM-cloningPattern detection

8.2 Tooling

  • Subex ROC — dominant in mature RA practice
  • WeDo Technologies (Mobileum) — strong on assurance + fraud
  • TEOCO — assurance + interconnect specifically
  • In-house — large operators (Vodafone, Telefónica) often build hybrid

The RA team typically runs:

  • 200-500 automated controls running daily
  • Monthly audit report to finance with leakage delta
  • KPI: leakage as % of revenue, recovery rate, time-to-detect

8.3 Fraud management

Fraud management overlaps with RA but focuses on bad-actor detection:

  • Subscription fraud — fake identity, no intent to pay
  • PBX hacking + IRSF (International Revenue Share Fraud) — fraudster routes calls to expensive numbers
  • Roaming fraud — fraudster activates service in roaming partner, generates usage, vanishes
  • SIM swap + account takeover — social engineering
  • Bypass / Refile — illegal voice termination through cheaper routes

Fraud platforms: Subex Fraud Management, Mobileum, FICO Telecom Fraud, in-house ML + rules. Detection latency target: <15 min for high-cost IRSF.

9. Number portability + roaming + settlement

9.1 Number portability

Mobile Number Portability (MNP) lets customers change operator while keeping number. Implementation per country:

  • US: NPAC (Number Portability Administration Center, operated by iconectiv) — centralized
  • EU: per-country databases coordinated under EU Universal Service Directive
  • UK: Ofcom-coordinated
  • India: MNPI / TRAI-managed
  • Japan: MNP managed by MIC

The platform integrates with country MNP database for:

  • Outbound port — customer leaves; we honor the port-out
  • Inbound port — customer joins; we receive the port-in
  • Port-out fraud prevention — PIN + ID verification

Port latency: hours in modern markets (was days a decade ago). Failures trigger regulator complaints.

9.2 Roaming

Inter-operator roaming:

  • Steering of roaming (SoR) — operator’s choice of preferred visited network per country
  • TAP (Transferred Account Procedure) — record format for roaming wholesale settlement (TAP 3.12 / NRTRDE for near-real-time)
  • CAMEL / GSM/UMTS legacy + 5G Roaming via SEPP (Security Edge Protection Proxy)
  • Pricing: home operator pays visited operator per minute / SMS / MB; wholesale rates set by Inter-Operator Tariff (IOT) negotiation
  • Settlement: monthly netting via clearing house (Syniverse, BICS, TOMIA, iBASIS)

Roaming hub partners: Syniverse (dominant in US), BICS (BICS BlueVerse), TOMIA, iBasis (KPN), Comfone, IDT Global.

9.3 Interconnect

Voice interconnect — when subscriber A on operator X calls subscriber B on operator Y, X pays Y a termination fee per minute. Wholesale CDRs reconciled monthly.

Rates: 0.001-0.1 USD/min depending on country + termination type (mobile vs fixed). MTRs (Mobile Termination Rates) regulated in EU + most markets; declining 2010-2024 as part of regulator push to lower retail prices.

10. Cloud + cloud-native deployment

10.1 The shift to cloud

Tier-1 operators ran BSS on Solaris + Oracle Exadata + bare-metal x86 in carrier-grade data centers for decades. The 2020-2026 shift:

  • Public cloud for some workloads — Telstra + Microsoft Azure (2022+), Vodafone + Google Cloud, Verizon + AWS (BSS workloads), AT&T + Microsoft, T-Mobile US + selective
  • Telco private cloud — VMware Telco Cloud (pre-Broadcom acquisition turmoil), Red Hat OpenShift, Wind River Studio, vRAN distributed cloud
  • Hybrid — most operators run mix: public cloud for digital-channel front-ends + analytics, private cloud for OCS + charging + sensitive workloads

10.2 CNF (Cloud-Native Network Function) + ODA

TM Forum ODA’s vision: BSS+OSS components ship as CNFs (Helm charts on Kubernetes) with TM Forum Open APIs as the standard interface.

Reality 2025-2026: ~40-50% of new BSS modules ship cloud-native; legacy migration of OCS + billing still slow. Matrixx Software, Optiva, Mavenir, Rakuten Symphony lead the cloud-native side; Amdocs + Netcracker + Ericsson modernizing existing stacks.

10.3 Tenant + multi-country deployment

Larger operators run multi-country instances:

  • Vodafone Group: shared platform with country instances (Italy, UK, Germany, Spain, etc.)
  • Telefónica: UNICA / UNICA Next as group platform
  • Orange: shared services + country instances
  • VEON: cross-country roll-up

Architectural patterns:

  • Shared catalog + per-country pricing
  • Shared product engineering + per-country compliance overlays
  • Country-specific tax engines + regulatory reporting
  • Per-country data residency

11. Lawful intercept + regulatory

Every telco operates under lawful intercept (LI) obligations — Communications Assistance for Law Enforcement Act (CALEA, US 1994), ETSI Lawful Interception (EU), country-specific equivalents.

11.1 Architecture

   LEA (Law Enforcement Agency) submits warrant
        |
        v
   Operator LI mediation function (LIMF / X1, X2, X3 interfaces)
        |
        |  X1: warrant administration (HI-1 interface)
        |  X2: intercept-related information (IRI, HI-2)
        |  X3: communication content (CC, HI-3)
        v
   Delivered to LEA monitoring facility (MF)

Standards: 3GPP TS 33.107 + TS 33.108 (LI architecture + handover); ETSI TS 102 232 (handover interfaces). 5G LI defined in 3GPP TS 33.126 + TS 33.127 / 128.

11.2 Compliance discipline

  • Segregation of duties — LI ops staff distinct from regular ops; cannot access non-warranted data
  • Audit trail — every LI activation logged with warrant reference + judicial authorization
  • Strict access control — fewer than 0.1% of telco staff have LI access; multi-person approval
  • Vendor isolation — LI provider isolated from main BSS+OSS

11.3 Post-Salt Typhoon (Oct 2024)

Disclosed Oct 2024: Chinese state-sponsored attacker “Salt Typhoon” infiltrated US telecom backbone networks (AT&T, Verizon, T-Mobile, Lumen, Charter) — including the LI / law-enforcement-portal systems. Implications:

  • Heightened scrutiny on LI architecture security
  • Push toward zero-trust networking inside telco IT
  • Renewed isolation of LI from production network
  • FCC + CISA + DoJ workgroup on telco cybersecurity 2025+
  • Anticipated 2026 FCC rule requiring stronger LI segmentation

See design-soc-incident-response-platform for the security-operations side.

11.4 Other regulatory obligations

  • E911 / 112 — emergency-call routing + location reporting to PSAP (Public Safety Answering Point); strict latency + accuracy SLAs
  • Number portability — coordinator participation per country
  • Universal service — coverage obligations + reporting
  • Net neutrality — varies by country; EU Open Internet Regulation; US after FCC 2024 net neutrality rule (subsequent litigation 2024-2025)
  • Spectrum compliance — license conditions, build-out milestones, regulator reporting
  • Privacy — GDPR EU + state laws in US (California, etc.) + national privacy frameworks
  • Anti-money laundering — for prepaid sales above thresholds; SIM-registration mandates in many countries
  • Tax + universal service fund — telco-specific tax structures, USF contributions

Regulators: FCC + state PUCs (US), Ofcom (UK), BEREC + national NRAs (EU), TRAI (India), MIC (Japan), MIC China, CMA (Italy), CRTC (Canada).

12. SLA monitoring + assurance

12.1 SLA tiers

Consumer: best-effort + statutory minimums (varies by country)

Enterprise:

  • Silver — 99.9% uptime, MTTR 4 hr business hours
  • Gold — 99.95% uptime, MTTR 2 hr 24×7
  • Platinum — 99.99% uptime, MTTR 1 hr 24×7
  • Custom — per-customer 99.999% with bonded credits

Mobile B2B + private slice: 99.999% with per-second granularity in some 5G URLLC contracts.

12.2 Measurement architecture

  • End-to-end probes — synthetic test transactions from customer site simulators or partner probe networks
  • Network counter aggregation — per-cell + per-customer counters from OSS PM
  • Customer device telemetry — opt-in via app for crowdsourced signal data
  • Per-slice KPI — 5G slice gives per-slice KPI by design

Assurance: when SLA missed → auto-credit + auto-notify + RCA (root cause analysis). Manual workflow common; emerging AI-driven RCA (Cisco Crosswork, Anuta, Blue Planet, Microsoft AI for IT Ops).

12.3 Workforce management + field operations

Field technician dispatch for site issues + customer premises equipment:

  • Vendors: ServiceMax, Salesforce Field Service, Oracle Field Service, IFS Cloud
  • Optimization: routing, scheduling, parts availability, skill matching
  • Integration: from fault management → auto-create work order → assign → dispatch

13. Trade-off matrix — key build decisions

DecisionOption AOption BTrade-off
OCSMatrixx (cloud-native)Amdocs CES (legacy modernized)Matrixx: future-proof + ODA-conformant, smaller install base; Amdocs: install base + maturity, but cloud transition incomplete
CRMSalesforce IndustriesTelco-native (Netcracker, Amdocs)Salesforce: stronger UX, weaker BSS-native; telco-native: better integration, weaker UX
CloudPublic (AWS / Azure / GCP)Telco private (Red Hat, VMware)Public: faster innovation + lower opex; private: regulatory comfort + LI segmentation
OSS orchestrationONAP-basedVendor (Ericsson EMM, Nokia FlowOne)ONAP: open + community + complex; Vendor: faster but lock-in
Slicing approachCustom per sliceTemplated catalogCustom: maximum flexibility for B2B; templated: faster operationally
Migration speedBig-bangStrangler-figBig-bang: 18-month vs 4-year, higher risk; strangler: lower risk + longer dual-run cost
Build vs buyGreenfield in-houseVendor BSS suiteIn-house: full control + costlier + slower; vendor: faster + lock-in

14. Failure modes

  • OCS down — all prepaid revenue stops, postpaid usage builds up unrated. Existential. Mitigation: active-active OCS clusters across data centers; graceful degradation (apply default rate, reconcile post-recovery).
  • Mediation backlog — Kafka lag, downstream charging stops. Customer service inquiries spike on bill anomalies. Mitigation: lag SLO + auto-scale; backpressure to upstream.
  • Tariff misconfig — wrong rate goes live, mass over-charge or under-charge. Regulatory exposure if over-charge. Mitigation: catalog change approval workflow + sandbox simulation + canary on rollout.
  • Provisioning fail — order succeeds in BSS but doesn’t activate on network. Customer paid but no service. Mitigation: order orchestration with compensation + rollback; reconciliation reports daily.
  • Bill shock — customer hit with massive bill (roaming, IRSF fraud, misconfig). Regulatory complaint magnet. Mitigation: spend caps + bill-shock alerts + transparent caps.
  • Number-port stuck — port-out hangs in coordinator; customer cannot leave. Regulator complaint. Mitigation: clear escalation runbook; reconciliation with NPAC daily.
  • Salt Typhoon-class intrusion — adversary in LI plane reading warrants + tapping non-warranted traffic. Existential national-security issue. Mitigation: zero trust + air-gapped LI + frequent forensic audit.
  • Privacy breach — CDR dump exposed via misconfigured cloud storage. GDPR fines up to 4% global revenue. Mitigation: encryption at rest + access logging + DSPM tooling (Wiz, Securiti, BigID).
  • Vendor financial collapse — small BSS module vendor folds mid-migration. Mitigation: contractual source-code escrow + transition rights + qualify financial health.
  • Country regulator change — sudden mandate (e.g., spectrum auction conditions, USF contribution, lawful intercept new format) requires 6-month delivery. Mitigation: configurable engine + compliance reserve budget.

15. Cost build

15.1 Build CAPEX (3-5 year strangler-fig)

ItemCost
Engineering + product (250 FTE × 4 yr × $180K loaded blended)$180M
Vendor licenses (Matrixx, Salesforce, Oracle, Pega, Veeva) Y1-3 commits$40M
Cloud infra (private cloud + selective public, dual-region)$20M
Migration consulting (Accenture, Infosys, TCS, Wipro, IBM consulting)$30-60M
Data migration + reconciliation tooling$5-10M
Compliance + lawful intercept rebuild$4-8M
Audit + cert + security review$2-4M
Build CAPEX total**450M+ on legacy complexity)

15.2 Annual OPEX

At 5M subscribers, 150-300M annual BSS+OSS opex. Decomposition:

ItemAnnual cost
Vendor software licences (Matrixx, Salesforce, Pega, Veeva, etc.)$40-80M
Cloud infrastructure (hybrid)$20-40M
Engineering + operations headcount (300 FTE blended $160K)$48M
Field workforce management$15-25M
Compliance + lawful intercept ops$5-10M
Vendor support + audit + cert renewals$5-8M
Customer-facing channels (call center IT, retail systems)$20-40M
Roaming + interconnect clearing fees$5-10M
Annual OPEX total$158-261M

15.3 Revenue + economics

5M subs × 1.5B annual revenue typical for tier-2.

BSS+OSS opex as % revenue: 10-20% — high for an IT spend but typical for telco where the BSS+OSS IS the product delivery vehicle. Margin compression continues 2024-2026 as ARPU stagnates and 5G CapEx mounts.

See industrial-organization for the telecoms competitive structure.

16. Observability + operations

16.1 KPIs to watch

KPITarget
OCS availability99.999%
Order-to-activate time<5 min for digital orders, <2 hr for complex
Bill cycle completion100% by SLA day (typically -2 days from delivery)
Mediation lag<30 sec p99
First-call resolution>70% on customer service
Churn rate<1% monthly postpaid, <3% prepaid
NPS30+ telco-industry benchmark
Revenue leakage<0.5%
Mean time to detect fraud<15 min

16.2 Instrumentation

  • OpenTelemetry for app traces + metrics (cloud-native services)
  • Splunk + Elastic for log aggregation across legacy + cloud
  • AppDynamics / Dynatrace / Datadog for APM (large operators have all three for different domains)
  • TM Forum TMF725 Resource Performance Management for network-side KPI exposure
  • Custom dashboards per business unit + per channel

16.3 NOC + SOC

  • Network Operations Center (NOC): 24×7 network event monitoring + incident response, often co-located with key core sites
  • Security Operations Center (SOC): cybersecurity-focused 24×7 — see design-soc-incident-response-platform
  • Crisis management: major-incident handling with cross-functional team + regulator + media comms

17. Adjacent

18. Glossary

  • 3GPP — Third Generation Partnership Project (mobile standards body)
  • AAA — Authentication, Authorization, Accounting
  • ARPU — Average Revenue Per User
  • B2B — Business-to-Business
  • BSS — Business Support System
  • CALEA — Communications Assistance for Law Enforcement Act (US)
  • CAMEL — Customized Applications for Mobile Networks Enhanced Logic
  • CDR — Call Detail Record
  • CHF — Charging Function (5G)
  • CNF — Cloud-Native Network Function
  • CPQ — Configure-Price-Quote
  • CRM — Customer Relationship Management
  • CSP — Communications Service Provider
  • DPI — Deep Packet Inspection
  • E911 / 112 — Emergency call services (US / EU)
  • eMBB — Enhanced Mobile Broadband (5G)
  • ETSI — European Telecommunications Standards Institute
  • eTOM — enhanced Telecom Operations Map (TM Forum)
  • FCC — Federal Communications Commission (US)
  • FWA — Fixed Wireless Access
  • GDPR — General Data Protection Regulation
  • HI-1/2/3 — Handover Interface (lawful intercept)
  • IMS — IP Multimedia Subsystem
  • IOT — Inter-Operator Tariff
  • IRSF — International Revenue Share Fraud
  • KPI — Key Performance Indicator
  • LEA — Law Enforcement Agency
  • LI — Lawful Intercept
  • mMTC — Massive Machine-Type Communications
  • MNO — Mobile Network Operator
  • MNP — Mobile Number Portability
  • MSC — Mobile Switching Center
  • MTR — Mobile Termination Rate
  • MVNO — Mobile Virtual Network Operator
  • NEF — Network Exposure Function (5G)
  • NPAC — Number Portability Administration Center
  • NRA — National Regulatory Authority
  • NSSAI / NSI — Network Slice Selection Assistance Information / Instance
  • OCS — Online Charging System
  • ODA — Open Digital Architecture (TM Forum)
  • OEM — Original Equipment Manufacturer
  • ONAP — Open Network Automation Platform
  • OSS — Operations Support System
  • PCF / PCRF — Policy Control Function / Policy + Charging Rules Function
  • PGW — Packet Data Network Gateway (4G)
  • PSAP — Public Safety Answering Point
  • PUC — Public Utility Commission (US states)
  • RA — Revenue Assurance
  • RAN — Radio Access Network
  • SEPP — Security Edge Protection Proxy (5G roaming)
  • SID — Shared Information / Data Model (TM Forum)
  • SLA — Service Level Agreement
  • SMF — Session Management Function (5G)
  • SoR — Steering of Roaming
  • SST — Slice / Service Type
  • TAM — Telecom Application Map (TM Forum)
  • TAP — Transferred Account Procedure (roaming)
  • TPS — Transactions Per Second
  • UPF — User Plane Function (5G)
  • URLLC — Ultra-Reliable Low-Latency Communications (5G)
  • USF — Universal Service Fund (US)
  • USSD — Unstructured Supplementary Service Data

Graph View

Backlinks