Skip to content

OrganizationPractitioner12 min read

3. Governance and roadmap

A four-phase roadmap, a RACI built with the teams, a choice between self-managed, SaaS and hybrid made with the same criteria, a solution evaluation grid and the anti-patterns that lie in wait for a CIO.

For: architects · team managers · executives and CIOsPrerequisites: Have read lesson 1 of the course.

Reading mode

Without governance, an observability strategy quickly turns into a collection of isolated projects that each go their own way. This lesson gives three tools to avoid that: a phased roadmap, a RACI and a selection method.

You build observability like a house: foundations first, then walls, then finishing. Each phase rests on the previous one. The durations below are indicative and can be adjusted to your scope; the order, however, must be respected.

flowchart LR
  P1["Phase 1<br/>Foundations<br/>0 to 3 months"] --> P2["Phase 2<br/>Structuring<br/>3 to 6 months"] --> P3["Phase 3<br/>Optimization<br/>6 to 12 months"] --> P4["Phase 4<br/>Maturity<br/>12 months and beyond"]

You take stock of tools, data sources and teams. You choose the target architecture based on the business case from lesson 1. You define service level indicators (SLIs) for critical services, meaning what you measure to know whether a service is doing its job: the share of orders that go through, the time a page takes to display. You also deliver two or three quick wins, achievable in under six weeks, to build trust.

You instrument priority services with OpenTelemetry, the open standard that describes how applications emit their observability data. You formalize service level objectives (SLOs), those numerical promises along the lines of “99.5% of orders go through in under two seconds”. Alerts rely on these SLOs and on burn rate, the speed at which you are using up the margin of error you allow yourself, like a fuel gauge dropping too fast. The RACI is signed and on-call is organized.

You set up distributed traces on critical journeys, to follow a request from one end of the system to the other the way you track a parcel. You link metrics, logs and traces together. Blameless post-mortems become systematic for major incidents. You steer the cost of observability data (see lesson 4).

Configurations, dashboards, alerts and SLOs are managed as code: written in files, under version control, reviewed before they are applied. You automate the response to some recurring incidents. You try AIOps, using AI to assist operations, on specific cases and if the data allows it (see lesson 6). If the organization uses AI systems, you observe them too (see the GenAI method).

A RACI is a table that says, for each activity, who does what. Think of the seating plan at a wedding: written in advance, it avoids arguments on the day. Here is a starting point to adapt.

ActivityCIOCISOPlatform, SREDevelopmentBusiness
Define the observability strategyACRIC
Decide the annual budgetAIRIC
Instrument servicesCIRAI
Define a service’s SLOsCICRA
Lead the response to a major incidentICARI
Report an incident to the authorityCARII
Produce management reportingACRIC
Steer the roadmapACRCI

R: responsible, does the work. A: accountable, approves, only one per row. C: consulted before the decision. I: informed after the decision.

Two rules avoid most drift. The first: only one A per row, because when two people are accountable, neither really is. The second: a review every quarter, because the organization changes. The interactive RACI matrix checks these rules and shows the load on each role.

Self-managed, SaaS or hybrid: decide with the same criteria

Section titled “Self-managed, SaaS or hybrid: decide with the same criteria”

The choice is a bit like buying a house, renting a furnished flat or doing a bit of both. Owning takes upkeep and skills; renting costs a rent that follows your usage; the mix means keeping an eye on both. No option is better in absolute terms. What separates them is your volume, your skills, your sovereignty constraints and the way you instrument.

CriterionSelf-managed open sourceSaaSHybrid
Upfront costinfrastructure and team timesubscription, little setup effortboth, on separate scopes
Cost over timemostly driven by people and storagemostly driven by billed volumeto be tracked on both lines
Skills requiredrunning the platform, on-calladministration, usage governanceboth, on a smaller scale
Time to first resultslongeroften shorterin between
Sovereigntyunder control if the hosting isdepends on hosting, applicable law and subcontractorsto be qualified per data flow
Dependencyon internal skillson the vendor and its pricingshared
Exit costlow if instrumentation is standard; dashboards and alerts to be rebuiltlow if instrumentation uses OpenTelemetry; high with proprietary agents and query languagesame, per scope

What weighs most on reversibility is not the commercial model. It is instrumentation. Think of electrical plugs: if all your appliances use a standard plug, you can switch electricity supplier without rewiring anything. Collection with OpenTelemetry plays that role and lets you change backend, the platform that stores and displays the data, without re-instrumenting the applications. Proprietary agents, whether they come from a vendor or an open source project, tie the organization to that backend.

The solution evaluation grid: six criteria

Section titled “The solution evaluation grid: six criteria”

To compare two or three options on a common basis, score each one from 1 (weak) to 4 (excellent) on six criteria, weighted according to your context. The interactive grid does the calculation and flags disqualifying scores.

CriterionQuestion to askWarning sign
Portabilitydoes the solution receive and export data in OpenTelemetry, in open formats?exclusive format, limited export
Exit costhow much does a migration cost: history, dashboards, alerts, integrations? Is it written into the contract?no exit clause
Data sovereigntywhere is the data stored, under which law, with which subcontractors?opaque location or subcontractors
Cost predictabilitydo the 12 and 36 month simulations hold on your realistic volumes?usage-based billing with no cap
Support and service levelsguaranteed response times, language, dedicated contact, active community for open source?no written commitment
Ecosystemintegrations with the CI/CD chain, ITSM, security; open and documented API?integrations limited to the same vendor’s products

eBPF: instrumenting without touching the code

Section titled “eBPF: instrumenting without touching the code”

Picture a sensor fitted on a building’s main pipes rather than in every flat: it sees the water go by without anyone opening the walls. eBPF works on that principle. It is a technology in the Linux kernel, the core of the operating system, that runs small verified programs at precise crossing points (system calls, network, functions). In observability, it captures HTTP, gRPC, TCP or DNS traffic without modifying applications. Well-known projects include Cilium and Tetragon for networking and security, Pixie for Kubernetes, plus Beyla, whose code Grafana Labs donated to OpenTelemetry in 2025.

For the CIO, eBPF lowers the cost of instrumentation and shortens the time before the first data arrives, especially on applications that cannot be modified. In return, it requires a recent Linux kernel and specific skills, to be checked tool by tool before a proof of concept.

An internal developer platform works like a well-equipped staff kitchen: each team helps itself without waiting to be served. Observability becomes a self-service offering there. Teams instrument their own services with ready-to-use templates, without waiting for the operations team. That requires shared OpenTelemetry standards and configurations versioned as code, which also brings auditability and reversibility.

Observability also serves frugality. It spots oversized services, inefficient queries, duplicate processing. It has a footprint of its own, hence a simple rule: every metric collected must serve an identified purpose, whether an alert, a dashboard or a decision. If your company publishes a sustainability report, infrastructure metrics can feed its energy section. The scope and timeline of the CSRD were revised by the EU in 2025 and 2026: check them for your case.

These are mistakes that turn up from one organization to the next, each with its remedy.

First, the empty dashboard: tools installed without SLOs or business indicators. The remedy is to start from the five indicators management expects, then link them to technical measurements. The invisible project is the same problem seen from finance: with no business case, it looks like a cost with no return and gets cut at the first budget review. The remedy is lesson 1.

The big bang means instrumenting everything at once; the phased roadmap and quick wins avoid it. The ghost RACI is when everyone thinks someone else is handling it: you need a RACI signed in phase 2 at the latest, reviewed every quarter.

Firefighting compliance discovers its obligations at the first inspection. You avoid it by building evidence capability into the roadmap (see lesson 2). That leaves the tool before the strategy: choosing a solution before defining the needs. The right order is strategy, then needs, a shortlist, a proof of concept and finally the decision.

Allow 45 minutes, in pairs or in a small group.

  1. Place your organization on the grid below, then on the site’s maturity self-assessment.
  2. Identify two or three quick wins achievable in under six weeks.
  3. Build phases 1 and 2: objectives, main deliverable, teams, budget, dependencies, quantified success indicator.
  4. Fill in the RACI matrix with your actual roles and spot the rows that spark debate.
  5. Compare two options in the evaluation grid.
DimensionLevel 1Level 2Level 3Level 4
Visionnoneimplicitdocumented and sharedaligned with group strategy
Budgetscatteredidentifiedstructuredmulti-year
Governanceno RACIinformal rolessigned RACIRACI reviewed every quarter
Compliancenothingfirefighting modebuilt into the roadmapevidence produced continuously
Communicationno reportingoccasionalmonthly dashboardregular review by the executive committee
Toolingdisconnected toolspartially integratedunified OpenTelemetry collectionconfiguration managed as code

The lowest dimensions point to the priorities. An organization at level 3 on tooling and level 1 on governance has invested in technology without organizing responsibilities: the RACI comes first.

To finish, write down your next action: which one, by what date, with whom?