Interaction of teams in operations

1) Why

The iGaming platform is dozens of domains (Payments, Games/Core, Risk/KYC, Data, Infra/SRE, Support, Compliance). Without formalized interoperability, MTTR, CFR and operational risks rise. The goal is to turn disparate functions into a single operating system: predictable contacts, transparent queues, common signals and consistent priority.

2) Principles

1. SLO-first: joint solutions are tied to SLO/error budgets.
2. A single source of truth: common dashboards, uniform statuses and artifacts.
3. Clear boundaries and interfaces: each pair of commands has a described contract (OLA/Runbook/API).
4. Small batches and reversibility: changes through phicheflags/canary, fast rollback.
5. No blame - yes data: parsing on facts, improvements - a mandatory part of the cycle.
6. Minimum required privileges and SoD: role separation for sensitive operations.
7. Automate the routine, standardize the rest.

3) Roles and RACI (end-to-end)

Head of Ops/SRE Lead is the owner of the operational framework, KPI/KRI. A

Service Owners (Payments/Games/KYC/Data) - domain goals, changes, risk. A/R

Platform/Infra - accessibility, performance, releases/canaries. R

Risk/Compliance/Security - SoD, RG/KYC/PII, audits. C/A

Support/CRM - front of complaints, communication to players. R/C

On-call IC/CL - incident management and external updates. R

Release Manager - calendar, CAB, status of changes. R

Data/Analytics - product and operational metrics, RCA support. R/C

4) Interaction Contracts (OLA/SLx)

• Areas of responsibility: whose area (for example, PSP routing - Payments; cache/DB - Infra).
• Objectives/threshold metrics: Incident MTTA, escalation response time, post-monitoring window.
• Queues and priorities: P1-P4, business criticality, freeze windows.
• Interfaces: channels, bot commands, API/Runbook, owner directories.
• Artifacts: what documents/logs/dashboards are required to accompany the event.

5) Communication channels and protocols

Operational chat (shift): daily updates, mini-rituals, handover.
Var rooms for incidents: created by a bot; IC/CL roles are assigned by the command.
CAB/Change channel: discussion of changes, risks, release calendar.

Read-only SLO/Incident/Planned Activity summaries

Escalation: command templates '/page ', '/escalate', SLA reporting.

Unified message protocol: "fact → impact → ETA/ETR → next update window → owner."

6) Handovers between shifts and regions

1. SLO/SLI: where is the risk of budget burnout.

2. Open incidents/escalations and their ETAs.

3. Planned works/releases in the next 24-48 hours.

4. Providers (PSP/KYC/studios): active tickets, expectations.

5. On-call composition and contacts (IC/CL/domains).

6. "Watchlist" - areas of increased attention (queues/replication/cache).

Handover is recorded in a shift log, links - to var-rooms and dashboards.

7) Incident collaboration

Start: alert → bot creates a card '# inc-YYYY-MM-DD-XXX', assigns IC/CL and domain leads.
One vote rule: IC is the final decision; CL - communications.
Facts and hypotheses: we separate; "red" signals - priority.
Guardrails: phicheflags/PSP routing only change via runbook with SoD/dual-control.
Communications: drafts of public updates via CL, partners - targeted.
Closure: post-monitoring, post-mortem generation and improvement tasks with owners/deadlines.

8) Collaborate on changes

Release calendar: public, with freeze periods and on-call slots.
Quality gates: unit/contract/e2e, security, SLO gates staging.
Canary rolling: step by step 5%→25%→100% for GEO/tenants/banks.
Auto-rollback: policies by key SLI/KRI, WORM magazine.
Comm packages: draft updates agreed in advance with CL/Legal.
RACI changes: RM (A/R), SO (A/R), SRE (R), Sec/Compliance (C/A), CAB (A), IC/CL (R/C).

9) Unified telemetry and artifacts

Common metrics directory: SLI/SLO, business metrics, KRI (queues, PSP, replication).
Dashboard "Operation Map": summary by domains, regions, status of incidents/works.
Timelines: uniform format (time, author, action, result, links).
Post-mortems: template without charges, prevention measures, revision date.
Runbooks/Checklists: versioned; link from alerts and incident cards.

10) Prioritization and planning

Weekly Ops plan (30-45 min): coordination of top risks, releases, limits, improvements from post-mortems.
Kanban of operations: columns' Backlog → Ready → In Progress → Validate → Done ', WIP limits.
Priority criteria: impact on SLO/revenue/compliance, size/reversibility, dependence on providers.

11) Escalation matrix (squeeze)

12) Policies and SoDs

SoD/4-eyes: conclusions/bonuses/PSP routing/PII export - only with double approval.
JIT rights: temporary escalation of privileges for runbook actions.
Data policies: PII ban in open channels/dashboards; geo-boundaries.
Audit - immutable activity logs (WORMs), policy revisions.

13) Interaction tools

Incident bot: '/incident new ', roles, update timers, comm drafts, '/runbook', '/flag ', '/config'.
Metrics API: common SLO view and KRI, exemplars (trace_id) for RCA.
Release-portal: manifests, gates, rolling/rollback status.
Owners directory/CMDB: domains, contacts, backup channels.

14) Collaboration Metrics (KPI/KRI)

MTTA/MTTR by domain and slot (day/night), proportion of incidents caught before complaints.
Handover Quality: transmission defects (checklist items not closed on time).
Change Collaboration:% of releases with ready-made comm packages and no rollbacks.
Guardrail Discipline: frequency of SoD/policy violations (target 0).
Comms Cadence: adherence to public update intervals when P1/P2.
Post-mortem SLA: proportion of post-mortems ≤ D + 5, completion of actions.
Fair-share Load: Distribution of nights/peaks by people/teams.
Customer Signal Lead: Lag between objective degradation and first complaints.

15) Implementation Roadmap (6-10 weeks)

Ned. 1-2: domain/owner inventory; OLA templates; launch of the replacement channel and handover checklist; base escalation matrix.
Ned. 3-4: incident-bot (MVP), shared status channel, single SLO/SLI/KRI card; runbooks directory.
Ned. 5-6: CAB/release calendar, comm packages and freeze windows; SoD/4-eyes for sensitive operations.
Ned. 7-8: canary rolling and auto rollback as standard; post-mortem template, Exec/Ops-dashboards collaboration.
Ned. 9-10: P1 exercises, cross-regional handovers, WORM audits, KPI/KRI reports, OLA adjustments.

16) Templates (fragments)

16. 1 OLA (Payments ↔ Infra/SRE)

yaml ola:
scope: "Payments-Auth & Routing"
contacts:
payments_so: "@pay-so"
infra_oncall: "@sre-oncall"
objectives:
mtta_p1: "≤5m"
rollback_ttr: "≤10m canary"
interfaces:
runbooks: ["psp-failover", "reroute", "auth-throttle"]
dashboards: ["auth_success", "psp_latency", "queue_lag"]
escalation:
p1: ["IC","Payments Lead","SRE L2"]
p2: ["Payments OnCall","SRE OnCall"]
artifacts:
status_templates: ["public","partners"]
postmortem_due: "D+5"

16. 2 Handover checklist (10 items)

1. SLO domain statuses

2. Open Incidents (ETA/Owners)

3. Planned activities/releases + observation windows

4. Providers (PSP/KYC/Studios) - Risks/Expectations

5. Queues/replication/cache - lag/anomalies

6. Limit/Phicheflag Changes

7. Complaints/tickets and load thresholds

8. Comma plans and status drafts

9. On-call composition and reserve

10. "Watchlist" per slot

17) Antipatterns

"Will anyone deal?" without RACI and owner.
Incidents without IC/CL and update timers.
Hidden changes (manual clicks), no Git/Audit.
Non-common telemetry: different numbers in different teams.
Releases without comm packs and canaries.

SoD violations "for the sake of speed."

Handovers orally, without records and checklists.
Post-mortems without actions and deadlines.

Total

The interaction of teams in operations is a contractual collaboration: OLA/SLx, clear channels and roles, handover discipline, general telemetry, coordinated releases and incident processes. Such a framework reduces MTTR and CFR, aligns priorities, protects SLO, revenue and compliance - and makes daily operations predictable and sustainable.