Specifications
Technical & Operational
Specifications.
A clean surface for decision-makers, with deeper layers for engineering and operations stakeholders.
Summary
The platform on one sheet.
Every parameter that governs an engagement, with the section that specifies it. Figures are standards, not promises; contractual commitments are set in the enterprise agreement.
Specification summary
CloudPresto managed platform
Availability standard
99.9% for managed infrastructure, measured monthly, excluding agreed maintenance windows
sla →
Operations coverage
24/7/365 monitoring with defined escalation; the monitoring stack is itself highly available
service →
Severity model
P1 to P4, with a triage priority, update cadence and escalation rule per level
sla →
Change governance
Standard, normal and emergency change types; agreed windows; a rollback path on every change
service →
Recovery targets
Tiered by workload criticality, from a 15-minute recovery target with near-zero data loss on Tier 1 to a 24-hour target on Tier 4
resilience →
Replication
Multi-region; synchronous, asynchronous or semi-synchronous by tier
resilience →
Recovery validation
Quarterly: failover drill, restore validation, component-failure simulation, full DR exercise
resilience →
AI thresholds
95% and above executes with logging; below 80% is mandatory human review; destructive actions never automatic
ai →
Review cadence
Weekly operational summary, monthly service review, quarterly business review
sla →
Onboarding
Discovery, architecture workshop, controlled rollout, steady state; acceptance criteria at every phase
onboarding →
Contractual SLAs
Time-bound response and escalation targets, available as part of an enterprise agreement
sla →
Service model
What is operated, and how the engagement is governed.
The managed service covers the full operational lifecycle: provisioning, hardening, patching, triage, optimization and governance, not monitoring alone.
Operational scope
24/7 platform operations
Round-the-clock coverage with defined escalation paths and response expectations formalized per engagement.
Incident triage and escalation
P1 incidents triaged on priority. Structured severity classification, a communication cadence, and a post-incident review for every event.
Patching and lifecycle
Governed patch lifecycle with impact analysis, staged rollouts, and rollback gates. Change windows agreed with stakeholders.
Capacity and optimization
Proactive capacity planning, right-sizing, and cost optimization. Trend analysis finds scaling needs before they become performance problems.
Engagement controls
01
Consultative onboarding
Discovery, architecture workshop, phased rollout. No template deployments.
02
Workload criticality mapping
Every workload carries a criticality tier that drives its targets, monitoring depth and recovery priority.
03
Change governance
Risk-assessed, approval-gated workflows for standard, normal and emergency changes.
04
Business reviews
Weekly operational summaries, monthly service reviews, quarterly business reviews. Metrics aligned to business impact.
SLAs
Severity, availability, cadence.
How incidents are classified and communicated, what the availability standard covers, and when you hear from us. Time-bound response and escalation targets are set contractually in the enterprise agreement.
Severity model
Applies to every managed engagement
| Severity | Definition | Triage priority | Update cadence | Escalation |
|---|---|---|---|---|
| P1 · Critical | Service down, data at risk, revenue impact | Top priority | Every 30 minutes | Automatic |
| P2 · High | Degraded performance, partial impact | High priority | Every 2 hours | Automatic |
| P3 · Medium | Non-critical, workaround available | Standard | Daily | On request |
| P4 · Low | Improvement or cosmetic issue | Scheduled | Weekly | On request |
Availability standards
Platform availability
99.9% uptime standard for managed infrastructure. Measured monthly, excluding agreed maintenance windows. Availability reports included in monthly service reviews.
Monitoring coverage
24/7/365 monitoring with no gaps and defined handoff procedures. The monitoring stack is itself highly available, with independent health checks.
Change windows
Routine maintenance during agreed windows, typically weekly and off-peak. Emergency changes expedited with stakeholder notification. Every change carries a rollback path.
Communication
Incident notifications on detection. Status updates at the cadence defined per severity. A post-incident review and resolution summary after every P1 and P2.
Reporting cadence
Weekly
Operational summary: incidents, changes, patch status, key metrics.
Monthly
Service review: availability, SLA adherence, capacity trends, cost analysis.
Quarterly
Business review: risk posture, strategic recommendations, roadmap alignment.
Resilience
Continuity, by tier, proven on a calendar.
Every resilience engagement is designed for survivability, recoverability and controlled restoration under disruption. Targets follow the workload tier; drills prove them four times a year.
Recovery targets by tier
Assigned per workload at onboarding
| Tier | Recovery time target | Data loss target | Replication model |
|---|---|---|---|
| Tier 1 · Critical | Under 15 minutes | Near zero | Matched to the tier |
| Tier 2 | Under 1 hour | Set per workload | Matched to the tier |
| Tier 3 | Under 4 hours | Set per workload | Matched to the tier |
| Tier 4 | Under 24 hours | Set per workload | Matched to the tier |
Continuity architecture
Multi-region replication
Data and services replicated to geographically separate regions. The replication model is matched to the workload tier and its data-loss target.
Blast radius containment
Failures contained within defined boundaries: bulkhead isolation, circuit breakers, graceful degradation.
Automated failover
Quorum-based failure detection with automated triggers. DNS updates, connection rerouting and health verification automated, with a person notified.
Recovery validation
Every quarter
Q1
Failover drill
Controlled failover in a maintenance window. Recovery time measured, data integrity verified, gaps documented and remediated.
Q2
Backup restore validation
Full restore from backup into an isolated environment. Application tested, restore time measured against the tier target.
Q3
Component failure simulation
Targeted failure injection: database node, network partition, credential revocation. Validates isolation and graceful degradation.
Q4
Full DR exercise
Combined scenarios, end to end. Findings documented and reported, remediation tracked to close.
AI workflow
Scoped, grounded, governed.
How AI is scoped and governed end to end: guided qualification, retrieval-grounded context, and decisions a person is accountable for.
Workflow capabilities
Guided qualification
Multi-step intake with structured requirement capture, routing to the right service path, and a qualification brief.
Context packaging
Retrieval from the curated knowledge base. Every response traceable to its source, with a confidence score.
Governed decisions
Human override on every action. Full audit trail. Confidence thresholds decide what executes and what a person reviews.
Governance
Enforced in the workflow definition
Automatic execution
Confidence of 95% and above, with full logging
Mandatory human review
Confidence below 80%
Destructive actions
Never without human approval, enforced in the workflow definition
Audit trail
Inputs, sources, recommendation, confidence, decision and outcome retained per compliance policy
Calibration
Accuracy reviewed against real outcomes at every review
Knowledge base
Refreshed after service, pricing or procedure changes
Onboarding
Four phases, each with acceptance criteria.
Nothing moves to the next phase until the current one is accepted. Durations are typical and scoped to the estate.
Phases and deliverables
| Phase | Typical duration | Deliverables | Acceptance |
|---|---|---|---|
| Discovery | Weeks 1 to 2 | Workload inventory, dependency map, risk assessment, criticality tiers | All in-scope workloads documented with criticality scores |
| Architecture workshop | Week 3 | Operating model, escalation matrix, monitoring baselines, change governance rules | Stakeholder sign-off on the operating model |
| Controlled rollout | Weeks 4 to 10 | Monitoring deployment, triage activation, patch management, governance | Each phase passes acceptance before the next begins |
| Steady state | Week 10 onward | Full 24/7 operations, reporting cadence, optimization cycle | First monthly service review completed |
What we need from you
Access
Read-only cloud account access for discovery. Roles defined in the workshop. Standing operational access under accountable stewardship, every session logged and attributable.
Stakeholders
A technical lead (3 to 5 hours a week during rollout) and a business stakeholder (1 to 2 hours a week) for design sessions, reviews and approvals.
Documentation
Existing runbooks, architecture diagrams and incident history. Completeness not required; discovery fills the gaps.
Communication
A shared Slack or Teams channel, and PagerDuty or equivalent for incident escalation.
Request
The full specification, scoped to your estate.
Detailed technical documentation is provided during qualification, scoped to your environment and the standards you are held to.
What the full specification adds
01
Your workloads, tiered
Criticality, targets and monitoring depth per workload.
02
The operating model, written down
Ownership boundaries, escalation matrix, change governance, review cadence.
03
The contractual terms
Time-bound response and escalation targets in the enterprise agreement.