A live digital replica
of every rack you operate.
Datafabrix Twin is the digital-twin module of the platform. It creates a continuously synchronised, behaviourally accurate replica of every rack, zone, and site in your fleet — enabling what-if simulation, change-impact preview, and capacity planning that is grounded in your actual operational data.
Digital Twin — engineered for AI-class workloads.
A living, continuously-synchronised replica of your fleet. Simulate before you ship.
The problem we solve.
Every infrastructure change is a calculated bet. New workload mix. New cooling setpoint. New rack topology. New tenant. Engineers and architects have to estimate the impact — and then ship it to production hoping the estimate holds.
Twin removes the hope. It runs the change in a simulation grounded in your real fleet's telemetry — thermal behaviour, power draw, signal integrity, failure profiles — and shows you the predicted outcome before you commit a single byte to production.
Capacity planning is no longer 'what's our average utilization plus 30%'. It's 'here's exactly how the next 18 months of growth fits, exactly how the cooling holds, exactly when we have to procure'. Twin turns infrastructure planning from an art into an engineering discipline.
What Datafabrix Twin does.
- Continuously synchronised replica
Every device, every connection, every thermal junction — replicated in the twin and kept in sync with live telemetry. Not a snapshot. A living model.
- What-if simulation
Propose a workload change, a cooling setpoint change, a topology change. Twin runs it forward against your actual fleet behaviour and shows the predicted outcome.
- Change-impact preview
Before any production change ships, Twin renders the predicted impact on thermal envelope, power draw, SLA risk, and tenant performance — at scale, in seconds.
- Capacity planning with constraints
The next 18 months of growth, modelled inside the actual thermal and power envelope of your sites. No more spreadsheets. No more surprises.
- Failure-mode rehearsal
Run your disaster recovery plan against the twin before you ever need it for real. Validate your runbooks against actual fleet behaviour.
- Scenario testing for new tenants
Onboarding a large new customer? Twin simulates their workload pattern against current fleet state and tells you exactly where it should land.
Real scenarios. Real outcomes.
Three representative engagements that illustrate the kind of value Datafabrix Twin delivers in the field.
The Problem
The $40M cooling decisionAn operator is planning a $40M expansion of liquid cooling capacity. Without it, they cap at 35 kW per rack. With it, 60 kW. CFO wants confidence the investment is right-sized.
Our Approach
Twin runs the next 24 months of projected workload growth — actual tenant patterns, real seasonality, modelled new business — against both topology options. The liquid expansion clears the growth curve with 6 months of margin.
The Outcome
The investment ships with confidence. The CFO has a defensible model. The architects know exactly when the new capacity has to come online.
The Problem
New tenant placementA new strategic customer wants 200 GPUs and an SLA of 99.99%. Two zones look viable, but each has different thermal headroom and tenant-mix considerations.
Our Approach
Twin simulates the customer's actual workload profile (provided in pre-sales) across both zones for 60 days. Zone A throttles 4 hours/month under their pattern; Zone B clears with 12% headroom.
The Outcome
Customer is placed in Zone B. SLA is honoured. The pre-sales team turns the Twin output into a customer-facing performance guarantee.
The Problem
Disaster recovery, validatedAn operator has a documented DR runbook that 'should' shed 30% of workload to a backup site within 8 minutes. The runbook has never been fully exercised.
Our Approach
Twin replays the DR scenario against the live fleet model. The actual recovery time is 14 minutes — the runbook misses a dependency on a shared control-plane component.
The Outcome
Runbook fixed. DR exercise re-run in Twin until it consistently clears 8 minutes. Compliance and operations both get the proof.
Datafabrix Twin works best with...
Ready to see Twin in action?
Tell us about your fleet and your top operational pain. We will map Datafabrix Twin to a 90-day pilot scope — and quantify the expected outcome — within five business days.