Compute resources
Select infrastructure resources for the model-serving workload.
Service · Inference Infrastructure · Gruban
Infrastructure for inference
Connect model inference workloads with the compute, data center, and network infrastructure they require.
The infrastructure beneath inference
Gruban provides infrastructure for inference workloads. Prisman serves the inference product through its API or a custom contract; Gruban’s role is the infrastructure layer beneath the workload.
Infrastructure services
Relate compute to model demand
Infrastructure requirements depend on where the workload runs, how it connects, and what capacity it needs.
Deployment placement
Choose a location and environment suited to the operating scope.
Network connections
Connect inference workloads to the applications and services that use them.
Inference Infrastructure
Keep the service layers distinct
Gruban provisions infrastructure; Prisman provides the inference service consumed by applications and Agents.
01
Gruban infrastructure
Bare Metal, data center, and network resources support the workload environment.
02
Prisman inference
Prisman provides inference through API usage or a custom contract.
03
Workload placement
Discuss deployment location and infrastructure requirements with Gruban.
Service · Inference Infrastructure
Plan infrastructure for inference
Contact Gruban about the compute and network environment supporting your inference workload.