Overview

Infrastructure at a glance.

Capacity, incidents and recovery checks across the selected operating boundary.

Updated just now

Services monitored4241 healthy

Compute instances126Across 2 regions

Open incidents1Action required

Restore checks due3This week

Service state

Production services

ServiceRegionStateOwner

Commerce APIMumbaiHealthyOps

GPU inferenceMumbaiReviewAI Ops

Agency clusterUS WestHealthyOps

Backup verificationUS WestDueRecovery

Capacity

Allocation

Compute72%

Memory64%

Storage48%

GPU81%

Recent activity

Operating log

Maintenance completedCommerce API · 22 min ago
GPU utilisation reviewInference cluster · 1 hr ago
Backup completedAgency cluster · 3 hr ago

Infrastructure

Compute and services.

Review inventory, ownership and the operating state of each environment.

Inventory

Active resources

ResourceTypeRegionConfigurationState

commerce-prod-01Cloud VMMumbai8 vCPU · 32 GBHealthy

gpu-inference-aGPU nodeMumbai4 × L40SReview

agency-web-03DedicatedUS West32 cores · 256 GBHealthy

recovery-targetBackupUS West24 TB usableCheck due

Incidents

One active incident.

Named ownership and an explicit handoff path for every production event.

P2

INC-2026-0823-01 · Investigating

Elevated GPU inference latency

Latency crossed the review threshold on the Mumbai inference pool. Capacity and queue depth are under review.

Owner: AI Ops
Started: 18 minutes ago
Next update: 12 minutes

History

Recent incidents

Backup verification delayedResolved · 2 days ago

Commerce API memory pressureResolved · 5 days ago

Agency cluster deployment holdResolved · 8 days ago

Backups

Recovery needs evidence.

Track completed jobs, verification state and the next restore exercise.

Protected workloads34Last job completed

Restore checks due3This week

Failed jobs0No action required

Verification plan

Upcoming restore checks

Commerce databaseDue tomorrow · Mumbai

Agency application setDue in 3 days · US West

GPU model storeDue in 5 days · Mumbai

GPU Capacity

Capacity by card.

Review mock allocation data. Availability is confirmed before any customer order.

L40S · 48 GB12 cards

9 allocated · 3 review capacity

A100 · 80 GB8 cards

6 allocated · 2 review capacity

H100 · 80 GB4 cards

4 allocated · quote required

Workloads

Current allocation

WorkloadGPURegionUtilisation

Inference API4 × L40SMumbai82%

Fine-tuning queue2 × A100Mumbai61%

Image generation2 × L40SUS West74%