What goes in an IT infrastructure dashboard in Metabase?

An IT infrastructure dashboard is the "what do we run, and where's the headroom?" page: asset inventory by type and age, virtualization CPU/RAM headroom, storage capacity trends, network utilization, and the licenses and warranties about to lapse. Where an IT monitoring dashboard watches the estate minute by minute, this one plans it quarter by quarter — Metabase builds it from CMDB, hypervisor, and storage snapshots in your warehouse.

What does an IT infrastructure dashboard look like?

Here’s the layout this guide builds. Inventory leads so the estate’s shape is established first; capacity and headroom sit in the middle — cluster allocation, storage growth, network utilization — because that’s the part that changes purchase decisions; lifecycle and renewals close the page, where expiring licenses and aging hardware queue up next year’s spend.

Which cards belong on an IT infrastructure dashboard?

The eight below cover the three planning questions in order: what exists, how much room is left, and what expires next.

• Assets by type — servers, network, storage, endpoints (donut)
• Asset age distribution, in years since purchase (bar)
• Cluster headroom — CPU and RAM allocated versus physical, per virtualization cluster (row)
• Storage used versus capacity, weekly trend per tier (area)
• Storage pool headroom — used against total capacity (progress)
• Network utilization on core links, busy-hour percent (line)
• License renewals in the next 90 days, with seats and cost (table)
• Assets aging out of warranty or EOL by quarter (bar)

What data does the dashboard need?

• assets — CMDB export with asset ID, type, model, site, purchase date, warranty_end, and eol_date.
• vm_hosts and virtual_machines — hypervisor inventory with physical cores and RAM per host, and vCPU/RAM allocations per VM.
• storage_snapshots — per-pool capacity and used bytes, one row per day, tagged by tier.
• link_utilization — per-link capacity and busy-hour throughput from SNMP polling.
• licenses — product, seats, renewal date, and annual cost from procurement or the SAM tool.

How do you build it?

1. Schedule nightly snapshots from the CMDB, hypervisor API, storage arrays, and SNMP into the warehouse — capacity cards chart trends, so append rather than overwrite.
2. Join hypervisor and CMDB data on asset ID and fix the mismatches first; every headroom number inherits its accuracy from this join.
3. Build a cluster-headroom model computing allocated versus physical CPU and RAM per cluster (the SQL below), and point the row card and KPI scalars at it.
4. Compute days-to-full per storage pool from 60–90 days of growth, and chart used capacity against total per tier.
5. Add filters for site, asset type, and date range, and subscribe the infrastructure team monthly for capacity reviews.

Example card SQL

SELECT
h.cluster,
COUNT(DISTINCT h.host_id)                          AS hosts,
SUM(h.cpu_cores)                                   AS physical_cores,
SUM(v.vcpus_allocated)                             AS vcpus_allocated,
ROUND(100.0 * SUM(v.vcpus_allocated)
  / NULLIF(SUM(h.cpu_cores), 0), 1)                AS cpu_alloc_pct,
SUM(h.ram_gb)                                      AS physical_ram_gb,
SUM(v.ram_gb_allocated)                            AS ram_allocated_gb,
ROUND(100.0 * SUM(v.ram_gb_allocated)
  / NULLIF(SUM(h.ram_gb), 0), 1)                   AS ram_alloc_pct,
SUM(h.ram_gb) - SUM(v.ram_gb_allocated)            AS ram_headroom_gb
FROM vm_hosts h
LEFT JOIN (
SELECT
  host_id,
  SUM(vcpus)   AS vcpus_allocated,
  SUM(ram_gb)  AS ram_gb_allocated
FROM virtual_machines
WHERE state = 'running'
GROUP BY host_id
) v ON v.host_id = h.host_id
WHERE h.is_active
GROUP BY h.cluster
ORDER BY ram_alloc_pct DESC;

Related

Metrics

Integrations

Dashboards

FAQ