Compute Right-Sizing
Compute Engine instances GCP Recommender thinks are over-provisioned — current vs recommended machine type and projected monthly saving.
Same idea on another cloud
What this pane shows
Resources where GCP Recommender's machine-type recommender thinks you could save money by moving to a smaller size. Each row shows the current machine type, the recommended target and the projected monthly savings.
Key use cases
- Identify the highest-value rightsize candidates before a cost sprint.
- Validate that a workload's machine type still matches its current load.
- Generate a target list for batch automation that resizes off-hours.
- Quantify total rightsize savings opportunity before committing to the work.
How to embed
lp.js is the single-entry loader: it injects the shared LightPane runtime
(and sd-embed.css) once per page. Add this pane's own stylesheet and script;
the pane mounts itself wherever the matching <div> is in the DOM.
<!-- In <head> — this pane's stylesheet -->
<link rel="stylesheet" href="https://lightpane.io/_pane-tests/gcp/rightsize/rightsize.css">
<!-- Your access key (inline, or set window.lpEmbed.accessKey from your own JS) -->
<script>window.lpEmbed = { accessKey: "<your-access-key>" };</script>
<!-- Where the pane should appear -->
<div class="lp-gcp-rightsize-panel" data-lp-account="<your-account-alias>"></div>
<!-- Before </body> — shared loader + this pane's script -->
<script src="https://lightpane.io/embed/lp.js"></script>
<script src="https://lightpane.io/_pane-tests/gcp/rightsize/rightsize.js" defer></script>
Tier
Available on Builder and above.
Build notes
Live-wired to gcp_recommender_rightsize (Cloud Recommender
google.compute.instance.MachineTypeRecommender), region-scoped. The
recommender payload carries the machine-type change + projected saving; absolute
current/recommended cost and CPU/memory p99 utilisation aren't in it (they'd
need a Cloud Monitoring enrichment), so those only appear on the mock showcase.
Mock renders on unconfigured surfaces; live shows an honest empty state when the
project has no recommendations.