They Deleted 93% of Their Grid Data
A major power utility had tens of thousands of sensors. After 7 days, the data got deleted - not because they wanted to, but because storing petabytes at cloud rates would bankrupt the department. Engineers learned to diagnose problems fast or not at all.
Evidence boundaryHistorical case-specific report. Figures describe the deployment documented at the time and are not current universal product guarantees.
Client
Major Power Utility
Industry
Energy & Utilities - Smart Grid
Use Case
Substation Indexing & Air-Gapped Analytics
Products
Timeline
Pilot in 8 weeks, full rollout in 5 months
ROI
12-month payback on local-storage investment
/ Challenge
The Challenge
The utility's grid operations team had a problem. Every week, engineers asked for sensor waveforms from 8 days ago. Every week, IT had to explain it was already gone. The 7-day retention limit wasn't a policy choice - it was a budget reality. Cloud storage for petabytes of high-frequency sensor data would cost more than the entire IT budget. Worse, regulatory rules prohibited sending raw customer-meter data off-site without anonymization.
Tens of thousands of sensors at thousands of samples per second
7-day retention limit - faults older than a week had no waveform evidence
Cloud egress quoted in the millions per year - more than the entire bandwidth budget
Critical infrastructure compliance prohibits raw operational data on the public internet
Customer meter data contains PII; manual redaction doesn't scale
No way to search waveforms - you had to know exactly when and where to look
/ Solution
The Solution
The deployment put analysis at the substation. Each substation cluster indexed voltage anomalies, frequency deviations, equipment signatures, and timestamps. Raw waveforms stayed on local storage while configured metadata and event outputs moved upstream.
Substation-Level Indexing
Each substation cluster runs anomaly detection and event classification for nearby sensors. Instead of storing petabytes of raw waveforms, we store gigabytes of metadata and flagged events. That's a 99% reduction in bandwidth and storage.
Local Long-Term Storage
Raw sensor data stays on local NAS at each substation. $0.003/GB/month locally versus $0.023 in the cloud. We went from 7-day retention to 5 years for less money. Engineers search the index and retrieve only the windows they need.
Customer-Defined Transformations
Configured rules selected and transformed records at the substation before approved outputs left the local network. Processing records remained available for customer review.
/ Results
The Results
Engineers stopped getting refused when they asked for week-old waveforms. Investigation time dropped from days to hours. Operators stopped watching raw streams and started responding to indexed alerts. And the budget actually went down.
99%
Data Reduction
5 years
Retention
Policy-bound
Configured Outputs
$1.8M
Annual Savings
Sensor retention extended from 7 days to 5 years
99% bandwidth reduction - metadata flows upstream, raw waveforms stay local
$1.8M annual savings vs. projected cloud architecture
Customer-defined transformations limited which records left each substation
Processing records supported the utility's own review process
Search returns relevant time windows in seconds across 5 years of history
Pilot validated in 8 weeks, full rollout in 5 months
/ Explore Related Solutions
/ What's next
Deleting data you might need?
If your retention limits are budget-driven, not policy-driven, we can evaluate a local-processing design against your requirements.
/ Typical engagement
30 days
Free assessment
90 days
Optional pilot
Measured
Baseline vs. after
