In high-frequency microservice meshes (e.g., Istio with successThreshold=1, periodSeconds=1), a standard health endpoint can consume 3-5% of CPU per pod solely for JSON serialization and dependency checks. FujizakuraWorks identified that existing health libraries treat readiness/liveness as cold, infrequent events. PHC v1.0 inverts this assumption: health checks are hot, continuous validation signals.
Environment: c5.large (2 vCPU, 4GB RAM), Linux 6.1, 10 concurrent probe requests.
| Framework | p99 Latency (µs) | Alloc/op (bytes) | Binary Size |
|-----------|------------------|------------------|--------------|
| Spring Boot Actuator 3.2 | 1240 | 2048 | 28 MB (JAR) |
| Go health pkg (std) | 380 | 256 | 6.2 MB |
| PHC v1.0 (Concise mode) | 78 | 0 | 212 KB |
| PHC v1.0 (Verbose JSON) | 95 | 0 | 212 KB + 4KB buffer | petitehealthcheckv10byfujizakuraworks hot
Load test (5000 pods → per-pod health checked every 1s):
True to its name, the PetiteHealthCheckV10 fits in a pocket. Key specs: In high-frequency microservice meshes (e
The device syncs via Bluetooth 5.3 and works offline — data can be stored for up to 90 tests before syncing.
As of May 2025, the PetiteHealthCheckV10 is sold: The device syncs via Bluetooth 5
Price: ¥22,000 JPY (~$145 USD) + shipping.
"Hot" drops happen every 2–3 weeks, with units selling out in 10–15 minutes.
“Most health checks take 20 minutes and tell you to drink water.
Petite Health Check v10 takes 90 seconds and tells you when your team will crash.
Built by Fujizakura Works — for humans who move fast.”