cost service v13: lower per-VIN estimates to 3mc/10MB

- Per-VIN marginal: 3 millicores / 10MB (was 50mc/80MB)
- Based on ~3000 VINs using ~10 cores / 30GB marginal on top of platform base
- Total fleet now shows realistic: 87 cores / 569GB for 2463 VINs
- Updated README with v13 output
This commit is contained in:
Chris Rai
2026-02-05 12:16:22 -05:00
parent a813a51696
commit 87f80d356a
3 changed files with 25 additions and 22 deletions

View File

@@ -17,7 +17,7 @@ spec:
spec:
containers:
- name: cost
image: fiskercloud.azurecr.io/cost:v12
image: fiskercloud.azurecr.io/cost:v13
imagePullPolicy: IfNotPresent
ports:
- containerPort: 8077

View File

@@ -39,9 +39,11 @@ Incremental resources needed for each additional connected vehicle:
| Activity Level | Messages/15min | CPU (millicores) | Memory (MB) |
|---------------|----------------|------------------|-------------|
| Low | < 100 | 50 | 80 |
| Medium | 100-1000 | 75 | 120 |
| High | > 1000 | 100 | 160 |
| Low | < 100 | 3 | 10 |
| Medium | 100-1000 | 5 | 15 |
| High | > 1000 | 6 | 20 |
*Based on ~3000 VINs using ~10 cores / 30GB marginal on top of platform base*
### Cost Rates
@@ -60,7 +62,7 @@ Incremental resources needed for each additional connected vehicle:
```
Platform Base Cost = (80 cores × rate + 544 GB × rate) × hours
Per-VIN Cost = (0.05 cores × rate + 0.08 GB × rate) × hours × activity_multiplier
Per-VIN Cost = (0.003 cores × rate + 0.01 GB × rate) × hours × activity_multiplier
Total Cost = Platform Base + (Per-VIN × VIN count) + Managed Services
Cloud Cost = Total with cloud rates
@@ -121,20 +123,20 @@ curl http://localhost:8077/cost/report
╠══════════════════════════════════════════════════════════════════╣
║ FLEET OVERVIEW ║
║ ─────────────────────────────────────────────────────────────── ║
║ Active Vehicles: 3301
║ Cloud Cost: $10662.95
║ On-Prem Cost: $770.97
║ Savings: $9891.98 (92.8%)
║ Active Vehicles: 2463
║ Cloud Cost: $695.52
║ On-Prem Cost: $62.55
║ Savings: $632.97 (91.0%)
╠══════════════════════════════════════════════════════════════════╣
║ RESOURCE USAGE MODEL ║
║ ─────────────────────────────────────────────────────────────── ║
║ Platform Base: 80 cores / 544 GB RAM (fixed)
║ Per-VIN Marginal: 50 millicores / 82 MB RAM
║ Total Fleet: 245.1 cores / 808.1 GB RAM
║ Per-VIN Marginal: 3 millicores / 10 MB RAM
║ Total Fleet: 87.4 cores / 568.6 GB RAM
╠══════════════════════════════════════════════════════════════════╣
║ COST FORMULA ║
║ ─────────────────────────────────────────────────────────────── ║
║ (Platform Base) + (Per-VIN × 3301 VINs) + Managed Services
║ (Platform Base) + (Per-VIN × 2463 VINs) + Managed Services
╠══════════════════════════════════════════════════════════════════╣
║ COST RATES ║
║ ─────────────────────────────────────────────────────────────── ║
@@ -144,19 +146,19 @@ curl http://localhost:8077/cost/report
╠══════════════════════════════════════════════════════════════════╣
║ ANNUAL PROJECTION (based on current usage) ║
║ ─────────────────────────────────────────────────────────────── ║
║ Cloud Annual: $127955.46
║ On-Prem Annual: $9251.64
║ Annual Savings: $118703.82
║ Cloud Annual: $8346.25
║ On-Prem Annual: $750.61
║ Annual Savings: $7595.65
╚══════════════════════════════════════════════════════════════════╝
TOP COST VEHICLES:
VIN CPU (mc) RAM (MB) Cloud $ On-Prem $ Savings %
─────────────────── ──────── ──────── ────────── ────────── ────────
VCF1EBU24PG007242 670 1032 15.36 1.17 92.4%
VCF1UBU21PG008884 670 1032 15.36 1.17 92.4%
VCF1ZBU29PG006267 694 1067 15.13 1.15 92.4%
VCF1EBU26PG007307 683 1051 14.48 1.11 92.4%
VCF1EBU22PG011385 676 1040 13.93 1.03 92.6%
VCF1UBU21RG013084 94 149 1.20 0.12 90.4%
VCF1EBU21RG012448 94 149 1.20 0.12 90.4%
VCF1EBU22PG011385 94 149 1.20 0.12 90.4%
VCF1EBU24PG011467 94 149 1.20 0.12 90.4%
VCF1EBU29PG007298 94 149 1.20 0.12 90.4%
...
```

View File

@@ -24,8 +24,9 @@ const (
// Per-VIN resource usage (marginal cost per vehicle)
// This is the incremental CPU/RAM needed for each additional connected vehicle
// Covers: telemetry ingestion, Kafka processing, storage writes, API queries
PerVinCPUCores = 0.05 // 50 millicores per VIN (marginal)
PerVinMemoryGB = 0.08 // 80MB per VIN (marginal)
// Based on: ~3000 VINs using ~10 cores / 30GB marginal on top of platform base
PerVinCPUCores = 0.003 // 3 millicores per VIN (marginal)
PerVinMemoryGB = 0.01 // 10MB per VIN (marginal)
// Platform base resources (fixed cost regardless of VIN count)
// Based on migration plan production environment: