Advances, Systems and Applications
From: Online architecture for predicting live video transcoding resources
Parameter | Description |
---|---|
Video files | 20 popular unseen video files from YouTube. Resolution: 1920 × 1080, 24 FPS, Video format: MKV |
Prediction algorithms | RF, RL, SGD |
Reference | vCPU request = vCPU limit |
Online model training | Every 30 min |
Target | Transcoding speed > 0.98 or 0.99; CPU consumption < 80 or 70% |
vCPU request | 45%*vCPU limit for VM with 4 vCPU, 52,5%*vCPU limit for VM with 6 and 8 vCPU, 57,5%*vCPU limit for VM with 16 vCPU |