System Requirements - Transcoding Server
Below is a rough guideline for server hardware requirements for youra RTMPtranscoding server
For 10 Concurrent Streams
CPU:
•26–8 cores- This gives headroom to run roughly 20 concurrent encoding tasks (
modern2dual‑coreperatstream)~2.5+ GHzplusshouldmanagebethesufficient)passthrough.
- This gives headroom to run roughly 20 concurrent encoding tasks (
RAM:
•48 GB minimum (816 GBrecommendedrecommended)for- Ensures
bufferingsmooth operation with multiple encoder processes andsystemOSoverhead)overhead.
extra- Ensures
Network:
• 1 Gbps NIC- Estimated
Bandwidth:Ingress:~40 Mbps total (10streams× 4Mbps)Mbps ≈ 40 Mbps PlentyEstimatedofEgress:headroom10is×available(1.5with+a2.5 + 4) ≈ 80 Mbps- The 1 Gbps link easily covers this with room for spikes.
- Estimated
For 100 Concurrent Streams
CPU:
•4–816–32 coresForEachingeststreamonly,produces 2 CPU-intensive transcoding jobs; aquad‑larger core3.0+ GHz processor may sufficeIf any processing (e.g., repackaging)count isadded,critical.leanThetowardhigher8rangecoresis advised for pure software encoding.
RAM:•8 GB minimum(16 GB recommended to comfortably handle buffering, connection management, and OS overhead)Network:•1 Gbps NIC might be borderlineif streams are high qualityEstimated Bandwidth:~400 Mbps total (100 streams × 4 Mbps)For extra reliability and headroom, consider a10 Gbps NICor NIC bonding
For 1,000 Concurrent Streams
CPU:•8–16 coresFor pure ingest, 8 cores might work if optimizedIf you perform any transcoding or heavy processing, 16+ cores are recommended
RAM:
• 16 GB minimum (32 GBrecommendedrecommended)to- More
higherconcurrentbuffering,encodingconnectiontasksmanagement,willandbenefitanyfromadditionalextraprocessingmemory.
accommodatetasks)- More
Network:
• 10 Gbps NIC (oraggregated/multipleaggregatedNICs)connections)- Estimated
Bandwidth:Ingress:~4 Gbps total (1,000 streams100 × 4Mbps)Mbps ≈ 400 Mbps AEstimated Egress: 100 × ~8 Mbps ≈ 800 Mbps- Although calculated bitrates are below 1 Gbps, using a 10 Gbps
connectionNIC providesthe necessaryample headroom for overhead, bursty traffic, andstabilitypotential increases in bitrate if you choose higher-quality settings.
- Estimated
KeyAdditional Considerations
-
oftenProcessingHardwareLoad:Acceleration:TheIfaboveyourecommendationscanassumeuseminimalGPUs or dedicated encoding hardware (NVENC aka NVIDIA), you can significantly reduce CPUloadrequirements.perForstreamlarge-scale(i.e.transcoding,simplethisingress).is a - more
Networkcost‑effectiveOverhead:Real‑worldandconditionsenergy‑efficient(protocol overhead, burstiness, etc.) might push bandwidth requirements higher. It’s wise to over‑dimension network capacity relative to the calculated total.approach. -
Scalability:
InForproduction,very high concurrency (hundreds to thousands of streams), consider a multi‑server or cloud‑based transcoding farm that distributes the loadbalancingratheracrossthanmultiplerelyingserversonifayousingleexpectbox.to -
approachEncoding Settings & Quality:
More aggressive encoding quality settings will increase CPU load. Tailor theselimits,recommendations based on your specific quality versus resource trade‑offs. -
Redundancy & Future Growth:
Always plan with extra headroom for unexpected spikes andensurefuturemonitoringscalingto adjust resources as needed.needs.
These guidelines provide a starting point to help you size your hardware. Actual requirements can vary significantly depending on your exact situation.