How do I go about estimating cost of self hosted livekit for only as a transport layer (livekit server, livekit sip) on AWS? Is there a unit level estimate or any process available to do that?

Thanks for reaching out, Vipul. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

I will have my own SIP trunk that terminates on my AWS cluster and livekit components will be deployed in the same cluster.

Thanks — that helps. Since your SIP trunk terminates in the same AWS cluster, the estimation will focus on AWS egress, EC2/SIP compute, and any intra‑cluster networking. To provide a concrete cost estimate I need these details: • Expected average video bitrate per outbound stream (e.g., 1.7 Mbps for 720p) and audio bitrate (e.g., 64 kbps). • Number of simultaneous viewers/participants receiving streams (concurrent downstreams). • Average session length (minutes or hours) and sessions per day. • Whether you’ll use simulcast/SLI or multiple quality layers per stream. • AWS region(s) you’ll deploy in. • Estimated number of LiveKit server instances (or desired concurrency target so I can recommend instance sizing). • Any additional AWS services (NAT Gateway, ALB, EBS size, CloudWatch retention) to include. If you provide those, I’ll calculate: 1) total hourly and monthly egress (GB) and cost using region pricing, 2) EC2 compute cost estimate, 3) SIP trunking minutes estimate impact (if relevant), and 4) a combined monthly cost range and recommended instance suggestions. , with interactive elements

• This is for standard voice call in India. • Number of simultaneous viewers/participants receiving streams (concurrent downstreams).: 1k concurrent calls/rooms with 40 seconds AHT. Each room will have 2 participants. • this will be deployed in ap-south-1

Thanks — I’m responding to your scenario: 1,000 concurrent 2‑party voice calls (standard voice) in ap‑south‑1, 40s average call length, SIP trunk terminating in same AWS cluster. I’ll keep this concise and show the calculation steps so you can adjust numbers. Assumptions • Audio bitrate: 64 kbps per direction (adjust if you use different codec/bitrate). • Each call has 2 participants; LiveKit server will send one downstream audio stream per participant (so per call, 2 outbound streams total). • All SIP trunking terminates in‑cluster (no external egress for call media). Only AWS outbound egress billed is traffic leaving AWS to external participants; since calls terminate in‑cluster, we assume negligible external egress. If participants are remote and media leaves AWS, include egress calculation below. • Use ap-south-1 data transfer pricing (you should confirm exact per‑GB rate in AWS Pricing Calculator). I’ll use a placeholder rate of $0.09/GB as an example — replace with your actual regional rate. Bandwidth / egress calculation (if media egress from AWS) 1. Per‑stream bitrate = 64 kbps = 64,000 bps. 2. Bytes/sec = 64,000 ÷ 8 = 8,000 B/s. 3. Per‑stream per‑call (40s): 8,000 × 40 = 320,000 B ≈ 0.000298 GB (320,000 ÷ 1024³). 4. Per call (2 outbound streams): 0.000298 GB × 2 = 0.000596 GB per call. 5. For 1,000 concurrent calls active simultaneously (each call 40s on average): concurrent count implies continuous concurrency, but monthly usage depends on call volume. To compute hourly or monthly egress assume sustained concurrency: • Egress per second for 1,000 concurrent calls = 8,000 B/s × 2 × 1,000 = 16,000,000 B/s ≈ 15.26 MB/s. • Egress per hour = 15.26 MB/s × 3600 ≈ 55,000 MB ≈ 53.7 GB/hour. • Egress per month (24×7): 53.7 GB × 24 × 30 ≈ 38,664 GB ≈ 37.8 TB. Or, if you want per‑call basis (given 40s AHT and X calls/day), multiply per‑call GB (0.000596) by total calls per month. Cost of egress (example rate $0.09/GB) • Monthly egress cost ≈ 38,664 GB × $0.09 ≈ $3,480. (Replace $0.09/GB with your exact ap‑south‑1 rate.) Compute (EC2) and infrastructure estimate • Minimal LiveKit + SIP control plane for voice is lightweight compared to video. A starting baseline could be: ◦ 2–3 medium instances for redundancy (e.g., c5.large or t3.medium equivalents), plus a small instance for signaling/SIP if separated. ◦ Example pricing (approximate): c5.large on demand in ap‑south‑1 ≈ $0.096/hr (use AWS Pricing Calculator for exact current rates). • Example EC2 cost (3 × c5.large, 24×30): ◦ Hourly: 3 × $0.096 = $0.288/hr → Monthly ≈ $208. • Other infra (example estimates): ◦ EBS storage (small): $10–$30/month. ◦ Elastic IPs / NAT / ALB: $10–$50/month depending on architecture. ◦ Monitoring/CloudWatch: $10–$50/month. SIP trunking • Since your SIP trunk terminates in‑cluster, trunk provider minutes/billing still apply per their pricing. If your provider charges (for example) $0.008/min, compute minutes used = calls × average duration and multiply by rate. Putting it together (sample totals — replace rates with your exact numbers) • Monthly egress cost (example): ~$3,480 (for 37.8 TB at $0.09/GB). • EC2 + infra: ~$300–$400/month (for a small redundant cluster). • SIP trunk provider: depends on provider rates and total minutes (not included here). Next step Would you like me to: • Recalculate using a different audio bitrate, or using per‑call monthly volume instead of sustained 1,000 concurrent 24×7? • Use the exact AWS ap‑south‑1 data transfer and EC2 rates and output a line‑by‑line monthly estimate?, with interactive elements

I will have sustained 1k concurrent channels for 10 hours. I'll be making outbound calls. Let's say if 40% of the calls are connected, to connect to 1M calls in an hour, I'll need to make around 2.5M calls. What's the estimate of entire AWS cost?

I don't have the answer you're looking for. You could also try asking your question: • in one of the other Slack channels or • to https://deepwiki.com/livekit/livekit_composite which is trained on all LiveKit source code If you find the answer, please post it here to help others!

ok what was the monthly aws estimate for livekit self hosted?