Ask The Expert: “How much bandwidth and what performance benchmarks do we need for a smooth Microsoft Teams experience in our 500-user Parallels RAS deployment?”

Aug 19, 2025 by Moin Khan

Q:  We’re rolling out Microsoft Teams across our 500-user Parallels RAS deployment, and management is asking tough questions about bandwidth requirements and performance guarantees. How do I calculate what we really need for a seamless Teams experience, and what performance benchmarks should I be targeting?

The Real-World Bandwidth Calculation

Forget Microsoft’s generic bandwidth recommendations—they don’t account for RAS overhead and concurrent usage patterns. Here’s my tested formula:

Per-User Bandwidth Requirements (Optimized)

Audio-Only Calls:

• Baseline: 67 kbps up/down
• RAS protocol overhead: +15%
• Buffer for quality: +25%
• Actual requirement: 95 kbps per concurrent audio session

Video Calls (720p):

• Baseline: 1.2 Mbps up/down
• RAS encoding overhead: +30%
• Network jitter buffer: +20%
• Actual requirement: 1.87 Mbps per concurrent video session

Screen Sharing:

• Static content: 350 kbps
• Dynamic content: 1.5 Mbps
• RAS compression benefit: -20%
• Actual requirement: 280 kbps – 1.2 Mbps

Concurrent Usage Reality Check

Don’t size for 100% concurrent usage—you’ll waste budget. My enterprise data shows:

• Peak concurrent calls: 35-40% of total users
• Video adoption rate: 65% of all calls
• Screen sharing: 25% of calls include sharing

Sample Calculation for 500 Users

Peak concurrent sessions (40%): 200 users

• Audio-only calls (35%): 70 users × 95 kbps = 6.65 Mbps
• Video calls (65%): 130 users × 1.87 Mbps = 243.1 Mbps
• Screen sharing overhead (25%): 50 sessions × 800 kbps = 40 Mbps
• Total peak requirement: ~290 Mbps

My recommendation: Size for 350 Mbps with QoS policies

Performance Benchmarks That Matter

Network Performance Targets

• Round-trip latency: <50ms (ideally <30ms)
• Packet loss: <0.1% for media traffic
• Jitter: <10ms variation
• Available bandwidth utilization: <80% at peak

RAS Host Performance Metrics

CPU Utilization:

• Baseline per user: 5-8% CPU per concurrent Teams session
• Video processing: +15% CPU overhead
• Target: <70% CPU utilization across all RAS hosts

Memory Allocation:

• Teams cache: 200-400MB per user session
• Active call memory: 150-300MB additional
• Sizing rule: 600MB RAM per potential Teams user

Storage IOPS:

• Teams generates 15-25 IOPS per active user
• Profile containers add 10-15 IOPS
• Minimum: 40 IOPS per concurrent user

QoS Configuration for Guaranteed Performance

Traffic Prioritization Strategy

Audio Traffic (DSCP 46 – Expedited Forwarding)

├── Guaranteed bandwidth: 10% of total link

├── Maximum latency: 150ms

└── Packet loss limit: 0.1%

Video Traffic (DSCP 34 – AF41) 

├── Guaranteed bandwidth: 60% of total link

├── Maximum latency: 400ms

└── Burst allowance: 2x guaranteed rate

Application Sharing (DSCP 18 – AF21)

├── Guaranteed bandwidth: 20% of total link 

├── Best effort burst capability

└── Lower priority during congestion

Real-World QoS Implementation

I always implement a three-tier approach:

1. Guarantee tier: Reserve 30% bandwidth for Teams
2. Priority tier: Teams gets first access to additional 40%
3. Best effort: Remaining 30% shared with other applications

Monitoring KPIs for Success

User Experience Metrics

• Call setup time: <3 seconds
• Media quality score: >4.0 (out of 5)
• Connection success rate: >99.5%
• Mid-call drops: <2%

Infrastructure Health Indicators

• RAS connection latency: <20ms internal
• Session host resource utilization: <75%
• Network utilization: <80% during peak
• Storage latency: <10ms response time

My Person Rules for Teams + RAS Success

1. Size for reality, not theory: Use actual usage patterns, not vendor assumptions
2. QoS is non-negotiable: Teams without proper QoS is a guaranteed user complaint generator
3. Monitor proactively: Set alerts at 60% of your capacity thresholds
4. Plan for growth: Size infrastructure for 18-month projected usage
5. Test under load: Always validate with concurrent user simulations before go-live

The Bottom Line

For your 500-user deployment, budget for 400 Mbps of dedicated Teams bandwidth, implement proper QoS policies, and size your RAS hosts for 40% concurrent usage with headroom for growth. Most importantly—establish baseline performance metrics before rollout so you can prove success to management.

Remember: Users don’t care about your technical architecture; they just want Teams to work as well as it does on their laptops. Make that your north star.