Skip to main content

​
Overview

This section provides comprehensive network performance data for Aviatrix cloud networking infrastructure, including throughput characteristics, latency measurements, and connection scaling limits across different deployment scenarios and cloud platforms.

​
Throughput Performance

​
Inter-Region Performance

| Source Region | Destination Region | Distance | Typical Latency | Max Throughput | Optimal Instance | |---------------|-------------------|----------|-----------------|----------------|------------------| | us-east-1 | us-west-2 | ~3,000 km| 65-75 ms | 10 Gbps | c5n.4xlarge | | us-east-1 | eu-west-1 | ~5,500 km| 80-90 ms | 8 Gbps | c5n.4xlarge | | us-east-1 | ap-southeast-1 | ~17,000 km| 180-200 ms | 5 Gbps | c5n.2xlarge | | eu-west-1 | ap-southeast-1 | ~11,000 km| 160-180 ms | 6 Gbps | c5n.2xlarge | | us-west-2 | ap-northeast-1 | ~8,000 km| 120-140 ms | 7 Gbps | c5n.4xlarge | | eu-central-1 | us-east-1 | ~6,000 km| 90-100 ms | 8 Gbps | c5n.4xlarge |

​
Same-Region Performance

Cloud ProviderSame AZDifferent AZCross-RegionBackbone Network
AWS< 1 ms1-2 msVariableUp to 100 Gbps
Azure< 1 ms1-3 msVariableUp to 100 Gbps
GCP< 1 ms1-2 msVariableUp to 100 Gbps
Performance Characteristics:
  • Same Availability Zone: Sub-millisecond latency, full instance bandwidth
  • Different Availability Zones: Low single-digit millisecond latency
  • Cross-Region: Varies based on geographic distance and network path
Instance ClassNetwork PerformanceBaseline BandwidthBurst BandwidthDuration
BurstableUp to 5 Gbps200 Mbps5 Gbps30 min
General PurposeUp to 25 Gbps1 Gbps10 GbpsSustained
Compute OptimizedUp to 50 Gbps10 Gbps25 GbpsSustained
Network OptimizedUp to 100 Gbps25 Gbps100 GbpsSustained

​
Latency Characteristics

​
Network Latency Factors

Distance-Based Latency (One-Way)
Distance RangeTypical LatencySpeed of Light LimitReal-World Factor
0-500 km2-8 ms1.7 ms1.2-4.7x
500-1,500 km8-20 ms5-10 ms1.6-2.0x
1,500-5,000 km20-60 ms10-33 ms1.8-2.0x
5,000-10,000 km60-120 ms33-67 ms1.8-1.8x
10,000+ km120-250 ms67+ ms1.8-3.7x
Factors Affecting Real-World Latency:
  • Fiber optic cable routing (not straight-line distance)
  • Number of network hops and routing equipment
  • Network congestion and quality of service
  • Terrestrial vs. submarine cable characteristics

​
Connection Scaling

​
Connection Limits by Gateway Type

Instance SizeConcurrent ConnectionsNew Connections/secBGP SessionsRoute Capacity
Small50,0002,50010010,000
Medium125,0006,25025025,000
Large250,00012,50050050,000
XLarge500,00025,0001,000100,000
XXLarge1,000,00050,0002,000200,000
Connection Distribution:
  • East-West Traffic: 60-70% of total connections
  • North-South Traffic: 20-30% of total connections
  • Management Traffic: 5-10% of total connections
Instance SizeConcurrent ConnectionsNew Connections/secApplication FlowsSession Capacity
Small10,0005001,0005,000
Medium25,0001,2502,50012,500
Large50,0002,5005,00025,000
XLarge100,0005,00010,00050,000
XXLarge200,00010,00020,000100,000
Connection Patterns:
  • Database Connections: Long-lived, low bandwidth
  • Web Service Connections: Short-lived, medium bandwidth
  • File Transfer Connections: Medium-lived, high bandwidth
  • API Connections: Very short-lived, low bandwidth
Gateway TypeUser ConnectionsSite ConnectionsVPN TunnelsFirewall Sessions
UserVPN Small100N/A10010,000
UserVPN Large2,000N/A2,000200,000
Site2Site SmallN/A252525,000
Site2Site LargeN/A200200200,000
Hybrid1,000100100100,000
Connection Characteristics:
  • UserVPN: High session count, moderate bandwidth per user
  • Site-to-Site: Low session count, high bandwidth per tunnel
  • Hybrid Deployments: Balanced approach for mixed requirements

​
Performance Monitoring

​
Key Performance Indicators

Primary Metrics
  • Bits per second (bps): Raw data transfer rate
  • Packets per second (pps): Packet processing rate
  • Connections per second (cps): Connection establishment rate
  • Utilization percentage: Resource usage levels
Monitoring Thresholds
  • Green Zone: < 60% of maximum capacity
  • Yellow Zone: 60-80% of maximum capacity
  • Red Zone: > 80% of maximum capacity
  • Critical Zone: > 95% of maximum capacity
Alerting Recommendations
  • Warning alerts at 70% utilization
  • Critical alerts at 85% utilization
  • Automated scaling triggers at 80% utilization

​
Performance Testing Methodology

​
Test Scenarios

Single Stream Tests
  • Purpose: Measure maximum single-connection performance
  • Tools: iperf3, netperf, custom applications
  • Duration: 60-300 seconds for statistical significance
  • Metrics: Throughput, latency, CPU utilization
Multi-Stream Tests
  • Purpose: Measure aggregate performance across multiple connections
  • Configuration: 1, 2, 4, 8, 16, 32 concurrent streams
  • Analysis: Scaling efficiency and resource utilization
  • Optimization: Connection balancing and resource allocation
Application Mix Testing
  • Web Traffic: HTTP/HTTPS with varying object sizes
  • Database Traffic: SQL queries and transaction patterns
  • File Transfer: Large file uploads and downloads
  • Real-time Applications: VoIP and video streaming
Traffic Pattern Simulation
  • Peak Hour Traffic: Maximum expected load
  • Normal Operating Conditions: Typical daily patterns
  • Burst Traffic: Sudden traffic spikes
  • Failure Scenarios: Failover and recovery testing
Connection Stress Tests
  • Maximum connection establishment rate
  • Connection table exhaustion scenarios
  • Rapid connection cycling patterns
  • Resource cleanup and garbage collection
Bandwidth Stress Tests
  • Maximum sustained throughput
  • Network interface saturation
  • Buffer overflow conditions
  • Quality of service under load
Network performance data is collected under controlled laboratory conditions and validated through real-world deployments. Actual performance may vary based on network topology, traffic patterns, and environmental factors.
Performance Optimization: For optimal network performance, monitor key metrics continuously, establish performance baselines, and implement proactive scaling based on utilization trends rather than reactive scaling based on performance degradation.