India VPS Performance Benchmark Report: CPU, Storage & Network Telemetry
Objective, verifiable hardware benchmark results for VisiHost India KVM virtual private servers hosted in Sector 62, Noida (Delhi NCR). Documents raw computational throughput via Sysbench 1.0.20, random 4K IOPS via FIO 3.36, stress-ng 4-hour thermal stability, and pan-India optical latency.
1. Executive Summary
When choosing cloud hosting for production databases, APIs, or Docker workloads, marketing claims like “ultra-fast compute” are meaningless without reproducible numbers. To provide full transparency, VisiHost conducted comprehensive benchmark runs on standard production nodes provisioned identically to customer virtual machines.
99.2% scaling efficiency between 1 and 2 threads. Less than 0.06 ms jitter between minimum and 99th percentile latency.
35,240 read IOPS and 22,110 write IOPS at queue depth 32. Sequential read bandwidth exceeds 1,840 MB/s.
Sub-30 ms nationwide transit to Mumbai (21 ms), Bangalore (24 ms), Hyderabad (25 ms), and Kolkata (18 ms).
Zero clock speed throttling, zero memory page faults, and zero hypervisor eviction across 4 continuous hours.
2. Test Environment Specification
All benchmarks were performed on a production-grade VisiHost Intel Xeon XS instance provisioned through our automated billing portal without custom kernel tuning or priority hypervisor pinning.
| Target Plan Tier | Intel Xeon XS (Entry ₹389/mo plan) |
| Virtual CPU Allocation | 2 vCores (Intel Xeon Scalable / Platinum Platform) |
| Memory Allocation | 6,144 MB (6 GB) DDR4 Registered ECC RAM |
| Storage Allocation | 50 GB Enterprise PCIe NVMe SSD (Hardware RAID) |
| Virtualization Layer | KVM (Kernel-based Virtual Machine) with virtio drivers |
| Operating System | Ubuntu 24.04 LTS (Noble Numbat) x86_64 |
| Linux Kernel | 6.8.0-31-generic |
| Datacenter Facility | Electronic City Sector 62, Noida (Delhi NCR), UP, India |
| Network Uplink | 1 Gbps switch port with NIXI Delhi upstream routing |
3. CPU Compute Benchmark: Sysbench 1.0.20
Sysbench prime number verification measures raw integer compute throughput and latency consistency under single-core and multi-core loads. The test computes prime numbers up to 20,000 across 10,000 total events.
sysbench cpu --cpu-max-prime=20000 --threads=2 run
| Metric | 1 Thread (Single Core) | 2 Threads (Dual Core) |
|---|---|---|
| Total Events Processed | 10,000 | 10,000 |
| Total Execution Time | 7.2148 s | 3.6421 s |
| Events Per Second | 1,385.98 / sec | 2,745.36 / sec |
| Latency: Minimum | 0.71 ms | 0.70 ms |
| Latency: Average | 0.72 ms | 0.73 ms |
| Latency: 95th Percentile | 0.73 ms | 0.74 ms |
| Latency: 99th Percentile | 0.75 ms | 0.76 ms |
Scaling Efficiency: 2-thread execution finished in 3.64 seconds vs 7.21 seconds on 1 thread (99.2% linear scaling). This confirms unthrottled guest vCPU assignment.
Jitter Analysis: The variance between minimum latency (0.70 ms) and the 99th percentile (0.76 ms) was 0.06 ms, demonstrating zero host core overcommitment.
4. Storage Subsystem Benchmark: FIO 3.36 Enterprise NVMe
FIO (Flexible I/O Tester) was executed to measure random 4K read/write I/O performance (simulating transactional SQL workloads) and sequential 1M throughput (simulating database dumps and container image pulls).
fio --name=fio-test --ioengine=libaio --iodepth=32 --rw=randrw --rwmixread=75 --bs=4k --direct=1 --size=1G --numjobs=4 --runtime=30 --group_reporting
| Test Profile | IOPS | Throughput | Average Latency |
|---|---|---|---|
| Random 4K Read (75%) | 35,240 IOPS | 137.6 MB/s | 0.48 ms |
| Random 4K Write (25%) | 22,110 IOPS | 86.4 MB/s | 0.58 ms |
| Combined 4K Random IOPS | 57,350 IOPS | 224.0 MB/s | 0.52 ms |
| Sequential 1M Read | 1,840 IOPS | 1,840 MB/s | 0.54 ms |
| Sequential 1M Write | 1,280 IOPS | 1,280 MB/s | 0.78 ms |
5. Network Performance & Pan-India Latency Telemetry
Round-trip ICMP echo latency measured from our Noida datacenter across major Indian telecommunications corridors. Each data point represents the average of 1,000 packets with zero measured packet loss (0.0%).
| City / Economic Zone | State | Round-Trip Ping | Transit Classification |
|---|---|---|---|
| Noida (Electronic City Sector 62) | Uttar Pradesh | 1.2 - 2.8 ms | Datacenter Local |
| Delhi NCR (Connaught Place / South Delhi) | Delhi | 2.4 - 5.6 ms | Direct Fiber Transit |
| Gurugram (Cyber City) | Haryana | 3.8 - 6.2 ms | Direct Fiber Transit |
| Chandigarh / Mohali | Punjab | 8.4 - 11.2 ms | Northern Corridor |
| Jaipur | Rajasthan | 9.2 - 12.8 ms | Western Route |
| Lucknow | Uttar Pradesh | 10.5 - 14.1 ms | Central Transit |
| Kolkata | West Bengal | 17.4 - 21.0 ms | Eastern Corridor |
| Ahmedabad | Gujarat | 18.6 - 23.5 ms | Western Corridor |
| Mumbai | Maharashtra | 20.8 - 25.2 ms | Fintech Corridor |
| Pune | Maharashtra | 22.1 - 26.8 ms | IT Corridor |
| Bengaluru | Karnataka | 23.4 - 28.2 ms | Tech Hub South |
| Hyderabad | Telangana | 24.2 - 29.6 ms | South Central |
| Chennai | Tamil Nadu | 26.8 - 32.1 ms | Southern Coastal |
Choosing a remote desktop for work? See how these routes relate to Windows RDP hosting in India.
6. Thermal Stability & Sustained 4-Hour Stress Test
Many budget cloud providers employ aggressive hypervisor burst-throttling that drops CPU clocks by 40–60% under continuous compute loads. We subjected our test instance to a 4-hour continuous 100% stress test using stress-ng:
stress-ng --cpu 2 --vm 1 --vm-bytes 4G --timeout 4h --metrics-brief
Zero Clock Degradation: CPU clock frequencies remained completely stable throughout the 4-hour window with zero thermal throttling or hypervisor core eviction.
Memory Allocation Integrity: Zero memory page faults, kernel allocation stalls, or Out-Of-Memory (OOM) killer terminations occurred across 4 GB of continually cycled physical memory.
Concurrent Storage Latency: Random I/O operations executed during the final 30 minutes of the stress run registered identical response times (0.52 ms average).
7. Reproduce These Benchmarks Independently
We encourage developers and sysadmins to verify their server performance independently. Run these standard commands on your own VisiHost Linux VPS:
Deploy on Verified India VPS Hardware
Experience the same enterprise Intel Xeon compute nodes, DDR4 ECC memory, and NVMe SSD arrays documented in this benchmark report starting at ₹389/month.