Benchmarks & QoE

Performance

Performance That Speaks for Itself

We instrument every layer — from NVMe queues to player buffer health — so capacity planning is numerical, not narrative. Below are representative outcomes from enterprise pilots; your NexCache team reproduces them against your traces.

Live metrics

Key metrics with in-view instrumentation

Counters animate when they enter the viewport so demos stay engaging without sacrificing accuracy — underlying figures come from fixed test harnesses, not vanity rounding.

0%

OTT egress avoided

p95 across 6-week pilot

0×

Faster first frame

vs. direct internet path

0

Concurrent 4K sessions

Single appliance class node

0%

Peak WAN smoothing

Corporate HQ trace

Cache hit ratio (VOD catalog)94% steady-state
SSD write amplification factor0.42 vs. 2.1 legacy cache
CPU spent in crypto setup per session−63%
Startup path

First-frame latency improvements

The first segment dominates perceived quality. NexCache removes cross-continent TCP slow-start from that transaction by serving ciphertext from a millisecond-local NIC. Players still perform license exchanges with authorized servers — only bulk data movement relocates to the edge.

  • p99 startup aligns with p50 when LAN is uncongested.
  • Fewer renditions stall waiting for initial buffer — ABR climbs smoothly.

Lab capture

Direct CDN1840 ms
NexCache edge260 ms

Capacity planning

Peak bandwidth smoothing

Enterprises pay for the top five minutes of the quarter. NexCache flattens those spikes by absorbing duplicate demand locally — finance sees it on the bill; network sees it on the SNMP graphs.

Telemetry-grade proof

Export before/after Mbps series to your data lake. We encourage third-party verification — no black-box claims.

QoS friendly

Shaped prefetch obeys DSCP and SD-WAN policies so streaming optimization never starves voice or ERP.

Stackable nodes

Horizontal scale adds near-linear hit capacity; hashing avoids hot partitions on mega events.

Load testing

Throughput under load

Synthetic load generators model thousands of simultaneous players with realistic ABR ladders. NexCache sustains line-rate ciphertext egress on modern NICs while keeping CPU headroom for TLS and housekeeping.

Profile
Sustained Gbps
CPU headroom
1080p ladder · 2k sessions
18.4
31%
4K HDR ladder · 400 sessions
22.1
24%
Mixed live + VOD burst
19.8
28%

Figures from dual-25G nodes with NVMe RAID1. Your hardware bill of materials may vary; NexCache publishes sizing guides per vendor SKU.

Comparisons

Benchmark comparisons

Apples-to-apples against legacy transparent proxies and DIY NGINX stacks that cannot retain encrypted segments. NexCache wins on correctness first — then on throughput.

Median time-to-first-frame

1.9 s on constrained WAN0.27 s from edge
7.0× improvement

Peak concurrent 4K sessions (single node)

38 before QoS collapse210 sustained
5.5× headroom

CDN egress cost proxy (normalized)

100% baseline9% residual
91% reduction

Benchmark NexCache on your traffic mix

Bring a PCAP or let Insights capture a read-only sample. We will replay representative workloads in our lab and return a signed performance report for your engineering record.