CONCEPT Cited by 1 source
Radio-access-network bottleneck¶
Definition¶
In the end-to-end Web protocol stack, the radio-access network (RAN) — the physical link between a mobile device and the carrier's cell tower — is typically the performance bottleneck for the last 100–1000 metres. Zalando's 2024-06 post identifies it as the load-bearing constraint that protocol-level improvements like HTTP/3 have to contend with but cannot eliminate (Source: sources/2024-06-17-zalando-next-level-customer-experience-with-http3-traffic-engineering).
Why it's the bottleneck¶
- Constrained radio spectrum. Frequency bands below 6 GHz are saturated. Adding more users is constrained by Shannon- limit physics, not engineering choices.
- Physical blockage effects. At mmWave (high-bands, 24–52 GHz), radio is highly sensitive to buildings, vehicles, trees, and the human body — small physical changes cause dropped packets that look indistinguishable from congestion to a TCP-era congestion controller.
- Handover loss. Moving between cells, or Wi-Fi ↔ cellular, causes bursty loss and RTT spikes. CC algorithms that treat this as congestion back off unnecessarily.
5G frequency-band trade-offs¶
Zalando's post summarises 5G's three frequency bands:
- Low-bands (sub-1 GHz) — wide-area coverage; low capacity.
- Mid-bands (1–6 GHz, commonly 3.3–4.2 GHz) — trade-off between coverage and capacity; most European commercial 5G deployments use this range.
- High-bands (mmWave, 24–52 GHz) — ultra-high data rates + ultra-low latency, but short range and blockage-sensitive. "Mass scale operating in mmWave spectrum, presents a demanding challenge in terms of its practical implementation and costs."
Zalando's stated expectation: mid-bands will dominate European 5G deployment, because low-bands are capacity- starved and high-bands are CAPEX-prohibitive.
Operational numbers at 5G mid-bands¶
From Zalando's cited field measurements (Ritola 2022 thesis, even at bad signal strength -100 dBm to -113 dBm):
- Downlink latency: p99.9 = ~13 ms
- Uplink latency: p99.9 = ~28 ms
- Peak data rate: up to 20 Gbps
- Average data rate: > 100 Mbps
- Device density: up to 1 million devices / km²
5G mid-bands outperform 3G/4G/LTE on both latency and packet- loss probability, but the radio-physics constraints remain.
Congestion-control implications¶
This is where Zalando's CC-algorithm analysis lands:
"High-reliability plays against the congestion control algorithms used by QUIC. Conventional algorithms are not able to differentiate between the potential causes of packet loss or congestion on the radio channel due to noise, interference, blockage or handover. NewReno and CUBIC have resulted in very poor throughput and latency performance. Only BBR exhibited the lowest round trip time values among all possible physical failure scenarios and can satisfy the typical 5G requirements."
Loss-signalled CC algorithms (CUBIC, NewReno) assume loss = congestion; on radio this is wrong. Model-based CC (BBR) estimates bottleneck bandwidth + RTT and is robust to RF loss.
4K UHD video as a RAN-limited workload¶
Zalando positions 4K UHD streaming as the canonical workload pushing RAN limits:
- H.265 at 3840×2160, 60 fps demands 30–50 Mbps bandwidth, 6–11 ms latency, 99.999 % packet-delivery reliability.
- Mid-bands can deliver this in urban areas only — the physical link is the constraint.
- 5G mmWave unlocks it but is deployment-limited.
Architectural implication¶
A protocol-layer optimisation (HTTP/3 + QUIC) can reduce handshake RTTs, fuse transport + TLS, multiplex streams, and migrate connections — but cannot add bandwidth to a saturated radio spectrum. The RAN bottleneck is the architectural constraint every mobile-targeting system must plan around.
Seen in¶
- sources/2024-06-17-zalando-next-level-customer-experience-with-http3-traffic-engineering — canonical wiki instance. Zalando uses the RAN bottleneck as the reason HTTP/3's congestion-control story has to include BBR and not just inherited TCP defaults.