Direct Answer

Meters all support 4G/MQTT — why do some scenarios still need an ESX at the cabinet for local computing? Three scenarios must push computation down: 1) no-network cannot stop (the site must alarm and trip locally); 2) high-frequency waveforms (microsecond data never goes to the cloud); 3) multi-protocol private devices (must be translated on site); otherwise direct-to-cloud can skip the edge layer. Landing = first ask whether the three must-have scenarios hit, then decide whether to drop the gateway.

A scenario: A project's meters all support 4G/MQTT — the customer asks: do we still need a gateway? — the answer: it depends — pure meter reading with tolerable outages can go direct-to-cloud; but a process loop that must trip when the network drops, microsecond waveform capture, or private-protocol devices — the ESX is required — one of the three hits, and the edge layer cannot be skipped. The decision logic is not "is the gateway expensive" but "does the site need reliability".

1. Direct-to-Cloud: Meters Straight to the Platform

What direct-to-cloud is — meters connect directly: meters carry 4G/MQTT and go straight to the platform; The direct advantage — one layer less: no gateway, one device less; The direct fit — standard protocols: Modbus/TCP or DL/T 645 uploaded directly; The direct premise — tolerable data loss: losing data during an outage is acceptable; The direct essence — the device is the endpoint: each meter uploads independently, the platform manages devices directly.

In one sentence: direct-to-cloud means meters upload via 4G/MQTT directly — one layer saved, on the premise that data loss is tolerable.

An example: A pure meter-reading project: energy meters direct-to-cloud — no gateway — the platform collects — lowest cost. The logic is that direct value lies in saving — one layer less means one cost less.

2. The Three Boundary Conditions of Direct-to-Cloud

Condition one, outage is tolerable — no alarming or tripping during an outage is accepted; Condition two, low-frequency data — second/minute-level data suffices, no high-frequency waveform need; Condition three, standard protocols — devices speak standard protocols, no on-site translation; All three met — direct is viable: the edge layer can be dropped; One unmet — edge is mandatory: computation goes down.

In one sentence: the three boundaries are outage tolerable, low-frequency data and standard protocols — all three met before dropping the edge.

An example: A park self-checks the three: outage tolerable, minute-level data, Modbus standard — all met — direct-to-cloud — gateway saved. The logic is that boundaries are a permission list — all three met before direct; one unmet, the edge layer comes back.

3. Scenario One: No-Network Cannot Stop — Local Alarm and Tripping

What no-network-cannot-stop means — process safety: the site must act even offline; The local alarm — judged on site: threshold exceeded alarms locally; The local trip — executed on site: linkage output acts directly; The cloud limit — remote dependency: cloud commands cannot reach the site offline; The scenario-one conclusion — must go down: the edge closes the loop locally, no-network never stops.

In one sentence: no-network cannot stop means a local loop — alarming and tripping still work offline — the edge must go down.

An example: A chemical loop: must trip above threshold when the network drops — ESX local linkage — still acts after five minutes offline — process safety held. The logic is that safety loops cannot depend on the network — offline, cloud commands never arrive; the local loop is the bottom line.

4. Scenario Two: High-Frequency Waveforms — Microsecond Data Never Goes Up

What high-frequency waveforms mean — recording needs: fault waveforms and harmonic transients at microsecond level; The cloud limit — bandwidth and cost: microsecond data would need too much bandwidth and cost; The local role — on-site recording: waveforms stored at the edge, features to the cloud; The local analysis — on-site criteria: waveforms recognized locally, conclusions only; The scenario-two conclusion — must go down: high-frequency data processed locally, the cloud receives only features.

In one sentence: high-frequency waveforms are recorded and judged locally — microsecond data never goes up, only features.

An example: A project needs fault waveforms: microsecond data stored at the ESX — waveform features to the cloud — 90% bandwidth saved. The logic is that high-frequency cost lies in bandwidth — microsecond data on the cloud is unrealistic — the edge processes locally, the cloud receives only features.

5. Scenario Three: Multi-Protocol Private Devices — On-Site Translation

What multi-protocol means — mixed devices: different vendors and protocols coexist; The private protocol — nonstandard: the platform cannot recognize proprietary protocols; The on-site role — protocol translation: proprietary protocols translated into a standard format at the edge; The on-site role two — data normalization: different devices normalized into one model on the cloud; The scenario-three conclusion — must go down: translated on site, so the cloud collects uniformly.

In one sentence: private-protocol devices are translated on site and normalized — the edge translates, the cloud collects uniformly.

An example: A site with three private-protocol devices: the ESX translates on site — one data model to the cloud — no platform adaptation. The logic is that private-protocol adaptation is expensive — one translation at the edge saves the cloud adaptation.

6. The Edge-Mandatory Checklist

Check one, can outage stop — no: the local loop is mandatory; Check two, any high-frequency waveforms — yes: on-site recording is mandatory; Check three, protocols standard — no: on-site translation is mandatory; Check four, local linkage needed — yes: local logic is mandatory; Check five, the three-question conclusion — any hit: the edge layer cannot be dropped.

In one sentence: the edge checklist is outage cannot stop, high-frequency waveforms, nonstandard protocols and local linkage — any hit, the edge must be there.

An example: A project self-checks: outage tolerable, no waveform, standard protocols — but local linkage needed — one hit — install the ESX. The logic is that the checklist is an OR — not all hits required to install — any one hit gives the edge value — run through each item before deciding.

7. ESX Local Logic: What Mistudio's 300+ Instructions Can Do

What Mistudio is — a local programming tool: graphical logic orchestration for the ESX; The local logic capability — 300+ instructions: conditions, arithmetic, linkage and sequencing covered; The local logic landing — on-site execution: threshold judgment and interlock tripping done at the cabinet; The local logic output — linkage execution: output relays, sound-light and breakers; The local logic value — usable offline: logic runs at the edge, independent of the cloud.

In one sentence: Mistudio is ESX graphical programming — 300+ instructions run at the cabinet — linkage works offline.

An example: A project orchestrates in Mistudio: over-temperature plus current judgment — local trip output — acts normally offline. The logic is that local-logic value lies in on-site execution — instructions run at the cabinet, independent of the cloud.

8. The Hybrid Architecture: The Typical Edge-Cloud Topology

What the hybrid is — edge-cloud collaboration: local loop at the edge plus global analytics in the cloud; The edge role — real-time response: local alarm, tripping and recording; The cloud role — global insight: cross-park comparison, AI algorithms and reports; The data layering — detail at the edge, features to the cloud: bandwidth saved, sovereignty kept; The hybrid form — typical topology: meters to the ESX to FEXCloud.

In one sentence: edge-cloud collaboration is edge response plus cloud insight — detail at the edge, features to the cloud.

An example: A group runs edge-cloud: ESX local loops in 30 parks — features to FEXCloud — global comparison — a clear layered architecture. The logic is that the hybrid gets both — site reliability from the edge, global intelligence from the cloud.

9. The Cost-Reliability Tradeoff

The direct cost — low: no gateway, no edge investment; The direct risk — out of control offline: alarming and tripping fail during outages; The edge cost — present: gateway hardware plus implementation; The edge value — reliable: local loop and on-site response; The tradeoff conclusion — reliability first: when a must-have scenario hits, cost gives way to reliability.

In one sentence: the tradeoff is direct saves money and edge keeps reliability — when a must-have hits, reliability comes first.

An example: A project: direct saves 30k a year — but five loops cannot stop offline — install the ESX — no saving but worth it. The logic is that cost must be weighed against the cost of losing control — one offline loss can exceed the gateway price — reliability first.

10. Back to YESLON: The ESX Edge Capability

The ESX position — the edge node: cabinet computing, protocol translation and local logic; ESX-0223-GR — a typical model: multi-protocol access plus edge computing; The Mistudio companion — graphical programming: 300+ instructions orchestrated locally; The FEXCloud collaboration — edge-cloud: edge response plus cloud analytics; The ESX fit — selected by scenario: the landing carrier of the three must-have scenarios.

In one sentence: the ESX is edge computing, protocol translation and local logic — the carrier for the three must-have scenarios.

An example: A project hits all three: ESX translates on site, alarms locally and records waveforms — Mistudio orchestration — offline loop closed. The logic is that the ESX is the landing of edge-mandatory — not an optional gateway but the executor of local reliability.

11. Self-Check: Five Questions to Fix the Architecture

Question one, can outage stop — reliability: can loss of control be accepted offline; Question two, any high-frequency waveforms — data: any microsecond recording need; Question three, protocols standard — adaptation: are device protocols uniformly standard; Question four, local linkage needed — loop: must actions execute locally; Question five, has the three-year account been done — cost: edge investment vs loss-of-control damage.

In one sentence: the five questions are outage, waveform, protocol, linkage and cost — all answered, the architecture is clear.

A self-check example: A project runs the five: no-network cannot stop (local trip needed), no waveform, standard protocols, linkage needed — two hits — install the ESX — architecture set as edge-cloud. The decision logic of direct-to-cloud vs edge computing is to first ask whether the three must-have scenarios hit — no-network cannot stop needs a local loop, high-frequency waveforms need on-site recording, private protocols need on-site translation — any hit, the edge must go down; none hit, direct-to-cloud drops the edge layer — reliability first, cost gives way to safety.

Scope and Limitations

  • The conclusions in this article are intended for LV distribution and electrical-safety sites. Standard numbers (DL/T 645) refer to their currently effective editions, and product parameters follow the latest FEXLINK selection handbook and product knowledge base.
  • Quantitative criteria (protection zone, discharge and residual voltage, earth resistance, thresholds and grading) hold under the preconditions stated in this article. Where the site voltage level, earthing arrangement, environmental category or load profile differs, the figures must be recalculated. Specific configurations involving ESX-0223-GR、Mistudio should be verified against actual circuit parameters.
  • Specific product models and parameters are subject to the latest selection handbook and product knowledge base; this article does not constitute selection or procurement advice.
  • Standard numbers cited here refer to the currently effective editions; when a standard is revised, the related quantitative conclusions must be re-checked.
  • Deployment and implementation for specific projects must be carried out by qualified engineers based on site survey and design drawings.

Frequently Asked Questions

Direct-to-Cloud vs Edge Computing

Meters all support 4G/MQTT — why do some scenarios still need an ESX at the cabinet for local computing