Direct Answer
How should edge gateways be sized when a company needs to upload data from hundreds of meters and controllers to one platform? Gateway selection starts with three capacity dimensions — devices and data points per gateway (for example, ESX supports about 30 devices/2,000 points), the downlink protocol (RS485/Modbus, Zigbee or LoRa, chosen by on-site cabling) and the uplink (wired Ethernet/4G, chosen by on-site network conditions) — then divide the total points by per-gateway capacity and add redundancy.
A scenario: A park with 300 meters needs to go on the platform — too few gateways cannot handle the load and data queues up — too many waste money — a wrong uplink choice leaves the site with no signal. The logic of selection is to size capacity first, then choose the protocol, then settle the uplink — when all three dimensions align, the gateway will not fail.
1. Why Gateway Selection Often Fails
Failure one, not enough capacity — too many devices: the gateway cannot handle the load and data queues; Failure two, wrong protocol — mismatch with the site: cabling conflicts with the protocol; Failure three, no uplink — no network: 4G without coverage, no cable reachable; The cost of failure — rework: changing the plan on site, rewiring; The root of failure — ignoring the site: selecting by model, not by scenario.
In one sentence: gateway failures are not enough capacity, wrong protocol and no uplink — the root is ignoring the site.
An example: A factory gateway carries 300 devices — the single unit supports only 30 — data jams — rework to add gateways. The logic is that selection starts with the site — device count, cabling and network — when the three dimensions match, it will not be discovered on site that it cannot handle the load.
2. The Three Capacity Elements: Devices, Data Points and Cache Days
Element one, devices — the access limit: how many meters/controllers one unit can connect; Element two, data points — the point limit: how many data points one unit can acquire; Element three, cache days — local storage: how many days it can cache offline; The relation — mutual constraints: more devices and points reduce the cache; The meaning — the capacity baseline: the first foundation of selection.
In one sentence: the three capacity elements are devices, data points and cache days — the first foundation of selection.
An example: ESX supports about 30 devices/2,000 points: 300 meters need about 10 gateways. The logic is to look at per-unit capacity first — divide the total points by per-unit capacity — get the gateway baseline — then adjust with a redundancy factor.
3. How to Choose the Downlink Protocol
Protocol one, RS485/Modbus — wired: cabling to the device, stable and reliable; Protocol two, Zigbee — wireless: no cabling, short range, good for retrofits; Protocol three, LoRa — long-range wireless: kilometer level, low power, good for scattered sites; The basis of choice — site cabling: choose RS485 when cabling is possible, wireless when it is hard; The reminder — match the devices: choose what the devices support.
In one sentence: the downlink protocol is RS485 when cabling is possible, Zigbee/LoRa when it is hard — follow the site and the devices.
An example: A park's meters are concentrated: RS485 cabling is stable — a factory's equipment is scattered: LoRa wireless avoids cabling. The logic is that the protocol follows the site conditions — wired for concentrated, wireless for scattered — device support is the premise.
4. The Uplink: Ethernet vs 4G
Link one, wired Ethernet — stable: cable/fiber available, sufficient bandwidth; Link two, 4G wireless — flexible: no cable, remote sites; Trade-off one, stability — wired wins: Ethernet is more stable than 4G; Trade-off two, bandwidth — wired wins: large data volumes go over Ethernet; Trade-off three, coverage — 4G wins: sites without cable can only use 4G.
In one sentence: the uplink is Ethernet when cable exists, 4G when it does not — stability first, coverage as the fallback.
An example: A factory's substation has fiber: Ethernet uplink is stable — a remote pumping station has no network: 4G uplink serves. The logic is that the uplink follows on-site network conditions — wired when available, 4G when not — bandwidth-heavy important sites prefer Ethernet.
5. Point Estimation and Redundancy Configuration
Step one, count the points — inventory: the number of meters and data points; Step two, calculate the baseline — division: total points divided by per-unit capacity; Step three, add redundancy — factor: reserve 20-30% margin; Step four, plan the distribution — location: distribute gateways by substation/workshop; The meaning of estimation — no rework: calculated once, with sufficient margin.
In one sentence: point estimation is counting, division, redundancy and location-based distribution — calculated once, no rework.
An example: A park with 300 meters × 10 points each = 3,000 points — ESX 2,000 points per unit — 2 units plus 20% redundancy = 3 units — distributed by substation. The logic is to count the total first, then the units, then add redundancy — reserve margin for expansion — distribute by on-site location. One unit fewer means jams, one more means waste — the redundancy factor is the balance between avoiding jams and saving money.
6. What Local Preprocessing the Edge Gateway Does
Preprocessing one, data cleaning — filtering: invalid frames and duplicates are filtered locally; Preprocessing two, local calculation — aggregation: second-level data aggregated before upload; Preprocessing three, offline buffering — caching: data cached locally during outages; Preprocessing four, local alarms — edge judgment: simple violations respond locally; The value — bandwidth savings: the uploaded volume drops sharply.
In one sentence: edge computing is cleaning, aggregation, buffering and local alarms — the upload volume drops sharply.
An example: A factory gateway aggregates locally: second-level becomes minute-level upload — bandwidth saved by 90% — three days offline without data loss. The logic is that a gateway is not a dumb transmitter — local processing makes the upload cheaper and more stable — edge computing is part of the selection.
7. Back to YESLON: How to Match the ESX and CW Series
ESX-0223-GR — high performance: about 30 devices/2,000 points, edge computing; CW-C1/C2/C3 — tiered: selected by point scale; The matching principle — by scale: ESX for large projects, CW for small and medium ones; The reminder — protocols: on-site device protocols must match the gateway; The YESLON promise — complete selection: capacity, protocol and uplink configured once.
In one sentence: the YESLON gateways are ESX for high performance and CW in tiers — matched by scale and protocol.
An example: A park: 3,000 points uses 3 ESX units — a small factory with 100 points uses CW-C1 — selected by scale. The logic is that gateways are tiered by scale — more points use ESX, fewer use CW — protocol matching is the key.
8. Self-Check: Four Questions to See Whether the Gateway Is Correct
Question one, is the capacity enough — capacity: are devices, points and cache sufficient; Question two, is the protocol right — protocol: does the downlink match the site and devices; Question three, does the uplink work — uplink: is Ethernet/4G available on site; Question four, is there redundancy — redundancy: is 20-30% margin reserved.
In one sentence: the four questions are capacity, protocol, uplink and redundancy — all answered, gateway selection does not fail.
A self-check example: A factory answers the four questions: capacity enough, protocol right, uplink works and redundancy present — the gateway configuration passes. Edge gateway selection starts with three capacity dimensions — devices and data points per gateway, the downlink protocol chosen by on-site cabling and the uplink chosen by on-site network conditions — then divide the total points by per-unit capacity and add redundancy — size the capacity first, then choose the protocol, then settle the uplink — when the three dimensions align, data from hundreds of meters can reach the platform stably.
Scope and Limitations
- The conclusions in this article are intended for LV distribution and electrical-safety sites. Criteria are compiled from public codes and FEXLINK product documentation, and product parameters follow the latest 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 intelligent edge computing gateway, CW-C1/C2/C3 industrial gateway 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
With Hundreds of Meters Pushing Data to the Platform, How Should Edge Gateways Actually Be Sized?
How should edge gateways be sized when a company needs to upload data from hundreds of meters and controllers to one platform