Network Readiness for 1:1: Sizing a District Network Before the Devices Land

Enrollment is the wrong denominator. The number that sizes a network is how many devices are actively transmitting in the worst hour of the year, and in most districts that hour is a state assessment window, not a normal Tuesday.

Build the count from the room outward. Take the largest class size the building actually runs, add the teacher device, add any fixed devices on wireless, and add the staff and guest devices that will be in the same coverage area. Then find the concurrency peak: assessment windows, all-school announcements pushed to displays, the first morning of a rollout when every device checks in for policy and updates at once. Design for that hour with headroom, because the whole point of the exercise is that the network stops being visible.

Two adjustments matter. First, devices per student is drifting above one in secondary schools — a district-issued laptop plus a personal phone on the same network is two associations, whether or not the phone is welcome. Decide the policy before you size, because a guest network you did not plan for still consumes capacity. Second, if the district is moving from carts to per-student devices, the concurrency change is far larger than the device-count change: carts serialize usage by design.

Hallway-mounted access points covering two or three rooms each were a defensible design when a classroom had four devices in it. In a 1:1 building they are the most common physical cause of poor performance, because the access point ends up serving three rooms' worth of clients through two walls.

The modern default in instructional space is an access point per classroom, ceiling-mounted inside the room. That is not a rule of physics, it is a consequence of client density and of how much of the useful spectrum is at 5 and 6 GHz, where range through masonry and metal stud is shorter than the 2.4 GHz behavior everyone's intuition is calibrated on. Concrete block, tile, plaster over lath, and rooms with metal casework all cost more signal than a floor plan suggests.

Two design consequences follow. First, more access points at lower transmit power beats fewer at high power: cranking power to cover a dead spot widens the co-channel interference problem across the building. Second, channel planning becomes real work once you have an access point per room — and this is where Wi-Fi 6E and Wi-Fi 7 change the arithmetic materially, because 6 GHz supplies channel width that the 5 GHz band in a dense building simply does not have.

Do a survey. A predictive survey against real floor plans and real wall construction, validated on site, costs a fraction of a rollout and is the only way to turn the paragraphs above into an actual mounting list.

Add up every powered device the closet will serve after the rollout — access points, cameras, phones, door controllers, display controllers — using each device's maximum published draw rather than its typical draw, and compare the total against the switch's published PoE power budget. That total is the constraint, not the port count. A 48-port PoE+ switch does not deliver full PoE+ on all 48 ports at once, and a design that assumes it will fail on the day the last camera is installed.

Access points are what usually forces the upgrade. A Wi-Fi 6E or Wi-Fi 7 access point can move a closet from 802.3af to 802.3at or 802.3bt, and once several rooms per closet make that jump the old switch is out of budget regardless of free ports. The catalog side of this is unglamorous and specific: a 48-port PoE+ smart switch such as the Netgear GS752TPP-300NAS, or a managed AV-line switch like the Netgear M4250-26G4XF-PoE+, or Ubiquiti's Enterprise 48 PoE where the district is standardized on that stack. Read the published power budget on the model you are actually buying rather than the family.

Then the uplink. An access switch serving twenty classrooms of 1:1 devices on a single gigabit uplink is the most common invisible bottleneck in K-12 networks, because every component tests fine individually. Size access-to-distribution uplinks against the concurrent load you calculated, use link aggregation or 10 Gb uplinks where the math says so, and remember that adding access points multiplies the demand on the same uplink rather than relieving it.

SETDA's Broadband Imperative III — published in 2019 and written for the 2023-24 school year, so read it as a planning floor rather than a current ceiling — gives the targets: peak internet capacity of at least 2.8 Mbps per user for small districts with a floor of 300 Mbps per district, at least 2 Mbps per user for medium districts of 1,000 to 10,000 students, and at least 1.4 Mbps per user for large districts, alongside wide-area network capacity of 10 Gbps per 1,000 users. The FCC's long-term external connectivity goal, as published by EducationSuperHighway, is at least 1 Mbps per student.

We do not supply circuits. Internet access, transport between buildings, and the contracts behind them come from carriers and regional networks, and the lead time on a circuit upgrade is frequently longer than the lead time on every piece of hardware in the rollout combined. If the sizing exercise says the circuit is short, that conversation starts first — before the switches, before the access points, and certainly before the devices.

On E-rate, and strictly as information: the program divides eligible services into Category One, which is connectivity to the building, and Category Two, which is connectivity within it — access points, routers, switches, hubs and cabling, along with basic maintenance of those internal connections. For the FY2026-2030 five-year cycle the FCC set the school and district Category Two pre-discount multiplier at $201.57 per student and the funding floor at $30,175, and applicants validate their student counts in the first year they apply during the cycle, which sets the budget for all five years. End-user devices — student laptops, tablets, Chromebooks — are not eligible under either category, which means the device half of a 1:1 program needs a different funding source than the network half.

To be explicit about our own position: Uniqcli is not an E-rate consultant or filing agent, and nothing here is a statement that any particular purchase is or is not eligible. Eligibility is determined by the Eligible Services List and by your own filing, with your consultant. We are describing the program's structure because it changes how a district sequences a rollout, and for nothing else.

The problems that produce a bad first Monday are rarely bandwidth. They are the services nobody load-tested.

DHCP scope is first. A /24 holds fewer than 254 usable addresses; a building with 900 devices needs scopes sized accordingly, and lease times short enough that addresses return to the pool but long enough that the server is not renegotiating constantly. Scope exhaustion presents to a teacher as "the internet is down" and to a network engineer as a phone call at 10:15.

Identity is second. If the district authenticates devices with certificates or 802.1X, the enrollment burst on rollout morning is a load event for the certificate authority and the RADIUS servers, not a background task. Stagger it — enroll by building, by grade band, by day — rather than discovering the ceiling with every student watching.

Third is management: an MDM or device-management platform pushing initial policy and application payloads to a whole fleet on the same morning will saturate the circuit you just sized correctly for steady state. Pre-stage as much as possible before devices ship, and schedule the rest.

Fourth, segmentation. One flat VLAN for student devices, staff devices, printers, cameras and building systems is a performance problem and a security problem at the same time. Segment before the devices land; retrofitting segmentation into a live 1:1 building is significantly harder than doing it in July.