Dell Networking S5148F-ON 25GbE Switch

The Dell Networking S5148F-ON is a 1RU top-of-rack switch with 48 line-rate 25GbE SFP28 server ports and six line-rate 100GbE QSFP28 uplinks, built on a 3.6 Tbps cut-through fabric with a programmable data plane. It is the OS10-generation sibling of the S5048F-ON and the more capable of the two on paper.

Dell shipped two switches with an identical port layout one year apart. Choosing between them is not about density - it is about what happens inside the forwarding pipeline, and about which network operating system you intend to run.

What "programmable data plane" actually means here

Most fixed-function switch silicon implements a pipeline decided at tape-out: a fixed set of headers it can parse, a fixed set of actions it can take. If a new encapsulation arrives after the chip ships, the chip cannot learn it, and support has to come from a later hardware generation.

The S5148F-ON is built differently. Dell's specification for this model calls out programmable packet modification and forwarding and programmable packet mirroring and multi-pathing, and states that the switch offers hardware support for future and emerging protocols, including hardware-based VXLAN as both a Layer 2 and a Layer 3 gateway. Dell positions it explicitly for "use cases that require customization to any packet processing steps or supporting new protocols".

The S5048F-ON specification makes no programmability claim of any kind. That single difference is the reason both models exist, and it is the honest way to describe the gap. We do not publish an ASIC vendor or part number for either switch, because Dell does not publish one on either specification sheet, and we would rather leave a gap than guess at silicon.

The practical consequence for a buyer is narrower than the marketing implies. If you are running standard Layer 2 and Layer 3 with a conventional VXLAN overlay, both switches do the job and the S5048F-ON is the more common part on the secondary market. If you are running a custom encapsulation, an in-house telemetry format, or a fabric that leans on non-standard header handling, the S5148F-ON is the one with the mechanism to support it.

Cut-through versus store-and-forward, and the scale gap

Both switches share a 3.6 Tbps full-duplex fabric, but they move packets differently. The S5148F-ON is cut-through: it begins forwarding as soon as it has read enough of the header to make a decision, so latency stays roughly flat as frame size grows. The S5048F-ON is store-and-forward: it buffers the whole frame before forwarding, which adds serialization delay proportional to frame size. For most enterprise traffic this is invisible. For RDMA, financial messaging and tightly-coupled compute it is the difference that matters.

The table scale is the other real gap, and it is large:

A 512K MAC table on a 48-port leaf is not there for the 48 servers in front of it. It is there for VXLAN, where a leaf learns remote MAC addresses from across the whole fabric, and for multi-tenant environments where every tenant's virtual machines land in the same hardware table. If you are building an overlay of any size, table scale is what decides whether a leaf ages gracefully or starts punting traffic to software.

25GbE server access, and why it replaced 10GbE

An SFP28 port and an SFP+ port are the same physical cage, the same single electrical lane and the same one-cable-per-server cabling. The difference is the lane rate. Moving a rack from 10GbE to 25GbE access multiplies usable bandwidth per port by 2.5x without adding a port, a fibre run or a rack unit. That is why 25GbE became the default server-access speed instead of a jump to 40GbE, where the same bandwidth costs four lanes and a fatter cable.

The 48 SFP28 ports on this switch are multi-rate and accept 1GbE and 10GbE optics and DACs, so a mixed 10/25GbE rack is a supported configuration rather than a workaround. Dell states this directly: the switch supports mixed 25G and 10G servers on front-panel ports without limitation. That matters on the secondary market, because it means an S5148F-ON can be dropped into an existing 10GbE rack today and grow into 25GbE as servers get replaced.

Uplinks, breakout and oversubscription

The six 100GbE QSFP28 uplinks are multi-rate and can also run at 10, 25, 40 or 50GbE. Each is four 25GbE lanes in one cage, so a breakout cable turns one QSFP28 port into four SFP28 ports. Spend all six on breakout and you get 24 additional 25GbE ports, for a maximum of 72 - which is where the "up to 72 ports of 25GbE" figure in Dell's own literature comes from. Dell also documents the switch in 16, 28, 40, 52 and 64 port 10GbE configurations depending on how you split access and uplink.

Left as uplinks, six 100GbE ports give 600 Gbps toward the spine against 1.2 Tbps of access bandwidth - a 2:1 oversubscription ratio, which is a conservative and well-tested leaf design point. Most deployments keep four uplinks to the spine and break out the other two for a handful of dense nodes.

Airflow: standard and reverse builds

This listing covers both airflow builds of the S5148F-ON. Select the direction under the Airflow option: air direction is set by the power supplies and fan modules physically fitted, so it is chosen when the unit is built rather than switched afterwards. Power supplies and fans must both face the same way - a mismatched pair raises the switch's incompatible-airflow alarm.

Standard airflow (I/O to PSU)

This SKU is the I/O-panel-to-PSU chassis - Dell's standard airflow direction. Cool air enters at the port face and exhausts at the power-supply and fan face. In a hot-aisle/cold-aisle row, that is the correct choice when the switch is racked with its ports facing the cold aisle.

Reverse airflow (PSU to I/O)

Dell names the two airflow directions by where the air goes. I/O Panel to PSU airflow is the standard build: cool air enters at the port face and exhausts at the power-supply and fan face. PSU to I/O Panel airflow is the reverse, drawing air in at the power-supply face and exhausting it out of the ports. This -RA SKU is the PSU-to-I/O-Panel unit.

Which one you need is decided by which way the switch faces in the rack, not by preference. In a hot-aisle/cold-aisle row, servers pull cool air from the cold aisle at the front and exhaust into the hot aisle at the back. If you rack a top-of-rack switch with its port face toward the cold aisle, standard airflow is correct. If you rack it with its port face toward the hot aisle - which is extremely common, because it puts the switch ports on the same side as the server NICs and halves your cable runs - then a standard-airflow switch would be inhaling your own hot exhaust from the first day, running hot and spinning its fans harder for the rest of its service life.

Two practical consequences. First, power supplies and fan modules are direction-specific. Dell's own specification for this model lists the power supply and the fan module twice, once per direction, under separate part numbers. A fan module pulled from a standard-airflow unit will not convert this chassis, and mixing directions inside a single unit is not a supported configuration. Second, mixing directions within one row is the most frequent cause of a "this switch runs hot" complaint that has nothing to do with the switch - it is simply pushing hot exhaust into a cold aisle.

A quick field test if you have the unit in front of you: stand at the port face with the switch powered on. If you can feel air blowing out at you, it is a reverse-airflow unit. If you feel air being pulled in, it is standard airflow.

This switch draws up to 516W and dissipates up to 1,956 BTU/h, with a typical draw of 421W once optics are loaded. On a fully populated 48-port leaf the optics are a meaningful share of that number, which is why the typical figure sits so close to the maximum. Dell rates the unit Fresh Air compliant to 45 C and specifies operation from 0 to 45 C at 5 to 90% relative humidity.

The reason this matters specifically for a reverse-airflow unit: it is exhausting into the aisle your patch panels and technicians are on. In a cabinet with dense horizontal cabling and no server airflow to help, verify the exhaust path is genuinely open and not blocked by a cable manager or a blanking panel before you commit.

OS10, ONIE and what a licence means second-hand

The S5148F-ON is an ONIE switch. Dell decouples the hardware from the network operating system, and the Open Network Install Environment on the box will install an alternate NOS with zero touch. The factory software is Dell EMC Networking OS10, which ships in an Open Edition decoupled from the L2/L3 protocol stack and a Premium Edition that adds Dell's switching and routing protocols, IP services, QoS and automation. Underneath, OS10 is an unmodified Debian Linux 8.4 distribution on a stock 3.16 kernel, which is what makes standard Linux tooling and scripting work on the switch itself.

What you are buying from us is the hardware. A Dell OS10 entitlement is a licence bound to the unit's service tag and to a Dell account, and it does not automatically travel with a switch on the secondary market. A refurbished unit may arrive with OS10 installed and fully functional and still carry no active entitlement or support contract. We do not promise a transferred Dell licence, a software subscription or a support contract with a refurbished switch, and you should treat any reseller who does with suspicion.

Plan the NOS as a separate line item: budget an OS10 subscription from Dell if you need vendor support and TAC escalation, or run an ONIE-compatible alternative if you are comfortable owning the software stack. Tell us what you intend to run and we will tell you the state of a specific unit before it ships.

Storage and converged traffic

The switch implements the full Data Center Bridging set - priority flow control (802.1Qbb), enhanced transmission selection (802.1Qaz), DCBx and the DCBx application TLV for iSCSI - which is what makes lossless iSCSI SANs and RoCE workable on a shared fabric. Dell also calls out rogue-NIC control, a hardware protection against a misbehaving server NIC flooding pause frames and stalling everything attached to the switch. FCoE transit via FIP snooping is supported for environments still bridging to Fibre Channel.

Common questions

What is the difference between the S5148F-ON and the S5048F-ON?

Identical port layout - 48x 25GbE SFP28 and 6x 100GbE QSFP28 in 1RU on a 3.6 Tbps fabric. The S5148F-ON is cut-through with a programmable data plane, runs OS10, and has far larger MAC, ARP and multicast tables. The S5048F-ON is store-and-forward, runs OS9, and publishes a 2,678 Mpps forwarding rate. See the comparison table above.

How many servers can one S5148F-ON support?

48 at one 25GbE port each, or 24 dual-homed for NIC-level redundancy. A VLT pair gives 48 dual-homed servers with no spanning-tree blocked links.

Can it run 10GbE servers?

Yes. The SFP28 ports are multi-rate and take 1GbE and 10GbE optics and DACs. Dell supports mixed 25G and 10G servers on the front panel without limitation.

Does it stack?

Not in the legacy stacking sense. It uses Virtual Link Trunking for multi-chassis active-active pairs and for stretching a Layer 2 VLAN within or across two data centres, which is the modern equivalent and does not share a control plane the way a stack does.

What is the maximum port count?

72 ports of 25GbE, using all six QSFP28 uplinks as 4x25GbE breakout. That leaves no uplink capacity, so it is a design for a standalone or edge switch rather than a leaf.

Where it sits in the Dell S-series

If you are starting a 25GbE build from scratch and want current-generation software support, the S5248F-ON is the newer platform. If you are extending or replacing an existing S5000-series estate, staying on this generation keeps the optics, DACs and operational tooling you already own.

Condition, testing and what ships in the box

This is a refurbished, professionally reconditioned unit. Every switch is powered up, inspected and functionally tested before it leaves us, and it ships with warranty coverage. Power supplies, fan modules, rail kits, console cables and optics are configurable above - choose the airflow direction that matches the chassis, because power supplies and fans are direction-specific parts.

If you need a specific firmware or NOS state, a particular optics loadout, or matched pairs for a VLT domain, tell us before you order and we will build to it.

Specifications