Dell Networking S5048F-ON 25GbE Switch

The Dell Networking S5048F-ON is a 1RU top-of-rack switch with 48 line-rate 25GbE SFP28 server ports and six line-rate 100GbE QSFP28 uplinks, on a 3.6 Tbps non-blocking fabric rated to 2,678 Mpps. It was Dell's first native 25GbE server-access switch, and it remains the cheapest way onto 25GbE on the secondary market.

Dell built it to solve a specific migration problem: how to move a rack from 10GbE to 25GbE server ports without rewiring anything, and how to uplink that rack at 100GbE using the same generation of optics.

Why 25GbE replaced 10GbE for server access

A 25GbE SFP28 port and a 10GbE SFP+ port are the same physical cage, the same single electrical lane, and the same one-cable-per-server structured cabling. The only 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 rather than a step up to 40GbE, where the same bandwidth costs four electrical lanes, a fatter cable and a QSFP+ cage on every server. Dell states the design goal plainly: 25GbE is backward compatible to 10G and 1G, and the switch supports mixed 25G and 10G servers on front-panel ports without limitation.

The practical consequence for a buyer is that an S5048F-ON drops into an existing 10GbE rack today, using the 10GbE optics and DACs already in the building, and grows into 25GbE as servers get replaced. You do not need to flip the whole rack on day one.

Uplinks, breakout and oversubscription

The six 100GbE QSFP28 uplinks are multi-rate and 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. Use all six for breakout and you get 24 additional 25GbE ports, for a maximum of 72 - which is where Dell's "up to 72 ports of 25GbE" figure comes from, and which is worth understanding because it is often mistaken for a physical port count. The chassis has 48 SFP28 ports and 6 QSFP28 ports. Everything above that is breakout.

Left as uplinks, six 100GbE ports give 600 Gbps toward the spine against 1.2 Tbps of access bandwidth - a 2:1 oversubscription ratio. For general-purpose virtualization, web tiers and most enterprise workloads that is a conservative and well-tested leaf design point. Most builds keep four uplinks to the spine and break out the remaining two for a handful of dense nodes.

The single-generation consistency is the underrated part. Access ports and uplink lanes are both 25G serdes, so the optics, the DACs and the breakout cables are all mainstream 25G-lane parts rather than an orphaned speed. That is what keeps this family economic to operate second-hand.

Fabric, buffer and forwarding

The switch runs a 3.6 Tbps full-duplex, non-blocking, store-and-forward fabric with a forwarding capacity of up to 2,678 Mpps. Store-and-forward means the switch buffers a complete frame before forwarding it, which adds serialization delay proportional to frame size. Dell qualifies the non-blocking claim to packet sizes larger than 250 bytes, which is worth knowing if your traffic profile is dominated by small packets.

The packet buffer is 22 MB, with Dell noting that 16 MB was supported in the initial software release. Behind it sit 8 GB of CPU memory, a 132K MAC table in scaled-l2-switch mode, an 82K ARP table in scaled-l3-hosts mode, up to 128K IPv4 routes and up to 64K IPv6 routes. Those are configurable-profile numbers rather than simultaneous maxima - the switch trades table space between L2 and L3 depending on the mode you select, which is a normal design in this generation and worth planning around if you intend to run a large overlay.

How it differs from the S5148F-ON

Dell shipped a second switch with an identical port layout - 48x 25GbE SFP28 and 6x 100GbE QSFP28 in 1RU on the same 3.6 Tbps fabric - about a year later. The differences are entirely internal:

Read that table honestly and the S5048F-ON wins on two counts - it publishes a forwarding rate and it has a larger packet buffer - and loses on table scale by a wide margin. If you are running a straightforward Layer 2 and Layer 3 leaf with a few thousand endpoints, this is the more common and usually cheaper part, and nothing about its tables will constrain you. If you are building a VXLAN overlay where a leaf learns remote MAC addresses from across the whole fabric, or a multi-tenant environment where every tenant's virtual machines share the same hardware tables, the S5148F-ON is the one with headroom.

We do not publish an ASIC name for either switch, because Dell does not publish one on either specification sheet.

Airflow, and why it is fixed at order time

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.

Dell builds this model in both directions and lists them as separate part numbers, with direction-specific power supplies and fan modules under their own part numbers again. Airflow is therefore a property of the physical chassis, not a setting, and a fan module pulled from a reverse-airflow unit will not convert this one.

Worth knowing before you order: unlike the S5148F-ON, we do not stock a reverse-airflow S5048F-ON. If your rack faces the hot aisle and you need reverse airflow at 25GbE, the closest equivalents we list are the S5148F-ON reverse-airflow, which has the same port layout, or the current-generation S5248F-ON. Ask us either way - reverse-airflow units of this model do exist and we can source them.

OS9, ONIE and what a licence means second-hand

The S5048F-ON is an ONIE switch, and that is the most important thing to understand before buying one used. Dell decouples the hardware from the network operating system: the Open Network Install Environment on the box supports zero-touch installation of an alternate network operating system, and the chassis does not care which one you run. The factory software is Dell Networking OS9, which brings the full routing stack - OSPFv2/v3, BGP-4 with IPv6 extensions, RIP, VRRP, PIM-SM and SSM, IGMP, MSDP and VRF-lite - plus the Open Automation Framework and an embedded Puppet agent for configuration management.

This model predates OS10. Dell's specification for the S5048F-ON names OS9 and third-party network operating systems and does not name OS10 anywhere. If OS10 is a requirement for your environment, that points you at the S5148F-ON or the current-generation S5200 series instead.

What you are buying from us is the hardware. Dell software entitlements are bound to a unit's service tag and to a Dell account and do not automatically travel with a switch on the secondary market. A refurbished unit may arrive with its factory software 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. Tell us what you intend to run and we will confirm the state of a specific unit before it ships.

Storage and converged traffic

Priority flow control (802.1Qbb), data center bridging exchange (DCBx) and enhanced transmission selection (802.1Qaz) are all implemented, along with the DCBx application TLV for iSCSI and explicit congestion notification. That combination is what makes a lossless iSCSI SAN workable over the same fabric as ordinary traffic, and it is the reason Dell positions this switch for converged deployments. FCoE transit via FIP snooping is supported for environments still bridging to Fibre Channel. The switch also acts as a high-speed VXLAN Layer 2 gateway, connecting hypervisor-based overlay networks to non-virtualized infrastructure.

Common questions

Does the S5048F-ON really have 72 ports?

No. It has 48 SFP28 ports and 6 QSFP28 ports. The 72 figure is the maximum 25GbE density with all six QSFP28 uplinks broken out into four SFP28 ports each, which leaves no uplink capacity at all.

Can it run 10GbE servers?

Yes. The SFP28 ports are multi-rate and accept 1GbE and 10GbE optics and DACs. Dell supports mixed 25G and 10G servers on the front panel without limitation, and this sheet's optics list even includes a 10GBASE-T SFP+ copper transceiver.

What is the oversubscription ratio?

2:1 with all six uplinks in use - 1.2 Tbps of access bandwidth against 600 Gbps toward the spine.

Does it stack?

Not in the legacy stacking sense. It uses Virtual Link Trunking and multiple VLT for multi-chassis active-active pairs, which is the modern equivalent and does not share a control plane the way a stack does.

Should I buy this or the S5148F-ON?

Same ports, same fabric. Take the S5048F-ON for a conventional leaf where a 132K MAC table is ample and the larger 22 MB buffer helps. Take the S5148F-ON for VXLAN overlays, multi-tenant fabrics, latency-sensitive workloads, or if you need OS10. See the comparison table above.

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 an existing S5000-series estate, or you want 25GbE at the lowest entry cost, this is the model to buy.

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 state, a particular optics loadout, reverse airflow, or matched pairs for a VLT domain, tell us before you order and we will build to it.

Specifications