Dell PowerSwitch S4348F-ON 10GbE Switch

The Dell PowerSwitch S4348F-ON is a 1RU top-of-rack switch with 48 ports of 10GbE SFP+ and six multi-rate 100GbE QSFP28 uplinks, forwarding at 1001.7 Mpps across a 1.08 Tb/s non-blocking fabric with a 32MB buffer, sub-650 nanosecond latency and hardware VXLAN gateway support, running Enterprise SONiC Distribution by Dell Technologies.

It is Dell's current-generation answer to a question most of the industry stopped asking out loud: what do you buy when the servers in the rack are still 10 gigabit?

10GbE to the server did not go away

Every switch launch for the last several years has been about 25, 100 and 400 gigabit, and if you only read launch material you would conclude that 10GbE server access is finished. Walk into the racks and it plainly is not. A very large installed base of servers has 10GbE LOM or a 10GbE adapter, will be in service for years, and does not saturate what it already has.

What has changed is not the access speed - it is everything the rack has to connect to. The spine runs at 100GbE. The storage fabric expects to be reached at 100GbE. The overlay is VXLAN and the switch is expected to terminate it in hardware rather than punt it to software. The control plane is expected to be automated rather than typed.

The S4348F-ON is built for that exact gap: unchanged 10GbE server access, with an uplink, buffer, control plane and software model that belong to the current generation.

What actually changed from the S4148F-ON

The obvious comparison is the S4148F-ON, which occupies the same position in the previous generation and is still the more familiar part number. Both are 1RU with 48 ports of 10GbE SFP+. The differences are in what sits above and behind those ports.

The uplinks. The S4148F-ON carries four 100GbE QSFP28 plus two 40GbE QSFP+ - 480 Gbps of uplink capacity split across two port types. The S4348F-ON carries six 100GbE QSFP28 and no QSFP+ at all: 600 Gbps, one port type, one optics inventory. 40GbE has not disappeared, it has just stopped being a dedicated port - the 100GbE ports are multi-rate and will run 10, 25, 40 or 100.

The buffer. 12MB becomes 32MB, with Dell's integrated SmartBuffer. On a top-of-rack switch this is the number that quietly decides how a rack behaves under incast - many servers answering one request at once, all bursting into the same uplink. A bigger shared buffer absorbs more of that before anything is dropped, which for storage and distributed-database traffic is the difference between a latency graph with spikes and a latency graph without.

The control plane. 4GB of CPU memory becomes 16GB. That is not a forwarding-performance figure; it is what makes a modern software stack, telemetry and container-based network operating systems practical on the box.

The software. The S4100 series is an OS10 platform. The S4348-ON series specification refers only to Enterprise SONiC Distribution by Dell Technologies for its operating system, with ONIE for zero-touch installation. If you are standardising an estate, that is a decision to make deliberately rather than to inherit.

One caution on the headline numbers, because it is a trap. Dell states this switch's fabric as 1.08 Tb/s half-duplex; the older S4100 sheets state their fabric on the full-duplex convention. Compared like for like the S4348F-ON is the larger fabric - 1.08 Tb/s half-duplex is 2.16 Tb/s full-duplex - but a straight numeric comparison between the two documents is not valid, and the same caution applies to the packet-per-second figures. The port arithmetic is the comparison that needs no interpretation: 48 by 10G plus 6 by 100G, all of it line-rate and non-blocking under full load.

Six multi-rate 100GbE ports and what to do with them

The six QSFP28 ports support 10, 25, 40 and 100 gigabit. That flexibility is the design's most useful property and it is worth being concrete about how it gets used.

As straight 100GbE uplinks, six ports is generous for a 480 Gbps access layer - enough for a redundant pair of uplinks to two spines with ports left over, without the uplink being the constraint during a failure.

As leaf interconnect, they let a small fabric be built out of these switches directly rather than requiring a separate spine tier - Dell explicitly positions the model for small-scale leaf-and-spine implementations alongside other S-series 1/10GbE top-of-rack switches.

As a migration ramp, running some of them at 25 or 40 gigabit lets the rack land in a network that has not finished moving to 100 yet, without buying a different switch for the interim. Dell publishes a maximum 10GbE density of 60 for this switch, which is the access ports plus what the uplink ports can be turned into.

On attach: the SFP+ access ports take 10GBASE-SR optics or passive 10GBASE-CU direct-attach copper, and the 100GbE ports are offered with 100GBASE-CR4 passive DAC. If you need 100G over distance, or breakout from a 100G port into four 25G links, tell us at quotation - both are sourcing conversations rather than menu selections.

VXLAN in hardware, and why it is on the spec sheet twice

Dell calls out hardware support for L2 and L3 VXLAN gateway as unique to this series among its 10GbE peers, and it is the feature that most changes what the switch can be asked to do.

VXLAN carries Layer 2 segments across a routed underlay, which is how a modern data centre gets virtual-machine mobility and tenant separation without stretching real Layer 2 domains across the building. Terminating it in hardware, at line rate, for both bridging and routing means the top-of-rack switch can be the boundary between the virtualised overlay and everything that is not virtualised - bare-metal servers, storage arrays, appliances - without a software gateway becoming the bottleneck.

The scale behind it is published and worth checking against your design: 4K Layer 2 VLANs, 4K Layer 3 interfaces per system, up to 1000 VRFs, 224K MAC addresses in L2 mode and 64K in L3 mode, 32K IPv6 hosts and 32K IPv6 routes rising to 65K in Route Scale mode, and 16K ARP entries. A thousand VRFs on a top-of-rack switch is a multi-tenancy number, not a checkbox.

Two other capabilities matter for specific buyers. IEEE 1588v2 is supported in hardware, which is what makes precision time distribution credible for financial, broadcast and industrial-measurement workloads. And sub-650 nanosecond latency puts this switch in a bracket where the switch is not the interesting term in a latency budget.

A note on software: a Dell software entitlement is bound to the unit's service tag and to a Dell account and does not automatically transfer on the secondary market. A refurbished unit may arrive with a working image installed and without an active entitlement or support contract. We do not promise a transferred Dell licence, a software subscription or a support contract with a refurbished switch. Tell us what you intend to run and we will confirm the state of a specific unit before it ships.

Airflow: choose it, and choose it once

The S4348-ON is built in both directions - I/O panel to PSU and PSU to I/O panel - and Dell offers the power supply and the fan module as direction-specific parts in each. That is deliberate: hot-aisle and cold-aisle layouts need opposite directions, and a switch pulling air the wrong way through a contained aisle is a thermal problem that gets blamed on everything except the switch.

Two practical rules. First, every airflow part in one chassis must match - five fan trays and two power supplies, all in the same direction. Mixed directions inside one box is the single most common and most avoidable way to make a switch run hot. Second, match the switch to the rack it is going into, not to the switch two rack units below it, unless those two are in the same aisle orientation.

Because both directions are power-supply and fan swaps on the same chassis, we sell one model page and you choose the direction as part of the build. Confirm it with your quote request - it is decided at build time, not in the rack.

Power and cooling

Typical operating power is 119W and maximum is 362W, with 1076.97 BTU/hr of thermal output at maximum. For a 48-port 10GbE switch with 600 Gbps of uplink that is a modest figure, and the gap between typical and maximum is worth noting when you are sizing a rack PDU: the realistic draw is far below the plate rating.

Power supplies are hot-swappable and redundant, 90-240 VAC at 50/60 Hz, with DC supplies (-40V to -72V DC) available as a custom kit for DC plant. Cooling is five hot-swappable fan trays with two fans each, so a single fan failure is neither an outage nor an emergency.

Common questions

Can I stack these?

No. The six 100GbE QSFP28 ports are uplinks. Rack-level redundancy is built from link aggregation - Dell publishes 128 port channels per system with up to 32 members per channel - with multi-chassis behaviour depending on the Enterprise SONiC release you run. Confirm the exact feature set against your intended software version.

Is this a replacement for the S4148F-ON?

It occupies the same position with more uplink capacity, a much larger buffer and a far larger control plane, but it is not a drop-in: the operating system is different and there are no 40GbE QSFP+ ports. If you are extending an existing OS10 estate, the S4148F-ON keeps the fleet uniform. If you are building the next rack rather than matching the last one, this is the newer platform.

What is the difference between the S4348F-ON and the S4348T-ON?

Access media only. This model is 48 ports of 10GbE SFP+ for fibre or direct-attach copper; the S4348T-ON is 48 ports of 10GBASE-T on RJ45. Same fabric, same six 100GbE uplinks, same buffer. SFP+ is lower latency (sub-650 ns against sub-3.5 microseconds) and lower power; 10GBASE-T reuses structured copper.

Should I buy this or a 25GbE switch?

Match the switch to the server adapters you actually have. If the fleet is 10GbE, a 25GbE switch buys ports you cannot use and costs more to populate with optics. If the next server refresh is 25GbE, look at the S5248F-ON instead - and note that a 25GbE port runs a 10GbE link, so the question is really about how soon the refresh lands.

Is a TAA-compliant build available?

Dell lists both TAA and NON-TAA builds of each airflow direction. Flag the requirement with your quote request so it can be verified against the specific unit.

Where it sits

Availability, condition and how to buy

This model is supplied to order. Airflow direction, power-supply type, fan configuration and optics loadout are all confirmed at quotation, so use the quote request on this page and tell us which way the aisle runs.

Every unit is refurbished, professionally reconditioned, powered up, inspected and functionally tested before it ships, and it ships with warranty coverage. If you need a specific Enterprise SONiC state, a TAA build, or a matched set for a leaf tier, say so with your request.

Specifications