The Dell Networking S3048-ON is a 1RU open-networking top-of-rack switch with 48 line-rate 1000BASE-T ports, four line-rate 10GbE SFP+ ports and a non-blocking 260Gbps full-duplex switch fabric, supplied with Dell Networking OS9.X and ONIE for installing an alternate network operating system.
Two jobs this switch does well
1GbE top-of-rack server aggregation. This is the role Dell designed it for: forty-eight 1000BASE-T ports facing a rack of servers whose onboard NICs are gigabit, with four 10GbE SFP+ ports going up to an S-series or Z-series spine. Forty-eight gigabits of access capacity against forty gigabits of uplink is a 1.2:1 oversubscription ratio, and the 260Gbps fabric is non-blocking across every port at every frame size, so the uplink is the only place contention can occur. Latency is 3.7 microseconds on the copper ports and roughly 1.8 microseconds on SFP+.
Out-of-band management and console aggregation. In practice this is where most S3048-ON units end up, and it is worth being explicit about because buyers search for it. Every server iDRAC, every storage controller management port, every PDU and every switch management interface in a row of racks is a 1GbE copper port, and they all need to land somewhere that is reachable when the production fabric is down. Forty-eight copper ports plus four 10GbE uplinks in one rack unit is exactly the right shape for that job, and Dell's own successor in this line - the S3248T-ON - is positioned explicitly for data center management and out-of-band orchestration. The S3048-ON is the affordable secondary-market way to build the same thing.
A third role follows from the first two: campus or branch distribution where the requirement is a lot of gigabit copper, real Layer 3, and 10Gb uplinks, but not PoE. If you need PoE at this port density, the campus-oriented Dell Networking S3148P is the right switch instead.
Airflow: standard and reverse builds
This listing covers both airflow builds of the S3048-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)
The S3048-ON is built in two cooling variants and the specification sheet is unambiguous that in-field change of airflow direction is not supported. The fans and power supplies are ordered as a matched set and the direction is fixed for the life of the unit. This page is the I/O-panel-to-PSU variant - forward airflow - where cold air is drawn in across the port face and exhausted at the power-supply end.
Both variants use hot-swappable redundant power supplies and three hot-swappable fans, accept 90-264 VAC at 50/60Hz, draw a maximum of 87W (65W typical), emit at most 290 BTU/h and run below 36 dBA - quiet enough for a comms room adjacent to occupied space.
Reverse airflow (PSU to I/O)
A data center rack has a cold aisle at the front and a hot aisle at the back. Servers obey that convention: intake at the front bezel, exhaust at the rear. Switches frequently do not, because a top-of-rack switch is usually mounted with its port face toward the rear of the rack so that the patch cabling stays in the same aisle as the server NICs it connects to.
Mount a forward-airflow switch that way and you have built a heat loop: the switch's intake is now in the hot aisle, drawing the exhaust of every server below it, and pushing its own exhaust out of the front into the cold aisle. Intake temperature climbs, the fans spin up permanently, and component life falls.
The reverse-airflow variant fixes that by inverting the path: air is drawn in at the power-supply end - which is in the cold aisle when the switch is mounted ports-rearward - and exhausted across the I/O panel into the hot aisle. The switch now moves air in the same direction as everything else in the rack.
Apart from the direction of the air, the two variants are the same switch. Identical ports, identical 260Gbps non-blocking fabric, identical 131Mpps forwarding rate, identical OS9.X and ONIE support, identical power envelope: 90-264 VAC at 50/60Hz, 87W maximum and 65W typical draw, at most 290 BTU/h, under 36 dBA.
Open networking: what the -ON actually buys you
The S3048-ON was the first 1GbE enterprise switching platform to ship both a hardened vendor OS and open-networking support on the same hardware, and that dual nature is the entire point of the product. It arrives with Dell Networking OS9.X - a mature, feature-complete image with OSPFv2/v3, BGP-4 with IPv6 multiprotocol extensions, IS-IS, RIP, VRRP, PIM-SM and IGMP, 16K IPv4 routes, 8K IPv6 routes and 64 VRF-lite instances for multi-tenant Layer 3 isolation including multicast and IPv6.
It also ships with ONIE, the Open Network Install Environment, which is the boot loader that makes the hardware operating-system agnostic. ONIE is what allows a third-party network operating system such as Cumulus Linux to be installed with zero touch, and it is the mechanism by which any ONIE-targeted NOS reaches this platform. If your estate is standardizing on a Linux-based NOS rather than a vendor CLI, this is a switch you can move onto that standard rather than replace.
On the automation side the platform carries Dell's Embedded Open Automation Framework, a Puppet agent and REST API extensions, so configuration and provisioning can be driven from the same tooling that builds the servers underneath it. OpenFlow 1.3 is supported and interoperates with standard OpenFlow controllers. For Layer 2 extension, VLT, Routed VLT and VLT Proxy Gateway allow a VLAN to be stretched within or across two data centers without spanning-tree blocking, which is what makes VM mobility regions larger than a single rack practical.
Uplinks, stacking, and how to spend the four 10GbE ports
The four SFP+ ports are the interesting design decision on this switch, because they are not dedicated to a single purpose. Each one can be an uplink to the spine or core, a stacking port to another S3048-ON, or a mix of both, and you decide the split per deployment.
- All four as uplinks: 40Gbps to the core against 48Gbps of access capacity, a 1.2:1 ratio. This is the right answer for a top-of-rack switch in a leaf-spine or Active Fabric design.
- Two up, two across: 20Gbps to the core and 20Gbps of stack interconnect. Useful when two switches in adjacent racks need to behave as one logical device but the north-south load is modest.
- Stacking-heavy: up to six units are supported in a single user-port stack, managed as one logical device, with link aggregation supporting 16 links per group and 128 groups per stack.
Because stacking runs over the same SFP+ ports as the uplinks, every stack member you add costs uplink capacity. That is the trade-off against the campus S3100 series, which has dedicated rear stacking ports and does not consume uplinks to stack - but which tops out at OS9 and offers no open-networking path. Choose on the basis of which constraint matters more to the deployment.
Optics and cabling are ordered separately: SFP 100BASE-FX, 1000BASE-T, SX, LX and ZX for the gigabit side, and SFP+ 10GbE SR, LR, LRM, ER and ZR for the uplinks, plus SFP+ copper twinax direct-attach cables from 0.5m to 7m and a 15m active optical cable for longer in-row runs.
Related Dell 1GbE switches
The campus counterparts to this switch are the S3100 series, which trade open networking for dedicated rear stacking ports and PoE+ options: S3124, S3124F (fiber), S3124P (PoE+), S3148 and S3148P (48-port PoE+).
Condition, testing and support
Every unit is professionally refurbished: inspected, cleaned, configuration-wiped, firmware-checked and functionally tested on every port before it ships, and it ships with warranty. Confirm the airflow direction you need before ordering - it cannot be changed after the fact - and tell us the optic types your uplinks require so the switch arrives ready to patch. Power supplies, fans, rail kits, console cables and optics are configurable above; for multi-rack quantities or a specific bill of materials, request a quote.
Chassis and Physical
| Chassis | 1RU, Dell ReadyRails |
|---|---|
| Dimensions (H x W x D) | 1.71 in x 17.09 in x 12.6 in (4.4 cm x 43.4 cm x 32.0 cm) |
| Weight | 12.8 lbs (5.84 kg) with one power supply; 14.8 lbs (6.74 kg) with two |
| Console port | 1x RJ45 console/management port with RS232 signaling |
| Airflow options | I/O panel to PSU (forward airflow) on this variant. In-field change of airflow direction is not supported. |
| Redundancy | Hot-swappable redundant power supplies; hot-swappable redundant fans (3 fans); user port stacking up to 6 units |
| Acoustics | ISO 7779 A-weighted sound pressure <36 dBA at 78.8 F (26 C) |
Switching and Routing Performance
| Switch fabric capacity | 260Gbps full-duplex (130Gbps half-duplex) |
|---|---|
| Forwarding capacity | 131Mpps |
| Line-rate switching | Layer 2 and Layer 3, IPv4 and IPv6, all protocols |
| MAC addresses | Up to 80K |
| IPv4 routes | 16K |
| IPv6 routes | 8K (shared CAM space with IPv4) |
| IPv4 host table | Up to 40K |
| IPv6 host table | 8K |
| IPv4 multicast table | 8K |
| Layer 2 VLANs | 4K |
| MSTP instances | 64 |
| VRF-lite instances | 64 |
| Link aggregation | 16 links per group, 128 groups per stack |
| LAG load balancing | Based on Layer 2, IPv4 or IPv6 headers |
| Priority queues per port | 8 |
| Latency | 3.7 microseconds for 1000BASE-T; approximately 1.8 microseconds for SFP+ |
| MTU | 12,000 bytes |
| Packet buffer memory | 4MB |
| CPU memory | 2GB |
Stacking and Optics
| Stacking | User port stacking up to 6 units over the 10GbE SFP+ ports |
|---|---|
| Optics supported | SFP 100BASE-FX / 1000BASE-T / 1000BASE-SX / 1000BASE-LX / 1000BASE-ZX; SFP+ 10GbE SR / LR / LRM / ER / ZR |
| Direct attach cable options | SFP+ to SFP+ 10GbE copper twinax at 0.5m, 1m, 3m, 5m, 7m; 15m active optical cable |
Power and Thermal
| Power input | 90-264 VAC, 50/60 Hz |
|---|---|
| Power consumption | 87W maximum, 65W typical |
| Max thermal output | 290 BTU/h |
| Max current draw | <1A at 100/120 VAC; <0.5A at 200/240 VAC |
Environmental
| Operating temperature | 32 to 113 F (0 to 45 C) |
|---|---|
| Operating humidity | 5 to 85% RH, non-condensing |
| Operating altitude | 0 ft to 10,000 ft above sea level |
| Storage temperature | -40 to 158 F (-40 to 70 C) |
| Storage humidity | 5 to 95% RH, non-condensing |
Software, Protocols and Compliance
| Operating system | Dell Networking OS9.X; ONIE for zero-touch install of an alternate network operating system such as Cumulus Linux |
|---|---|
| Automation | Embedded Open Automation Framework, Puppet agent, REST API extensions |
| Virtualization / L2 extension | VLT, Routed VLT, VLT Proxy Gateway; VRF-lite for multi-tenant L3 isolation including multicast and IPv6 |
| Routing protocols | OSPFv2/v3, BGP-4 with IPv6 extensions, IS-IS, RIPv1/v2, VRRP, PIM-SM, IGMPv1/v2/v3, MSDP |
| IEEE compliance | 802.1AB, 802.1D, 802.1p, 802.1Q, 802.1s, 802.1w, 802.1X, 802.3ab, 802.3ac, 802.3ad, 802.3ae, 802.3az, 802.3u, 802.3x, 802.3z, ANSI/TIA-1057 LLDP-MED, Force10 PVST+ |
| Monitoring | sFlow v5, RMON / RMONv2, SNMP v1/v2c/v3, TACACS+, RADIUS |
| SDN | OpenFlow 1.3, interoperable with industry standard OpenFlow controllers |
| Certifications | TAA available, USGv6 Host and Router certified on Dell Networking OS 9.7 and greater, IPv6 Ready (Host and Router), UCR DoD APL core and distribution ASLAN switch, EU RoHS compliant |

