CONFIGURE & QUOTE - Dell PowerEdge R760 24-Bay 2.5" Drives [16th Gen: New]

The Dell PowerEdge R760 24-Bay 2.5" Hot-Swap is the maximum-density flagship configuration of Dell's 16th-generation 2U dual-socket rack platform: twenty-four 2.5" hot-plug front bays with backplane variants supporting PCIe Gen5 NVMe, the full 4th/5th Generation Xeon Scalable compute stack (up to 64 cores per socket on Emerald Rapids), 32 DDR5 DIMM slots, and PCIe Gen5 throughout. This is the 16th gen direct successor to the 15th gen R750 24-Bay we already document elsewhere in the catalog, with all the generational advantages: DDR5 memory at 4800-5600 MT/s, PCIe Gen5 NVMe bandwidth (double Gen4 per drive), Sapphire Rapids or Emerald Rapids Xeon, current 16th gen security baseline (Silicon Root of Trust), and active Dell ProSupport with new-server warranty options.

The R760 is current-production at Dell. For 16th gen platform currency at maximum SFF density, refurbished R760 24-Bay is the cost-correct call vs. buying new at full list price. Stepping up to the 17th gen R770 (Granite Rapids, PCIe Gen5 refined, DDR5 6400 MT/s, iDRAC10) makes sense only when the workload specifically benefits from those incremental gains or when iDRAC10 is required by tooling; for most 16th-gen-class NVMe workloads, refurbished R760 24-Bay is the right answer at meaningfully lower acquisition cost.

The R760 24-Bay 2.5" is the maximum-density SFF variant of the R760 family. The R760 chassis also ships in 8-Bay 2.5", 16-Bay 2.5", 12-Bay 3.5" LFF, and EDSFF E3.S configurations, but the 24-Bay is the variant most R760 customers deploy when local NVMe density is the design driver: large-scale vSAN ESA clusters, NVMe-oF disaggregated storage targets, all-NVMe Ceph clusters, AI training data tiers with local NVMe scratch, and converged SAS SSD + NVMe database deployments where 24 SFF bays per 2U node is the right design point. Wholesale Servers currently stocks only the 24-Bay R760 variant; for the lower-density R760 chassis variants, contact us to discuss sourcing.

What's Different About This Chassis

1. 24 SFF bays with multiple NVMe-capable backplane variants. The defining R760 24-Bay differentiator. Dell ships three distinct backplane SKUs for the 24-Bay chassis: a Universal backplane (mixed SAS/SATA + universal NVMe-direct slots), a Passive NVMe backplane (all 24 PCIe Gen4 NVMe direct), and a Switched NVMe backplane (all 24 with PCIe switch, Gen5 NVMe support). Specify which backplane at quote time; they are not field-swappable post-purchase.
2. PCIe Gen5 NVMe is the headline storage capability. Up to 14 GB/s sequential read per drive on Gen5 NVMe. The Switched backplane is the variant that surfaces Gen5 NVMe bandwidth across all 24 bays. Aggregate per-node throughput exceeds 300 GB/s theoretical; in practice limited by PCIe fabric, network, and application concurrency.
3. vSAN ESA is the canonical workload. The R760 24-Bay all-NVMe configuration is the highest per-node NVMe density configuration available for vSAN ESA in a standard 16th gen 2U chassis. ESA wants all-NVMe; the 24-Bay R760 delivers it at Gen5 bandwidth. HBA355i pass-through required for ESA (ESA manages drives directly, no RAID controller in the data path).
4. Compute and memory architecture is full R760 flagship. Up to 64-core 5th gen Emerald Rapids per socket, 350W TDP support with DLC, 32 DDR5 DIMM slots, 8 TB max memory with 256 GB RDIMMs. The 24-Bay does not constrain CPU or memory.
5. PCIe slot budget reduced vs. 16-Bay variants. The 24-Bay chassis trades rear PCIe slot budget for the additional front-bay storage; typical 24-Bay configurations have 4-6 PCIe Gen5/Gen4 slots vs. up to 8 on lower-bay-count R760 variants. For GPU-heavy plus 24-NVMe configurations, slot count is the binding constraint to plan around.
6. Direct Liquid Cooling available for high-TDP configurations. Dell offers optional DLC on the R760 for CPUs above 250W TDP and high-density configurations where air cooling cannot maintain the thermal envelope. DLC requires rack manifolds and a cooling distribution unit; verify rack infrastructure supports DLC before specifying.
7. No Smart Flow on 24-Bay. Smart Flow chassis cooling is a 16th gen option specific to the R760 8-Bay and 16-Bay variants where the chassis can sacrifice 2 SFF positions for an airflow grill. The 24-Bay chassis is fully populated and uses standard or HPR Gold fan configurations plus optional DLC for high-TDP CPUs.

Storage — 24 SFF Bays with NVMe-Capable Backplane

Backplane variants

• Universal 24-Bay backplane: Mixed-protocol configuration. Common variants: 16 SAS/SATA + 8 universal NVMe-direct slots, or 24 SAS/SATA with 8 universal slots (NVMe direct on those 8). The Universal backplane supports hardware RAID for SAS/SATA via PERC controllers and direct-attach NVMe (no hardware RAID on the NVMe bays via this backplane). For mixed-tier deployments where partitioning bays by protocol is desired.
• Passive 24-Bay NVMe backplane: All 24 PCIe Gen4 NVMe direct. No PCIe switch; lower latency on individual drives but higher PCIe lane consumption from the CPU. For all-NVMe deployments at Gen4 bandwidth.
• Switched 24-Bay NVMe backplane: All 24 PCIe NVMe with PCIe switch. Supports Gen5 NVMe drives at full Gen5 bandwidth (14 GB/s per drive). For maximum aggregate NVMe throughput; the canonical vSAN ESA cluster building block.

Common configurations

• vSAN ESA all-flash maximum density (Gen5): 24 PCIe Gen5 NVMe drives on the Switched backplane, HBA355i pass-through, 100 GbE networking minimum (ESA at 24 Gen5 NVMe drives saturates 100 GbE under sustained load). vSphere 8.x ESA required. Per-node capacity is the design driver: maximum NVMe drives per chassis = maximum capacity per node = lower cluster node count for a given TB target.
• NVMe-oF target node (Gen5): 24 PCIe Gen5 NVMe drives serving multiple compute hosts via NVMe over Fabrics (RoCE or TCP). 100 GbE or InfiniBand for fabric connectivity. Sub-100 microsecond storage latency to client hosts.
• Ceph all-NVMe OSD nodes: 24 NVMe OSDs per node on the Switched or Passive NVMe backplane. Bluestore on HBA355i pass-through. Memory budget: 4-8 GB per OSD plus headroom = 192-256 GB for well-provisioned all-NVMe Ceph nodes.
• Mixed SAS SSD + NVMe (Universal backplane): 16 high-endurance SAS SSDs at RAID 10 on PERC H965i for write-intensive database tier, 8 NVMe drives for hot-tier transaction log or temp tablespace. SQL Server, Oracle, and PostgreSQL deployments where local SSD capacity at multiple performance tiers is the design.
• High-density SAS SSD database storage: 24 mixed-use SAS SSDs at RAID 10 on PERC H965i = 12 drives usable at maximum write endurance. Large SQL Server, Oracle, or SAP deployments requiring substantial local SSD capacity without NVMe latency requirement.
• AI training data tier: 24 Gen5 NVMe drives as local scratch and training data cache alongside up to 2x 350W double-wide GPUs (A100, H100, L40S configurations) on the no-rear-drive riser configuration.

NVMe drive selection

• Gen5 NVMe (the headline capability): Up to 14 GB/s sequential read per drive. Requires the Switched backplane to surface Gen5 bandwidth.
• Gen4 NVMe (universal compatibility): Up to 7 GB/s per drive. Supported on all three backplane variants.
• Mixed-use NVMe (1-3 DWPD): For ESA, write-intensive database, NVMe-oF, Ceph bluestore. Never use read-intensive drives for write-heavy workloads.
• Read-intensive NVMe (0.1-1 DWPD): For ESA capacity tier, read-dominant database, distributed object storage. Lower cost per TB with equivalent read performance.

Every NVMe drive we ship is assessed for remaining endurance via SMART before shipment. Drives with significant endurance consumption are disclosed and priced accordingly.

Boot — BOSS-N1 hot-swap: The R760 ships with BOSS-N1, the 16th gen successor to BOSS-S2. BOSS-N1 supports 2x M.2 NVMe drives in hardware RAID 1, hot-pluggable from the rear of the chassis. Standard configuration on every R760 we ship; all 24 front bays remain for data.

RAID Controllers

• HBA355i (pass-through): Required for vSAN ESA, NVMe-oF, Ceph, and any software-defined storage. The standard choice for the 24-Bay configuration when NVMe is the design driver. Presents all drives directly to the OS; no RAID controller in the data path.
• PERC H965i (Series 12 / PERC12, 8 GB flash-backed cache): The 16th gen flagship hardware RAID controller. Tri-mode SAS4/SATA/Gen4 NVMe support. Handles RAID 0/1/5/6/10/50/60 across SAS, SATA, and Gen4 NVMe in a single controller. The right pick for hardware RAID configurations on the Universal backplane. Note: Gen5 NVMe RAID requires either the Switched backplane with H965i or HBA355i pass-through.
• PERC H755 (PERC11 carryover): 8 GB flash-backed cache. Lower cost than H965i if PERC12-specific features (tri-mode in one card) are not needed. Handles SAS/SATA RAID; NVMe RAID via H755N variant.
• PERC H755N: NVMe-specific PERC11 variant. Hardware RAID for Gen4 NVMe. Less commonly deployed than H965i on 16th gen platforms since H965i covers the same use case more flexibly.
• PERC H355: Entry-tier RAID, no cache. RAID 0/1/10 only; no RAID 5/6 cache acceleration. For boot/log arrays only; not appropriate for production data tiers that need parity.
• HBA355e / H965e: External HBA variants for attaching external storage shelves. The 2U R760 chassis has the PCIe budget to accommodate external HBAs alongside front-bay NVMe and other expansion cards.
• S160 software RAID: Embedded SATA/NVMe software RAID. For boot volumes and low-cost OS configurations only. Not recommended for production data tiers.

PCIe Gen5 Expansion

Up to 8 PCIe slots in the R760 chassis depending on riser configuration. The 24-Bay variant typically deploys with 4-6 slots usable after backplane and front-PERC consumption. PCIe Gen5 doubles Gen4 bandwidth per lane (64 GB/s at x16 vs. 32 GB/s Gen4). For Gen5 NVMe SSDs, Gen5 GPUs, and 200/400 GbE NICs, the bandwidth headroom is meaningful.

Available slot configurations include x8 Gen5, x16 Gen5 (full-height full-length for GPU), x8 Gen4 mixed configurations, and SNAPI x16 Gen5 for high-density single-card configurations. Specific riser config selection happens at order time and depends on workload (GPU configuration, network speed requirement, external HBA needs).

GPU Support

The R760 24-Bay is GPU-capable on chassis configurations that do not have rear drive bays. Dell qualifies up to 2x 350W double-wide GPUs (A100, H100, L40S, A30 configurations) or up to 6x 75W single-wide GPUs (L4, T4, A2) on the appropriate riser configurations.

• NVIDIA H100 (PCIe Gen5, 350W TDP): Maximum AI training and inference performance on this platform. Requires DLC for sustained operation; air cooling pushes the thermal envelope. Standard for AI/ML training data tiers with co-located 24-NVMe storage.
• NVIDIA L40S (PCIe Gen4, 350W TDP): AI inference and graphics-accelerated workloads. Up to 2x in standard riser configurations.
• NVIDIA A100 (PCIe Gen4, 250-300W TDP): Carryover from R750-class GPU deployments. Strong AI training throughput at slightly lower power than H100.
• NVIDIA L4 / T4 / A2 (single-width, 70-72W): Up to 6 single-width GPUs for multi-tenant inference, transcoding, and edge AI workloads on the appropriate riser configuration.

Note: 12x 3.5" LFF and rear-drive R760 configurations explicitly do not support GPU cards per Dell's service manual. The 24-Bay 2.5" with no-rear-drive riser configuration is the GPU-capable R760 chassis variant.

Networking

1 x OCP 3.0 mezzanine slot plus PCIe Gen5 expansion slots. The 24-Bay R760 storage performance ceiling is the network for most deployments. A single PCIe Gen5 NVMe drive can saturate 100 GbE; 24 drives in aggregate can sustain throughput that requires 200-400 GbE to surface.

• Dual-port 25 GbE SFP28 (OCP 3.0): Minimum recommended for production. Acceptable for smaller ESA clusters with modest east-west traffic.
• Dual-port 100 GbE QSFP28: Standard recommendation for the 24-Bay R760 in production. Required for NVMe-oF targets and high-throughput vSAN ESA clusters.
• Dual-port 200 GbE / 400 GbE (where qualified): For the most demanding NVMe-oF and HPC storage targets where Gen5 NVMe aggregate throughput requires the higher-speed fabric.
• DPU / SmartNIC options: NVIDIA BlueField-3 and similar DPU variants available for offloading network and storage processing. Note per Dell documentation: 25 or 100 GbE x 2 configurations may require DPU installation in specific risers.
• 2 x 1 GbE LOM: Standard built-in management network ports alongside the dedicated iDRAC port.

Power Supplies

PSU range: 800W, 1100W, 1400W, 1800W, 2400W, 2800W Platinum or Titanium. The 2800W is required for the most demanding GPU + Gen5-NVMe + high-TDP-CPU combined configurations. Mixed PSU wattages are not supported; both PSUs must match.

Management & Security

iDRAC9 Enterprise required for production deployments. Note that the R760 uses iDRAC9, NOT iDRAC10; iDRAC10 is the 17th gen platform (R670/R770). If your tooling chain specifically requires iDRAC10, the R770 is the platform. For most operational needs, iDRAC9 covers the requirement.

16th gen security baseline: Silicon Root of Trust (cryptographic verification from boot ROM through OS handoff), TPM 2.0 standard, Secure Boot, System Lockdown, Multi-Factor Authentication, Role-Based Access Controls, and Secured Component Verification. Required for federal compliance baselines (FIPS, FedRAMP, DoD). Active Health System v3 monitoring across NVMe, GPU, and platform health.

Physical Specs & Platform Notes

• Form factor: 2U rack, standard 19" rack-mount.
• Chassis dimensions: 86.8 mm H (3.41 in) x 482 mm W (18.97 in) x 772 mm D (30.4 in). Deeper than the R750 (715 mm) and the R650 (1U) chassis; verify rack depth at quote time.
• Cooling: Up to 6 fans. Three cooling tiers: standard (STD) for CPUs to 165W, 2U HPR (high-performance) for CPUs 165-350W non-GPU configurations, Direct Liquid Cooling (DLC) for CPUs above 250W and GPU-heavy configurations. Mixing fan tiers within a single chassis is not supported.
• Bezel: Optional security bezel with LCD.
• Operating temperature: Standard 10-35 degrees C ambient; ASHRAE A2/A3/A4 configurations available for higher-temperature data centers with appropriate fan and DLC selection.
• Maximum chassis weight: Full-loaded R760 24-Bay with all 24 NVMe drives, dual PSUs, and GPU configurations exceeds 75 lbs. Two-person lift mandatory.

Our Assessment

The R760 24-Bay 2.5" is purpose-built for the maximum-storage-density extreme of the 16th gen 2U lineup. Workload profile is specific and demanding: large-scale vSAN ESA at Gen5 NVMe bandwidth, NVMe-oF disaggregated storage targets, all-NVMe Ceph clusters, AI training nodes with co-located NVMe scratch, and converged SAS SSD + NVMe database deployments where 24 SFF bays per 2U node is the right design point. For workloads where 16 SFF bays or fewer cover the requirement, the lower-density R760 chassis variants (8-Bay or 16-Bay) cost less.

Where it falls short of the right answer: workloads where 12 LFF bays of capacity storage is the design driver (the R760 12-Bay 3.5" variant is the correct chassis; note no GPU support on that variant), or workloads where PCIe slot count for GPU plus networking expansion is the binding constraint (the 16-Bay R760 trades 8 drive bays for more PCIe slot headroom on certain riser configurations).

Bottom line: this is the 16th gen 2U platform for maximum NVMe density and AI-class GPU compute. For most current-generation NVMe workloads, the R760 24-Bay is the platform; for cost-primary procurement, the 15th gen R750 24-Bay remains the value play at refurbished pricing.

Workload Fit

Honest Limitations

• Three distinct backplane variants — specify at quote time. Universal (mixed SAS/SATA + NVMe-direct slots), Passive NVMe (all 24 Gen4 NVMe direct), and Switched NVMe (all 24 with PCIe switch, Gen5 support) are different SKUs and not field-swappable. Universal supports hardware RAID for SAS/SATA but only direct-attach NVMe; for Gen5 NVMe hardware RAID, the Switched backplane plus H965i (or HBA355i for pass-through) is the configuration.
• Reduced PCIe slot count vs. 16-Bay R760. 4-6 PCIe slots typical on 24-Bay configurations vs. up to 8 on 16-Bay variants. For deployments combining 24 NVMe with GPU, 100 GbE, and dedicated HBA, the slot budget can be tight; plan PCIe layout at design time.
• No Smart Flow on the 24-Bay chassis. Smart Flow is a 16th gen R760 cooling option only on 8-Bay and 16-Bay variants where 2 SFF positions can be sacrificed for an airflow grill. The 24-Bay is fully populated; high-TDP CPU deployments at warm ambient temperatures require DLC.
• DLC adds infrastructure requirements. Direct Liquid Cooling requires rack manifolds, a cooling distribution unit (CDU), and supply/return plumbing. Verify rack infrastructure supports DLC before specifying. Air-cooled R760 24-Bay caps at roughly 250W CPU TDP at standard ambient temperatures; above that, DLC is mandatory for sustained operation.
• Storage performance ceiling is the network. 24 PCIe Gen5 NVMe drives generate aggregate throughput that 25 GbE cannot surface and 100 GbE can saturate. Plan for 100 GbE minimum on NVMe-serving deployments; 200/400 GbE for sustained-throughput Gen5 NVMe workloads.
• 2800W PSU territory for maximum combined configurations. 24 Gen5 NVMe + dual 350W CPU + 2x 350W GPU pushes into 2800W Titanium PSUs and dedicated power cabling. Verify rack PDU capacity at quote time; this is not a workload that fits standard datacenter PDU planning.
• NVMe drive endurance is a real procurement decision. Mixed-use (1-3 DWPD) vs. read-intensive (0.1-1 DWPD) costs differ significantly at Gen5 NVMe price points. Right-size endurance to workload; do not over-buy and do not under-buy.
• vSAN ESA requires HBA355i pass-through specifically. H965i or H755N in front of ESA capacity drives breaks the platform certification. ESA wants direct drive access, not RAID-fronted drives. We will substitute HBA355i for the H965i on ESA-bound quotes.
• Aggregate NVMe failure events are statistically more likely. 24 drives = 24 failure-domain components. RAID/redundancy planning matters more, not less; mirror or erasure-code appropriately for the failure model.
• Hardware NVMe RAID via H965i is rarely the optimal choice at scale. Software-defined redundancy (vSAN, Ceph, ZFS, mdadm) generally outperforms hardware NVMe RAID controllers, especially at Gen5 bandwidth where the controller becomes the bottleneck. Use H965i only where hardware RAID semantics are explicitly required for the application.
• Full-loaded weight is significant. 24 drives + dual PSU + chassis exceeds 75 lbs. Two-person lift mandatory; consider lift-assist for rack installation above shoulder height.

Generation Context

vs. R750 24-Bay 2.5" (15th gen Ice Lake predecessor): The R760 24-Bay delivers PCIe Gen5 NVMe (vs. Gen4 on R750 = double bandwidth per drive), DDR5 memory (4800-5600 MT/s vs. DDR4 3200 MT/s), 4th/5th Gen Xeon Scalable (Sapphire Rapids/Emerald Rapids vs. Ice Lake), more cores per socket (up to 64 on Emerald Rapids vs. 40 on Ice Lake), 16th gen security baseline (Silicon Root of Trust improvements), BOSS-N1 boot, and active Dell new-server warranty options. For workloads that genuinely use Gen5 NVMe bandwidth or DDR5 capacity, the R760 24-Bay is the platform. For 15th-gen-class storage performance at meaningfully lower cost, the R750 24-Bay 2.5" remains a valid option at refurbished pricing.

vs. R770 (17th gen Granite Rapids successor): R770 brings Granite Rapids Xeon, PCIe Gen5 refined, DDR5 6400 MT/s, iDRAC10, and Granite Rapids on-chip AI acceleration. For workloads that specifically benefit from those incremental gains, R770 earns its premium. Most current 16th-gen NVMe workloads do not yet saturate DDR5 5600 MT/s or PCIe Gen5 at the per-drive level, so for refurbished 16th gen pricing, R760 24-Bay is the cost-correct call.

vs. R760 16-Bay 2.5" / 8-Bay 2.5" / 12-Bay 3.5" siblings (not currently stocked at Wholesale Servers): The 8-Bay and 16-Bay R760 variants trade SFF density for additional PCIe slot headroom and Smart Flow cooling options. The 12-Bay 3.5" is the LFF capacity-tier variant (NL-SAS capacity drives, no GPU support, no NVMe on the LFF backplane). Wholesale Servers does not currently stock these R760 variants; contact us to discuss sourcing if your workload fits one of those configurations.

vs. R760xs 8-Bay (15th gen-equivalent value-tier sibling at 16th gen): The R760xs 8-Bay is the single-socket-optimized economics variant of the 16th gen 2U platform. For single-socket NVMe deployments at 8 bays, R760xs delivers significant cost savings. Dual-socket + 24 bays + Gen5 NVMe is the R760 flagship's territory; single-socket + 8 bays is R760xs's territory.

vs. R760xa (16th gen GPU-specialist sibling, not currently stocked): R760xa is the GPU-specialized variant with up to 4 double-wide GPUs (vs. 2 on R760 24-Bay). For maximum GPU density workloads where 4+ A100/H100/L40S per node is the design point, the R760xa is the platform. For GPU compute with co-located 24-NVMe storage at 2-GPU density, R760 24-Bay is the integrated answer.

vs. R660 (1U 16th gen sibling): The R660 10-Bay 2.5" is the 1U dense-storage 16th gen option. R660 has 10 SFF bays maximum; R760 24-Bay has more than double. For 1U rack density at moderate NVMe count, R660. For maximum SFF density and GPU support, R760 24-Bay.

Ready to Configure?

24-Bay R760 configurations benefit from upfront discussion on backplane variant (Universal vs. Passive NVMe vs. Switched NVMe), CPU generation (Sapphire Rapids 4th gen vs. Emerald Rapids 5th gen) and TDP, drive type and endurance selection, network sizing (100 GbE minimum for NVMe-heavy), vSAN/NVMe-oF/Ceph architecture, GPU requirements (if any), DLC infrastructure availability, and PSU sizing. Tell us your storage architecture, workload type, and quantity. We respond within 24 hours. Volume pricing applies at 5 units and above.

Every Wholesale Servers R760 ships after a 12+ hour burn-in test covering every PCIe slot, every memory channel, and every drive bay. Standard 180-day warranty included; 1-Year, 2-Year, and 3-Year Premium warranty options available. New-server pricing with Dell manufacturer warranty available on select configurations. Call 1-800-778-1545 or use the quote form on this page.