VMware vSphere host
The default platform in enterprise VMware clusters across EU and US data centers — typically deployed in pods of 16-32 hosts behind vSAN or external storage.
2U dual-socket — the world's best-selling enterprise rack server.
HPE ProLiant DL380 Gen12 is the flagship 2U dual-socket rack server. It supports Intel Xeon 6 (5th Generation Scalable) processors, up to 8 TB of DDR5 memory across 32 DIMM slots, 12 EDSFF E3.S NVMe SSDs and up to four double-slot GPUs. It is the workhorse platform for enterprise virtualization, OLTP databases, VDI and AI inference workloads in EU, UK and US data centers.
HPE ProLiant DL380 Gen12 is HPE's flagship 2U rack server and the twelfth generation of the most successful rack server in the 25-plus-year history of the ProLiant line. According to IDC and Gartner, the DL380 is consistently the single highest-unit x86 server family shipped each year, and Gen12 extends that lead with a major platform refresh: Intel Xeon 6 processors (E-core and P-core variants), 32 DDR5 DIMM slots, PCIe Gen5 expansion, EDSFF E3.S NVMe storage and a redesigned thermal envelope built for accelerator-heavy AI inference workloads. For our EU, UK and US enterprise customers, it is the default new-build platform for virtualization, database, virtual-desktop infrastructure and inference workloads alike.
Compute and memory subsystem. DL380 Gen12 supports two Intel Xeon 6 processors. On the high-core E-core (Sierra Forest) SKUs the platform supports up to 144 cores per socket — 288 cores per server — which dramatically improves VM density and Kubernetes pod density at constant rack-U footprint. On the P-core (Granite Rapids) SKUs the platform supports the highest single-thread frequencies and AVX-512 throughput in the Intel server portfolio, which is the right choice for OLTP databases, technical computing, electronic design automation (EDA) and financial-services low-latency workloads. With 32 DDR5 DIMM slots and 256 GB RDIMM modules, the platform reaches 8 TB of memory per server, more than doubling the upper bound of DL380 Gen11. Eight DDR5 memory channels per socket sustain over 600 GB/s of aggregate memory bandwidth — a critical specification for in-memory databases such as SAP HANA, Oracle In-Memory, Microsoft SQL Server in-memory tables and Redis Enterprise. The platform also supports Compute Express Link (CXL) memory expansion, enabling future memory-pooling architectures across multiple servers — an emerging design pattern for memory-bound workloads.
Storage flexibility. The Gen12 chassis exposes three primary backplane options. The first is 12 EDSFF E3.S NVMe SSDs — the new HPE-standard form factor that replaces both SATA and 2.5-inch U.2 in this server generation. EDSFF E3.S delivers approximately 30 percent higher IOPS density than U.2, hot-swap from the front bezel, substantially better thermal characteristics and a longer device lifetime; leading hyperscalers have already standardized on the form factor and the enterprise tier is following. The second is 24 SFF SAS/SATA/NVMe — preserved for customers with existing SAS storage investments or large multi-tier flash deployments. The third is 8 LFF — the right choice for capacity-bound NAS, archive, backup target and large-disk Ceph-style storage roles. All three backplanes are paired with the HPE Smart Array MR416i-o RAID controller (or the new MR408i-o for tri-mode SAS/SATA/NVMe), delivering hardware RAID 0/1/5/6/10/50/60 with optional flash-backed cache for write-intensive workloads.
Accelerator and GPU support. DL380 Gen12 supports up to four double-slot GPUs in a single 2U chassis — NVIDIA L40S for inference and graphics, NVIDIA H100 NVL for transformer inference (with 94 GB of HBM3 per card and NVLink across pairs), NVIDIA A100 for legacy training and inference, plus Intel Gaudi 3 and AMD Instinct accelerators on selected configurations. With PCIe Gen5 x16 lanes per GPU slot, the platform delivers full bandwidth to large-language-model inference workloads, retrieval-augmented-generation (RAG) pipelines, computer-vision serving, scientific computing and rendering. For dense training clusters where eight GPUs per node are required with NVLink and NVSwitch interconnect, HPE Compute XD685 (formerly Apollo 6500) in a 6U form factor is the recommended sibling platform. For very dense inference deployments, three to four NVIDIA L40S cards per DL380 Gen12 deliver excellent inference throughput at reasonable cost — a popular configuration for enterprise AI co-pilot and RAG deployments.
iLO 6 management. HPE Integrated Lights-Out 6 is the embedded baseboard management controller (BMC) on every DL380 Gen12. It provides KVM-over-IP for full graphical remote console, virtual media (mount ISO files from the management workstation), full sensor telemetry (temperature, power draw, fan speed, voltage, current), BIOS and firmware management, comprehensive event logging and a hardware security stack that includes Silicon Root of Trust (immutable hardware-based attestation) and Server Configuration Lock (prevents unauthorized BIOS or firmware modification). The optional iLO Advanced license adds federation (group management of dozens of iLOs from a single browser session), video record-and-playback for post-incident analysis, rich integrations with VMware vCenter, Microsoft System Center, Red Hat Satellite and ServiceNow, and the Insight Online remote-support telemetry feed. For US federal customers and EU regulated industries, iLO 6 includes FIPS 140-2 and FIPS 140-3 validated cryptographic modes and supports the CNSA Suite 2.0 algorithm set required by Department of Defense procurement. The TPM 2.0 chip provides hardware-backed key storage and measured-boot attestation suitable for Windows Server VBS (Virtualization-Based Security) and credential guard scenarios.
HPE OneView lifecycle management. Where iLO manages a single server, OneView orchestrates fleets. Template-based provisioning lets operators define a golden configuration once and apply it to a hundred DL380 Gen12s simultaneously — BIOS settings, firmware levels, RAID layouts, network MAC pool allocation, boot ordering, secure-boot policy, OS deployment and post-install configuration are all driven from a single declarative profile. OneView also provides full firmware compliance reporting, generating the documentation trail that simplifies audits against SOC 2 Common Criteria, ISO 27001 Annex A, NIST 800-53 controls SC-12 (cryptographic key establishment) and SI-7 (software, firmware and information integrity), PCI DSS requirement 6 (develop and maintain secure systems) and HIPAA technical safeguards. For VMware shops, the OneView for vCenter plug-in surfaces hardware health, firmware compliance and lifecycle operations directly inside the vSphere Client. For organizations standardized on Ansible or Terraform, the OneView REST API and the official Ansible collections enable full infrastructure-as-code workflows.
Workload fit and TCO. DL380 Gen12 is the default choice for general-purpose enterprise compute. For VMware vSphere and VMware Cloud Foundation clusters, the platform delivers an excellent balance of cores, memory capacity and storage IOPS — and with Broadcom's per-core VMware Cloud Foundation licensing model, the option to use Intel's E-core (Sierra Forest) SKUs to maximize VMs per host while controlling license cost has become particularly attractive. For Microsoft SQL Server and Oracle Database deployments, the 8 TB memory ceiling enables substantial buffer-pool sizing that often eliminates the need for a separate flash array; for Oracle Real Application Clusters (RAC), DL380 Gen12 paired with InfiniBand or 100 GbE RDMA interconnect delivers excellent throughput. For Citrix Virtual Apps and Desktops and VMware Horizon VDI environments, the GPU slots support NVIDIA vGPU profiles for GPU-accelerated desktops at densities of 100-200 power users per host. For Red Hat OpenShift, Kubernetes, EKS Anywhere and other container platforms, the high core count and PCIe Gen5 networking deliver excellent pod density and pod-to-pod latency.
Sustainability and power efficiency. Gen12 introduces significant power efficiency improvements relative to Gen11. The Intel Xeon 6 platform delivers approximately 20-30 percent higher performance per watt on typical virtualization and database workloads; the new high-efficiency Titanium-class power supplies (96 percent efficient at 50 percent load) reduce data-center energy consumption further; and the redesigned airflow path supports higher inlet temperatures (up to 35 degrees Celsius), allowing operators to raise the data-center setpoint and reduce cooling overhead. For organizations tracking ESG metrics under the EU Corporate Sustainability Reporting Directive (CSRD) or California's SB 253, these efficiency improvements translate directly to reported Scope 2 emissions reductions. HPE also publishes per-SKU product carbon footprint disclosures to support customer Scope 3 reporting.
Sora Yazılım engagement model. We deliver DL380 Gen12 as a turnkey engagement for our EU, UK and US enterprise customers. The engagement begins with workload sizing — we run HPE Server Sizer Tool against the projected workload mix and combine the output with our own VM-density modeling, IOPS calculation, network bandwidth analysis and accelerator-requirements assessment. From there we develop a co-engineered bill of materials with the right CPU SKU (E-core or P-core), memory configuration (capacity versus bandwidth optimization), storage layout (EDSFF NVMe versus SFF SAS), accelerator selection (if required) and network adapter choice (OCP 3.0 25/100 GbE or InfiniBand HDR/NDR). OEM-grade rack-and-stack installation includes physical mounting, structured power and network cabling, iLO and OneView baseline configuration, BIOS hardening per CIS Benchmark and NIST 800-53 SI-7 guidance, firmware compliance lock and a complete asset-tracking and handover dossier. For VMware vSphere, Proxmox VE, Microsoft Hyper-V and Red Hat Enterprise Linux or Ubuntu installations we deliver the platform fully provisioned and ready to host workloads on day one. Tech Care 24/7 with 4-hour onsite response is co-quoted on every proposal, and warranty renewals are tracked 90 days before expiry to avoid coverage gaps. For multi-server orders we coordinate staged delivery, regional logistics and on-site installation timelines with the customer's project schedule.
Compliance and regulatory posture. For regulated industries — financial services, healthcare, government, defense, energy and telecommunications — DL380 Gen12 supports the full cryptographic compliance baseline required by EU, UK and US enterprise customers. FIPS 140-2 and FIPS 140-3 validated cryptographic modes are available via iLO 6 and TPM 2.0; secure boot with measured-boot attestation provides hardware-anchored chain-of-trust; complete audit-log retention satisfies NIST 800-53 AU-2 and AU-12 controls; and the Silicon Root of Trust provides supply-chain integrity attestation that addresses NIST 800-161 (supply chain risk management) requirements. For our US federal customers we deliver against the DISA STIG (Security Technical Implementation Guide), FedRAMP-aligned baselines, NIST 800-53 moderate and high baselines, and CMMC (Cybersecurity Maturity Model Certification) level 2 and 3 requirements. For EU customers, the platform supports the NIS2 directive technical and organizational measures, the EU AI Act infrastructure attestation requirements where applicable and the GDPR Article 32 (security of processing) requirements. For UK customers, Cyber Essentials Plus alignment is delivered. Industry-specific compliance includes PCI DSS, HIPAA HITECH, SOX IT general controls, FDA 21 CFR Part 11 and the ISO 27001 / ISO 27017 / ISO 27018 control families. Sora Yazılım delivers the compliance baseline documentation, the configuration evidence packs and the audit-readiness reports required by both internal audit and external assessors.
The default platform in enterprise VMware clusters across EU and US data centers — typically deployed in pods of 16-32 hosts behind vSAN or external storage.
For high-volume OLTP databases, the combination of 8 TB of memory and EDSFF NVMe storage delivers exceptional buffer-pool and tempdb performance — common pairing for SAP HANA scale-up nodes.
Four GPU slots support vGPU-accelerated VDI for engineering, design and finance users — a typical deployment serves 100-200 power users per host.
Two to four NVIDIA L40S or H100 NVL GPUs serve mid-scale large-language-model inference and retrieval-augmented-generation pipelines for enterprise AI workloads.
Hardware FIPS 140-2/3 modes plus Silicon Root of Trust meet the cryptographic-attestation requirements of US and EU financial regulators.
Mid-to-large enterprise data centers — the default general-purpose compute platform for virtualization, database, VDI and AI inference workloads.
Opens the vendor's original technical documentation and product page in a new tab.
HPE Servers (ProLiant) — HPE ProLiant DL380 Gen12 →1U dual-socket — the dense data-center rack server.
DetailsAMD EPYC 9004/9005 2U rack server.
DetailsSingle-socket AMD EPYC 1U rack — built for VDI and branch.
DetailsDual-socket enterprise tower server.
DetailsSora Yazılım handles licensing, deployment, training and ongoing management — all from a single team.