HPE Integrity Servers
HPE Integrity Servers are a family of mission-critical server systems developed by Hewlett Packard Enterprise (HPE) to deliver high-performance, reliable computing for enterprise workloads.[1] These servers emphasize security, scalability, and continuous availability, making them suitable for demanding applications in sectors such as finance, healthcare, and large-scale data processing.[2][3] Introduced as part of HPE's enterprise portfolio, the Integrity line initially featured Intel Itanium processors and has evolved to incorporate x86 architecture in later models, supporting operating systems including HP-UX 11i, Linux, and Windows Server.[4][5] Key features across the series include HPE Integrated Lights-Out (iLO) for remote management, error-correcting code (ECC) memory for data integrity, and built-in redundancy mechanisms to minimize downtime.[6][7] The product range spans entry-level options like the rx2600 and rx2800 series for branch offices and mid-sized deployments, to high-end configurations such as the Superdome 2, which scales up to 8 TB of memory with support for up to 32 sockets.[5][7] Advanced models like the MC990 X Server extend capabilities for in-memory computing, utilizing Intel Xeon E7 v4 processors to achieve up to 48 TB of memory and 32 sockets for Linux-based applications requiring extreme scale and reliability.[8][9] As of late 2024, HPE announced the end of standard support for HP-UX 11i v3 and certain Integrity i4/i6 models by December 31, 2025, reflecting a shift toward newer x86-based mission-critical solutions while maintaining legacy compatibility.[10]Overview
Introduction
HPE Integrity Servers are a line of mission-critical server systems developed and produced by Hewlett Packard Enterprise (HPE), originally introduced by Hewlett-Packard in 2003 as high-reliability platforms primarily based on Intel Itanium processors for demanding enterprise computing environments.[11] These servers emphasize reliability, availability, and serviceability (RAS) features to support continuous operations in environments where downtime is unacceptable.[7] The Integrity brand traces its roots to Tandem Computers' NonStop technology, acquired by Compaq in 1997 and subsequently integrated into Hewlett-Packard's portfolio following the 2002 merger, infusing the line with a legacy of fault-tolerant and scalable computing designed for transaction-intensive applications.[12] Initially focused exclusively on the Itanium architecture, the series expanded in 2015 with the introduction of the Superdome X models, which adopted Intel Xeon E7 processors to incorporate x86 compatibility while maintaining mission-critical capabilities.[13] Targeted at sectors such as database management, financial transaction processing, and telecommunications infrastructure, Integrity Servers handle high-volume workloads with configurations supporting up to 64 CPUs in early Itanium-based models and up to 32 processors with up to 768 cores in later x86 variants (such as the MC990 X with Intel Xeon E7 v4 processors).[14][9] They feature rack-mounted designs ranging from compact 1U to 2U form factors for entry-level systems to full cabinet-scale deployments for high-end configurations, incorporating hot-swappable components for enhanced high availability.[6] The platform supports operating systems including HP-UX and OpenVMS, enabling robust virtualization and legacy compatibility.[15] As of late 2024, HPE announced the end of standard support for HP-UX 11i v3 and certain Integrity i4/i6 models effective December 31, 2025, emphasizing the shift to x86-based solutions while supporting legacy systems.[10]Key Features
HPE Integrity Servers are distinguished by their robust Reliability, Availability, and Serviceability (RAS) features, designed to ensure continuous operation in mission-critical environments. Hardware partitioning through nPartitions provides electrically isolated hard partitions, allowing independent allocation of CPUs, memory, and I/O resources across multiple workloads without compromising isolation.[16] Dynamic processor resilience monitors correctable errors in processors, idling degraded cores to prevent failures and deconfiguring them upon reboot for seamless recovery.[16][17] Memory protection employs Double-Chip Sparing (DDDC), which tolerates up to two DRAM failures per rank, reducing DIMM replacements by 17 times and memory crashes by three times compared to single-chip sparing.[16] Predictive failure analysis is facilitated by the Superdome 2 Analysis Engine, a firmware-based tool that proactively monitors system health, performs self-repairs, and centralizes error reporting independent of the operating system.[16] Scalability in HPE Integrity Servers supports mid-range configurations with up to eight sockets, suitable for demanding enterprise applications, while high-end models utilize a cell-based design for enhanced modularity.[18] In cell-based systems like the Superdome 2, up to 16 cells enable configurations reaching 32 sockets and 256 CPU cores, providing massive parallel processing capabilities for large-scale computing.[19] I/O capabilities include support for PCI-X and PCIe slots, enabling high-bandwidth connectivity for expansion cards and peripherals.[20] Integrated RAID controllers, such as the Smart Array P400 and P800 series, offer RAID levels 0, 1, 1+0, 5, 50, ADG, and 60, with features like battery-backed write cache and online spares for data redundancy.[20] Entry-level models accommodate up to eight hot-plug SAS or SATA drives, supporting transfer rates up to 6 Gb/s for SAS and 3 Gb/s for SATA, ensuring flexible storage options.[21][20] Management tools feature Integrated Lights-Out (iLO), which enables remote monitoring, configuration, and firmware updates via a dedicated management processor.[22] Hot-plug components, including redundant power supplies and fans, allow for maintenance without system downtime, enhancing operational continuity. Security features encompass secure boot to verify firmware integrity during startup and Trusted Platform Module (TPM) support for hardware-based encryption and measured boot processes.[23][24] Later models incorporate a silicon root of trust, embedding cryptographic verification at the hardware level to prevent unauthorized code execution from boot.[25] Energy efficiency is achieved through power capping via iLO, which dynamically limits total system power consumption to optimize data center resources, alongside idle processor states in Itanium-based systems that reduce energy use during low-activity periods.[26][27]History
Origins and Launch
The origins of HPE Integrity Servers trace back to the fault-tolerant NonStop servers developed by Tandem Computers, a company founded in 1974 that specialized in high-availability systems for financial and transaction-processing applications. Tandem's NonStop architecture, introduced in the mid-1970s, emphasized redundancy and continuous operation to prevent downtime in mission-critical environments. In June 1997, Compaq Computer Corporation acquired Tandem for approximately $3 billion in stock, integrating the NonStop line into Compaq's server portfolio while preserving its focus on reliability for enterprise workloads.[28] The path to the Integrity brand accelerated with Hewlett-Packard's acquisition of Compaq, completed in May 2002 in a $25 billion stock transaction that created one of the world's largest technology companies. This merger brought together HP's PA-RISC-based HP 9000 Unix servers and Compaq's inherited NonStop systems, prompting HP to unify these disparate lines under a new Itanium-focused architecture to streamline offerings and leverage Intel's 64-bit processors for future-proof enterprise computing. The Integrity branding was specifically chosen to reflect this consolidation, drawing from Tandem's earlier use of "Integrity" in NonStop product names while signaling a broader, integrated server family.[29] HP officially launched the Integrity Servers line in June 2003, aligning with the release of Intel's Itanium 2 Madison processors and introducing models built on HP's zx1 chipset. The debut entry-level rx1600 and midrange rx2600 servers supported up to two Itanium 2 processors, including the 1 GHz Deerfield variant, in compact 1U and 2U rack form factors designed for scalable performance. Marketed as mission-critical enterprise platforms for demanding workloads, these initial systems highlighted seamless compatibility with HP-UX for Unix environments and OpenVMS for high-reliability applications, positioning Integrity as a versatile alternative to legacy HP and Compaq servers.[30][11][31] A significant 2003 milestone was the November release of the rx8620 midrange server, which extended the lineup to support up to eight Itanium 2 CPUs in a 4U chassis, enabling larger-scale deployments with enhanced memory and I/O capacity for enterprise data centers. This model further solidified Integrity's role in bridging high-availability NonStop heritage with modern Itanium scalability.[32]Evolution and Milestones
Following the initial launch of Integrity Servers, Hewlett-Packard (HP) advanced the platform with the introduction of the zx2 chipset in 2007, which underpinned entry-level models such as the 2U rx2660 and 4U rx3600. These systems supported up to two dual-core Intel Itanium 2 processors, including the Montvale series (Itanium 2 9100) clocked at up to 1.66 GHz with 18 MB L3 cache per processor, enhancing performance for mission-critical workloads while maintaining compatibility with HP-UX, Windows Server, and Linux.[33][34][35] Between 2007 and 2010, HP shifted focus to mid- and high-end scalability with the sx1000 and sx2000 chipsets, which powered Superdome systems capable of supporting up to 64 Itanium processors for extreme compute density in enterprise environments. The sx1000 chipset, introduced in 2006 for the original Superdome, facilitated non-uniform memory access (NUMA) configurations across multiple cells, while the sx2000 enhanced I/O bandwidth and reliability for larger deployments. A pivotal milestone came in April 2010 with the launch of Superdome 2, featuring a modular bladed architecture compatible with c-Class BladeSystem enclosures, up to 32 sockets, and a maximum of 8 TB DDR3 memory with double-chip sparing for fault tolerance.[36][37][19] In 2011, HP announced Project Odyssey, a strategic initiative to unify Itanium and x86 architectures under the Integrity branding, extending mission-critical features like HP-UX's fault-tolerant capabilities to x86 platforms via technologies such as the Crossbar Fabric for resilient data routing. This project aimed to bridge ecosystems, allowing seamless integration of Intel Xeon processors into Integrity systems while preserving reliability standards. Concurrently, OpenVMS support expanded to newer Integrity models, including early availability on the rx2800 i2 server in early 2011, broadening the operating system's footprint for high-availability applications.[38][39][40] Throughout the 2010s, Integrity Servers evolved through successive generations tied to Intel's Itanium roadmap: the i2 series with Itanium 9300 (Tukwila) processors in 2010, quad-core (4 cores per socket with SMT for up to 8 threads) and introducing PCIe Gen2 for improved I/O throughput; the i4 series with Itanium 9500 (Poulson) in 2012, 8 cores per socket with enhanced multi-threading (16 threads); and the i6 series with Itanium 9700 (Kittson) in 2017, offering up to 8 cores per socket alongside PCIe Gen3 support in select configurations for higher bandwidth in data-intensive tasks. These updates, seen in models like the rx2800 i2/i4/i6, emphasized energy efficiency and virtualization, with up to 384 GB DDR3 memory and six PCIe slots per system.[41][6][27][42][43] A landmark in this era occurred in 2015 with the release of Superdome X, the first x86-based Integrity server, incorporating up to 16 Intel Xeon E7 v3 or v4 processors in a bladed chassis for up to 24 TB DDR4 memory, realizing Project Odyssey's vision of hybrid scalability while delivering four-socket RAS (reliability, availability, serviceability) features to x86 workloads. In November 2015, Hewlett-Packard split into HP Inc. and Hewlett Packard Enterprise (HPE), with the Integrity Servers line continuing development under HPE.[44][45] The trajectory shifted in 2019 when Intel announced the discontinuation of Itanium development, accepting final orders for the 9700 series until January 30, 2020, with shipments ceasing by July 29, 2021, marking the end of new Itanium-based Integrity advancements.[46]Architecture and Technology
Processors and Chipsets
The HPE Integrity Servers were built around the Intel Itanium processor family, which utilized the Explicitly Parallel Instruction Computing (EPIC) architecture to enable high levels of instruction-level parallelism by allowing compilers to explicitly schedule multiple instructions for simultaneous execution, reducing reliance on complex hardware speculation.[47] This design aimed to deliver superior performance in enterprise workloads such as transaction processing and scientific computing, though it required specialized software optimization.[48] The Itanium lineage in Integrity Servers began with the McKinley generation in 2002, featuring single-core processors at up to 1.0 GHz, marking the debut of Itanium 2 with improved branch prediction and cache hierarchy over the initial Merced.[49] Subsequent generations included Madison in 2003, reaching 1.5 GHz while enhancing floating-point performance; Montecito in 2006, introducing dual-core configurations at up to 1.6 GHz with hyper-threading support; and Montvale in 2007, achieving 1.66 GHz with better power efficiency through fine-grained thread scheduling.[50] Later iterations comprised Tukwila in 2010, a quad-core design at up to 1.73 GHz incorporating a new memory controller and QuickPath Interconnect; Poulson in 2012, scaling to 2.53 GHz with eight cores per socket and improved integer execution units; and Kittson in 2017, the final generation with up to eight cores at 2.66 GHz on a 32 nm process, before the full discontinuation of Itanium development.[43][51] These evolutions progressively increased core counts and clock speeds to address scalability demands in mission-critical environments. Supporting these processors were custom chipsets developed by HP (later HPE). The zx1 chipset, introduced in 2003, targeted entry- and mid-range servers, supporting up to four CPUs and eight PCI-X slots for I/O expansion.[52] The sx1000 chipset, launched in 2004 for high-end configurations like the Superdome, enabled up to 64 CPUs across multiple cells and 192 PCI-X buses, facilitating massive scalability.[53] In 2006, the zx2 chipset succeeded zx1, adding support for dual-core Itanium processors and higher memory bandwidth to accommodate Montecito and later generations.[54] The sx2000 chipset, released in 2008, transitioned to PCI Express for improved I/O throughput and supported up to eight sockets per cell, enhancing performance in multi-socket systems like the Superdome 2.[55] As the Itanium ecosystem faced challenges, including limited software adoption and competition from x86 architectures, HPE transitioned Integrity Servers to Intel Xeon processors starting in 2015 with the Superdome X, which supported E7-8800 v3 (Haswell-EX) and v4 (Broadwell-EX) families offering up to 22 cores per socket at clock speeds reaching 2.4 GHz.[44] This shift maintained Integrity's mission-critical focus while leveraging broader x86 compatibility.[8] Despite advancements, Itanium-based Integrity Servers encountered limitations, notably high power consumption—up to 185 W per socket in later multi-core designs—and difficulties building a robust software ecosystem, which hindered widespread adoption beyond specialized HP-UX and OpenVMS environments.[41]System Design and Management
The HPE Integrity Servers employ a modular, cell-based architecture in their high-end configurations, such as the Superdome series, where the system is constructed from independent cell boards, each supporting 2 to 4 processor sockets, interconnected via high-speed fabrics like the sx3000 crossbar to enable scalable resource pooling for compute, memory, and I/O.[36] This design allows for flexible expansion, with cells grouped into larger complexes while maintaining isolation for reliability.[56] A core aspect of the system design is nPartition (nPar) technology, which provides hard partitioning by dividing the server into up to four electrically isolated domains, each capable of running an independent operating system instance without interference.[57] This hardware-level isolation ensures fault containment, allowing maintenance on one nPar while others remain operational. Within an nPar, vPartition (vPar) technology enables finer-grained virtual partitioning, allocating specific resources like CPUs, memory, and I/O to multiple virtual OS environments for optimized workload distribution.[58] System management is facilitated by the Integrated Lights-Out (iLO) processors, specifically iLO 2, 3, and 4 variants tailored for Integrity Servers, which deliver out-of-band access for remote monitoring, configuration, and control independent of the host OS.[59] These processors integrate with standards like WBEM for systems management and SNMP for network-based alerts, while supporting secure firmware updates to maintain system integrity.[60] Redundancy features enhance availability, including dual-port network interface cards (NICs) for failover connectivity and mirrored I/O buses that duplicate data paths to prevent single points of failure. In the Superdome series, automatic failover mechanisms, such as redundant Onboard Administrator (OA) modules, ensure seamless transition during component failures.[19] Power and cooling systems prioritize efficiency and uptime, with redundant power supply units (PSUs) configured in N+1 or 2N modes and supporting 48V DC distribution for stable internal delivery. The Superdome 2 offers liquid cooling options alongside air cooling to manage high-density thermal loads in demanding environments.[36]Operating Systems and Software Support
Supported Operating Systems
HPE Integrity Servers primarily supported HP-UX, a proprietary Unix operating system developed by Hewlett-Packard. HP-UX 11i v1, released in 2000, introduced features such as the Process Resource Manager (PRM) for workload partitioning and Online Journaling File System (Online JFS) for enhanced file system reliability on Integrity hardware. Subsequent versions included HP-UX 11i v2 in 2003, which expanded PRM capabilities for finer-grained resource control and improved Online JFS scalability for large-scale deployments, and HP-UX 11i v3 in 2007, adding advanced virtualization support and security enhancements while maintaining compatibility with earlier Integrity models. Standard support for HP-UX 11i v3 ends December 31, 2025.[61] OpenVMS, a multi-user, multitasking operating system originally from Digital Equipment Corporation, was ported to the Itanium architecture for Integrity Servers in 2003, with initial pre-production releases (v8.0 and v8.1) available that year. The first production version, OpenVMS 8.2, followed in 2006, enabling high-availability clustering of up to 96 nodes across Integrity systems for distributed transaction processing. Support continued through versions up to 8.4-2 in 2016, which added compatibility for later Itanium processors like the 9500 series and enhanced networking features for blade servers. HPE standard support for OpenVMS 8.4 ended December 31, 2020; VSI OpenVMS continues support on compatible hardware until Integrity server EOL.[62][63] HPE NonStop, a fault-tolerant operating system derived from Tandem Computers' heritage, was adapted for Integrity hardware to provide continuous availability for mission-critical applications. It supports SQL/MP for parallel database operations and Pathway for managing transaction-driven workloads, ensuring no single point of failure in clustered environments. NonStop support on Itanium-based Integrity ended with hardware transitions; current NonStop systems use x86 architecture (NonStop X) with ongoing support as of 2025.[64] Linux distributions offered broad compatibility on Integrity Servers, particularly through Itanium editions. Red Hat Enterprise Linux (RHEL) versions 3 through 5 provided full support for Itanium-based Integrity models until the end of full support in 2017, with extended lifecycle support until 2020, emphasizing stability for enterprise workloads on servers like the rx series. SUSE Linux Enterprise Server (SLES) extended support through versions 9 to 11, with SLES 11 offering optimized kernel support for Itanium until support ended in 2019.[65][66] Windows support was limited to Itanium architectures, with Microsoft Windows Server 2008 R2 certified for Integrity Servers until mainstream support ended in 2013 and extended support in 2018.[67] Following the shift to x86-based models in 2015, such as the Integrity Superdome X, full support expanded to include Windows Server 2016 and later, alongside RHEL 7 and above, SLES 12 and above, enabling seamless migration for mixed workloads. For x86 Integrity models, supported operating systems as of 2025 include recent RHEL, SLES, and Windows Server versions.[68]Virtualization and Compatibility Features
The HP Virtual Server Environment (VSE), now known as HPE Insight Dynamics - VSE, integrates hardware partitioning technologies such as nPartitions (nPars) and virtual partitions (vPars) with advanced workload management to optimize resource allocation across HPE Integrity servers. This environment enables dynamic provisioning of compute resources based on business policies, allowing seamless workload balancing and migration while supporting high availability through integration with HPE Serviceguard clustering. VSE extends its capabilities beyond Integrity systems by incorporating workload management across both Integrity and ProLiant servers, facilitating unified infrastructure utilization and reducing operational silos in heterogeneous data centers.[69][70][71] Integrity VM serves as a type-1 hypervisor embedded within HP-UX on Itanium-based HPE Integrity servers, enabling the creation of multiple isolated virtual machines (VMs) within a single physical or virtual partition. It virtualizes CPU, memory, and I/O resources, supporting up to four virtual CPUs per guest VM and allowing flexible sub-CPU entitlements for fine-grained resource control. This setup permits running multiple instances of HP-UX or other compatible operating systems concurrently on the same hardware, enhancing server consolidation and isolation without requiring hardware-level partitioning.[72] For backward compatibility, HPE Integrity servers leverage Aries, a transparent binary translator integrated into HP-UX, to execute legacy PA-RISC applications on Itanium processors without recompilation. Aries emulates both 32-bit and 64-bit PA-RISC 2.0 binaries, providing near-native performance for migrated workloads while preserving investments in existing HP-UX software developed for PA-RISC systems. Developers can also use compiler options, such as those specifying portable architectures, to build applications that run across both PA-RISC and Itanium environments, easing the transition for mission-critical deployments.[73][74] Container support on HPE Integrity servers is limited, particularly for native HP-UX environments, but post-2017 advancements introduced HPE 9000 Containers to facilitate migration of HP-UX workloads into containerized formats. These containers enable rehosting of PA-RISC and Itanium-based applications on x86 Linux systems using Docker-compatible structures, supporting orchestration with Kubernetes for modern cloud-native deployments without full OS emulation. This approach provides isolation and portability for legacy HP-UX applications, though direct Docker or Kubernetes runtime on HP-UX remains constrained to specialized resource partitions like Secure Resource Partitions (SRP).[75][76][77] While earlier Integrity models based on Itanium do not offer native x86 emulation, the Superdome X series introduces direct x86 compatibility by utilizing Intel Xeon processors in a modular blade architecture. This shift allows seamless execution of x86-optimized operating systems and applications, such as Linux and Windows Server, within the same high-availability framework as prior Integrity systems, eliminating the need for translation layers.[78] Key management tools for virtualization include GlancePlus, a performance monitoring package that tracks CPU, memory, and I/O metrics across virtualized environments on Integrity servers, including support for Integrity VM guests. It provides real-time dashboards and alerts to identify bottlenecks in multi-tenant setups, aiding administrators in optimizing resource distribution.[79][80]Current Status and Migration (as of November 2025)
With the end of standard support for Itanium-based Integrity servers and HP-UX 11i v3 on December 31, 2025, HPE recommends migration to x86-based mission-critical servers like Superdome X or ProLiant, supporting modern Linux and Windows versions. Legacy OS like OpenVMS and NonStop continue via VSI or HPE on compatible hardware, with containerization options for HP-UX workloads.[61][64]Server Models
Entry-Level Models
The entry-level models of HPE Integrity Servers were designed for smaller-scale deployments, such as departmental applications, development environments, and testing workloads, offering compact rack-mounted or pedestal form factors with 1 to 4 sockets. These systems emphasized cost-effectiveness, reliability, and compatibility with mission-critical software, targeting organizations needing scalable entry points without the complexity of higher-end configurations.[81] The rx1600 and rx1620 series, introduced in 2004, represented the initial 1U rack-mounted offerings for basic Itanium-based computing. These 1- or 2-socket servers supported Intel Itanium 2 processors ranging from 1.0 GHz to 1.6 GHz with 3 MB L3 cache, utilizing the zx1 chipset for foundational I/O connectivity. They featured up to 16 GB of PC2100 ECC registered DDR SDRAM across eight DIMM slots (though configurations often capped at 12 GB for optimal performance), two PCI-X slots for expansion, and support for up to two internal hot-plug SCSI or SATA drives. Ideal for software development and light testing tasks, these models provided a low-entry barrier for Itanium adoption in space-limited environments.[81][82][83] Building on this foundation, the rx2600, rx2620, and rx2660 series (2003–2007) offered 2U rack-mounted platforms for more balanced mid-sized applications, supporting 1 to 2 sockets with Itanium 2 processors up to 1.66 GHz Montvale variants featuring 9 MB or 12 MB L3 cache, paired with zx1 or zx2 chipsets. Memory capacity reached up to 48 GB of PC2100 or PC2-3200 ECC registered DDR SDRAM across 12 to 24 DIMM slots, enabling handling of database and application servers. Storage options included 4 to 6 hot-plug drives (SCSI, SAS, or SATA), with I/O expansion via 4 to 6 PCI-X slots providing up to 8.2 GB/s bandwidth. These servers suited departmental workloads like ERP testing and web services, offering improved density over the rx1600 series.[35][84][31][85] The rx2800 i2, i4, and i6 variants (2008–2014) advanced entry-level capabilities in a 2U rack form factor, exclusively using 2 sockets with later-generation Intel Itanium processors: dual-core or quad-core 9300/9500 series (1.6–2.4 GHz) for i2, up to 9700 series (2.6–3.1 GHz) for i4/i6, supported by sx1000 or sx2000 chipsets. These models scaled memory to 384 GB for i4 (24 DDR3 RDIMMs at 1.35V) and up to 768 GB for i6 using low-voltage DDR3, with 8 small form factor (SFF) SAS/SATA bays and PCIe Gen3 slots for modern I/O demands. Designed for secure edge computing and virtualization in smaller enterprises, they integrated features like hot-plug components and advanced RAS (reliability, availability, serviceability) for continuous operations.[69][6][86] As a transitional 4U rack-mounted model launched in 2006, the rx3600 bridged earlier zx2-based designs with emerging multi-core needs, accommodating 2 sockets (up to 4 cores) of Montecito Itanium 2 processors at 1.42–1.67 GHz with 12–18 MB L3 cache. It supported up to 96 GB of DDR2 SDRAM across multiple slots, 6 PCI-X or PCIe slots for flexible expansion, and up to 8 internal hot-plug SAS/SATA drives. This server targeted evolving development and consolidation tasks, providing a stepping stone for users upgrading from 2U models.[87][88][89] For space-constrained pedestal deployments, the rx5670 (2005, discontinued 2010) offered a 7U form factor with up to 4 sockets of Madison or Montecito Itanium 2 processors at 1.3–1.6 GHz and 6–9 MB L3 cache, using the zx1 chipset. Memory topped at 96 GB (some configurations up to 192 GB) via 48 DIMM slots with PC2100 DDR SDRAM, alongside support for multiple internal SCSI/SAS drives and standard PCI-X I/O. It served as an entry for high-availability testing in labs or offices where rack space was limited.[90][91][92] Finally, the rx6600 (2007) provided high-density entry in a 7U pedestal chassis, scaling to 4 sockets (up to 8 cores) of Montvale Itanium 2 processors at 1.4–1.6 GHz with 6–18 MB L3 cache and zx2 chipset support. It handled up to 384 GB of DDR2 SDRAM across 48 slots, 12 hot-plug PCI-X slots, and 16 SAS hot-plug bays for robust storage. Suited for dense virtualization and application consolidation in smaller data centers, it maximized compute per footprint.[93][34][94]| Model Series | Form Factor | Sockets/CPUs | Max RAM | Key Storage/I-O | Primary Use Case |
|---|---|---|---|---|---|
| rx1600/rx1620 | 1U Rack | 1-2 (Itanium 2, 1.0-1.6 GHz) | 16 GB DDR | 2 drives, 2 PCI-X | Testing/Dev |
| rx2600/rx2620/rx2660 | 2U Rack | 1-2 (Itanium 2, up to 1.66 GHz Montvale) | 48 GB DDR | 4-6 drives, 4-6 PCI-X | Mid-sized Apps |
| rx2800 i2/i4/i6 | 2U Rack | 2 (Itanium 9300-9700, 1.6-3.1 GHz) | 768 GB DDR3 | 8 SFF bays, PCIe Gen3 | Secure Edge |
| rx3600 | 4U Rack | 2 (Montecito, 1.42-1.67 GHz) | 96 GB DDR2 | 8 drives, 6 PCI-X/PCIe | Transitional |
| rx5670 | 7U Pedestal | Up to 4 (Madison/Montecito, 1.3-1.6 GHz) | 96-192 GB DDR | Multiple drives, PCI-X | Space-Constrained |
| rx6600 | 7U Pedestal | Up to 4 (Montvale, 1.4-1.6 GHz) | 384 GB DDR2 | 16 SAS bays, 12 PCI-X | High-Density Entry |
Mid-Range Models
The mid-range models of HPE Integrity Servers targeted departmental and enterprise applications requiring scalable performance for growing workloads, offering modularity through multi-cell designs that allowed expansion up to 8 or 16 sockets while maintaining a balance between cost and capacity. These servers utilized Non-Uniform Memory Access (NUMA) architectures to optimize balanced performance in multi-processor configurations, supporting key business applications such as database management and high-availability services.[95][96] The rx4600 series, introduced between 2003 and 2006, provided entry into mid-range computing with compact form factors suitable for space-constrained environments. The rx4610 model featured a 7U pedestal or rack-mountable chassis supporting up to 4 Itanium processors on the zx1 chipset, with a maximum of 64 GB RAM using PC2100 DDR SDRAM; it was designed for initial deployment in growing departmental setups.[97][95] The rx4640 extended this with a 4U rack-optimized design accommodating up to 4 dual-processor modules (8 cores total) based on Itanium 2 processors, up to 128 GB RAM, and 8 PCI-X slots for enhanced I/O connectivity, targeting workloads demanding higher throughput without excessive footprint.[98][4] The rx7600 series, available from 2004 to 2008, emphasized cell-based modularity for easier upgrades in enterprise settings. The rx7620 offered a 10U to 13U rack or pedestal configuration with 2 cells supporting up to 8 Itanium 2 processors on the sx1000 chipset and up to 256 GB RAM, enabling cell interconnects for seamless expansion in multi-user environments.[99][100] Complementing this, the rx7640 provided similar 8-socket scalability in a 10U form factor but upgraded to the sx2000 chipset with PCIe support, up to 256 GB RAM, and improved I/O bandwidth for demanding applications like online transaction processing.[96][101] The rx8600 series, spanning 2003 to 2008, represented the upper end of mid-range modularity with NUMA-optimized multi-cell designs for balanced enterprise performance. The rx8620 utilized a 17U chassis with up to 2 cells supporting 8 Itanium 2 processors on the sx1000 chipset and 128 GB RAM per cell, facilitating modular growth for core business operations.[102] The rx8640 advanced this to 4 cells in a 17U rack form factor, scaling to 16 processors on the sx2000 chipset with up to 512 GB RAM and extensive PCIe/PCI-X slots, ideal for high-density departmental consolidation without entering high-end clustering.[103]High-End Models
The high-end models of HPE Integrity Servers represent cabinet-scale systems engineered for extreme scalability and reliability in mission-critical data centers, supporting massive parallel processing and resource isolation for applications like financial services, telecommunications, and large-scale databases. The original Superdome series, introduced in the early 2000s for the Integrity line, featured configurations such as the SD-16, SD-32, and SD-64, scaling up to 64 CPUs across 4 to 8 cells using the sx1000 chipset. These systems supported Itanium 2 processors at speeds like 1.5 GHz with 6 MB cache, enabling configurations from 2 to 64 CPUs depending on the model. Maximum memory reached up to 1 TB of RAM, distributed across cells for balanced performance. A key capability was massive partitioning via nPartitions (nPars), allowing up to 8 independent partitions per system for mainframe-like isolation, fault tolerance, and resource dedication in high-availability environments.[104][105] The Superdome 2, released in 2010, introduced a bladed, modular design that scaled to 32 sockets using Intel Itanium processors from the Tukwila (9300 series) and Poulson (9500 and 9700 series) families, each with up to 8 cores. Built on the sx3000 chipset, it supported up to 8 TB of DDR3 memory across 512 slots and incorporated liquid cooling for sustained high-density operations. The high-end 32-socket variant delivered up to 256 cores and 64 built-in 10 GbE ports, with connectivity via 96 PCIe 2.0 slots. It enhanced partitioning with support for up to 16 nPars, enabling granular resource allocation while maintaining compatibility with HP-UX and other Integrity OSes.[106][19] Shifting to x86 in 2015, the Superdome X scaled to 16 sockets using Intel Xeon E7 v4 processors, such as the E7-8890 v4 with 24 cores at 2.2 GHz, for a maximum of 384 cores in fully populated configurations. It provided up to 24 TB of DDR4 memory across 384 DIMM slots with ECC and double-chip sparing, PCIe Gen3 expansion with 24 mezzanine slots, and integrated iLO 4 management for automated monitoring and recovery. Supporting up to 8 blades in an 18U enclosure, it offered hybrid compatibility between Itanium and x86 workloads through nPars hard partitioning and vPars virtualization, allowing seamless migration and mixed-OS environments. A standout feature was dynamic resource sharing via flexible partitioning (including FlexScalers-like capabilities), enabling real-time reallocation of CPU, memory, and I/O across up to 16 partitions for optimized utilization in consolidated data centers.[107][78] The MC990 X Server, introduced in 2017, extended high-end capabilities for in-memory databases and extreme-scale Linux applications, scaling to 32 sockets with Intel Xeon E7-8800 v4 family processors (up to 24 cores each at 2.2 GHz). It supported up to 48 TB of DDR4 memory across extensive DIMM slots, with 20 PCIe Gen3 slots for I/O expansion and integrated management via iLO 5. Designed for Oracle Linux and in-memory computing like Oracle 12c, it featured advanced RAS features including double-chip kill and patrol scrubbing for data integrity, enabling configurations in a rack-mountable 11U chassis for massive shared-memory environments.[9][8]Discontinuation and Legacy
End-of-Life Timeline
In January 2019, Intel announced the end of manufacturing for the Itanium 9700 series processors, with final orders accepted until January 30, 2020, and the last shipments scheduled for July 29, 2021.[108] This marked the definitive closure of the Itanium product line, which had powered most HPE Integrity Servers since their inception. HPE aligned its support lifecycle accordingly, retiring Itanium-based models progressively and shifting focus to x86 alternatives. HPE began announcing end-of-life (EOL) phases for specific Integrity models following Intel's decision. The rx2800 i6 server reached EOL on December 31, 2020, though end-of-service-life (EOSL) support extended to December 31, 2025.[109] Similarly, the BL860c i6 blade server followed an EOL in 2020 with EOSL until December 31, 2025.[110] For the Superdome 2, support concluded earlier for i4 configurations on March 31, 2023, while i6 variants maintained EOSL through December 31, 2025.[111] By the end of 2025, all Itanium-based Integrity models will be obsolete, with no new hardware or standard support available.[61] Operating system support tied to Integrity hardware also faced defined endpoints. Standard support for HP-UX 11i v3, the primary OS for many Integrity systems, ends on December 31, 2025, after which only mature support—limited to existing fixes and parts—will continue through at least December 31, 2028.[61] OpenVMS support on Integrity servers has been extended by VMS Software Inc. (VSI), with ongoing compatibility for i6 models under VSI OpenVMS V8.4-2 and later releases, though full ecosystem migration to x86 is encouraged beyond 2025.[112] HPE NonStop, another key OS for Integrity, remains actively supported with no announced EOL, as evidenced by the June 2025 introduction of the fifth-generation NonStop Compute NS5 X5 systems.[113] Key announcements framed these transitions. In December 2014 (with general availability in 2015), HPE introduced the Superdome X as an x86-based bridge from Itanium, supporting up to 16 Intel Xeon processors and designed for mission-critical workloads previously handled by Integrity Itanium systems.[13] This positioned Superdome X for longer support under standard HPE x86 cycles, avoiding the Itanium timeline. In October 2024, HPE issued a product change notification confirming that standard support for all i4 and i6 Integrity servers—and related HP-UX sales—would end on December 31, 2025, urging customers to plan migrations.[61]| Milestone | Date | Description |
|---|---|---|
| Intel Itanium EOL Announcement | January 2019 | End of manufacturing for 9700 series; final orders January 30, 2020.[108] |
| Last Itanium Shipments | July 29, 2021 | Final delivery of processors to HPE and partners.[108] |
| rx2800 i6 and BL860c i6 EOL | December 31, 2020 | Cessation of new sales for these models.[109][110] |
| Superdome 2 i4 EOSL | March 31, 2023 | End of service for i4 variant.[111] |
| Superdome X Introduction | 2015 | x86 transition platform announced.[13] |
| HP-UX 11i v3 and i4/i6 EOSL | December 31, 2025 | Standard support ends for OS and hardware.[61] |
| HP-UX Mature Support | Through December 31, 2028 | Limited ongoing access to fixes.[61] |
| NonStop Ongoing Support | No EOL Announced | Continued with new x86 generations in 2025.[113] |