HP ProLiant DL980 G7 Servers in Highly Available Linux Environments With Intel Xeon Processor 7500 Series and HP PREMA Architecture, the DL980 is an ideal platform for your enterprise applications Technical white paper Table of contents Introduction... 2 Linux and the DL980 as a Scale-up, Consolidation platform... 2 HP ProLiant DL980 G7 servers: Your IT Foundation... 3 The DL980 and HP Converged Infrastructure... 4 Red Hat Enterprise Linux 6 Enhanced Feature Set... 4 SUSE Linux Enterprise Server... 5 Linux High Availability Clustering vs. Proprietary RISC/UNIX Clustering Solutions... 5 High Availability Options with Linux... 6 DL980 Technical Design Overview... 8 HP ProLiant DL980 G7 Server PREMA Architecture... 8 Superior Manageability with HP Insight Control for Linux... 10 Red Hat Enterprise Virtualization on HP ProLiant DL980 G7 Servers... 11 Enterprise-level Capabilities of the Intel Xeon 7500 Processor... 11 Running Linux on Proprietary Hardware Platforms... 11 IBM z-series Mainframe: Integrated Facility for Linux (IFL)... 11 Running Linux on Proprietary POWER... 12 Capability Comparison Linux, Solaris and AIX... 13 Migrating Proprietary Applications to Linux-based HP ProLiant DL980 servers:... 14 Migrating from Solaris to Linux... 14 Migrating from AIX to Linux... 15 HP Migration Services... 16 TCO Comparison: RISC Servers and the HP DL980... 16 Comparison #1: IBM POWER 770 vs. HP ProLiant DL980 G7... 16 Comparison #2: Oracle SUN M9000 vs. HP ProLiant DL980 G7... 18 The DL980 as the Smart Enterprise Choice... 19 Links and Resources... 20
Introduction Many enterprises are migrating their core enterprise applications from expensive proprietary platforms or older x86 servers to Linux on standards-based servers. In the past, the lack of an x86 server with the performance, availability, and scalability found in proprietary platforms hindered adoption of these servers at the enterprise core. However, the new HP ProLiant DL980 G7 server with Intel Xeon Processor 7500 series overcomes these limitations. The DL980 is not just a larger x86 platform; it was designed with enterprise in mind. It features large internal storage capacity, network and I/O expandability, built-in availability and reliability features, and advanced power management capabilities. The DL980 is an excellent platform for scale-up consolidation and for virtualization of legacy UNIX and Linux workloads. The DL980 can help you to meet your enterprise-level Information Technology (IT) challenges, such as: Inability to scale resources effectively to handle high available demands and exponential data growth. Business demand for lower cost, standards-based Linux and x86-based servers to lower costs, where the only alternative is to run Linux as a secondary operating environment on a RISC or proprietary hardware platform. Insufficient reliability from existing servers to handle large single system databases and high-density virtualization. The need to lower support and operating costs, but the inability to do so due to legacy server sprawl. The HP DL980, featuring PREMA Architecture with Smart CPU Caching, along with a resilient system fabric, is designed to increase processing power, reduce throughput bottlenecks, and enhance reliability features not previously available in a lower TCO x86 environment. This white paper highlights the benefits of the Linux operating system (OS) running on the HP DL980, and the advantage of running enterprise applications in a lower cost clustered Linux environment. Linux and the DL980 as a Scale-up, Consolidation platform According to a Linux Foundation survey, Linux is the overwhelming operating system (OS) of choice for enterprises planning to move mission-critical applications from proprietary silos to a Cloud platform model based on standards-based servers. 1 HP is the top provider of Linux server systems, accounting for 40% of unit shipments worldwide. HP supports the broadest range of Linux distributions, including Oracle Enterprise Linux (OEL), Red Hat Enterprise Linux (RHEL), and SUSE Linux Enterprise Server (SLES). The Reliability, Availability, and Serviceability (RAS) value that the HP DL980 brings to your environment starts with the Intel Xeon processor 7500. Linux and scalability Since the introduction of the Linux 2.6 kernel into the major distributions with RHEL4 and SLES9, the scalability of Linux on multiprocessing systems has improved. The schedule, memory, and I/O subsystems now recognize the internal topology of the platform and are able to optimize for best performance. HP works with the commercial Linux distributions to ensure that workload scaling is effectively implemented on HP servers. 1 http://www.linuxfoundation.org/node/6792 2
Many larger, data intensive enterprise workloads are better suited to scale-up server technology. There are several advantages to running 64-core DL980 servers for scale-up enterprise workloads, such as: Large enterprise databases and data warehouse workloads function better when deployed in scaleup systems. Now, with the HP DL980, you can effectively scale workloads to 64 cores in a single server and beyond, if running clustered applications such as Oracle Real Application Clusters. Single system latency and bandwidth is always lower in a scale-up versus scale-out server implementation. While solutions such as InfiniBand can overcome many latency issues, it adds another layer of tuning and complexity. Trying to run large, data intensive workloads on RISC or proprietary servers with Linux can be risky. Proprietary systems are being marketed as consolidation platforms for older (non-critical) Linux workloads running on outdated x86 servers. Proprietary systems are not designed with Linux as their ideal operating environment while the DL980 is. The ability to cache all or most of your active database buffer blocks into a single large memory cache, as in the DL980, increases performance and throughput, while reducing application latency. You can operate your large, scale-up workloads on Linux with the HP DL980 at much lower TCO. Deploying large Oracle databases on the DL980 gives you the following advantages: Provides predictable user response times. The design delivers excellent transactional workload (OLTP) and data warehouse (DW) performance. Ability to grow your database workloads over time, scaling to thousands of users. Data growth in today s IT environment is nearly exponential the DL980 can handle dynamic changes and meet future requirements. Sustainable high-performance I/O and system throughput. HP ProLiant DL980 G7 servers: Your IT Foundation The HP ProLiant DL980 G7 is the newest member of HP s industry-leading scale-up x86 ProLiant family. HP designed this server to take full advantage of Intel s latest and most capable Intel Xeon processor, the Intel Xeon processor 7500 series with Quick Path Interconnect (QPI). Building on the advanced performance and reliability capabilities of the Intel processor technology and alignment with industry standard operating systems such as Linux, the DL980 G7 delivers an enhanced level of scale-up x86 performance, availability, and reliability. The ProLiant DL980 G7 is the first server to use the HP PREMA Architecture, incorporating a node controller design with smart CPU caching and resilient system fabric. HP PREMA Architecture represents HP s technology direction for scale-up x86 servers. The PREMA Architecture is the design foundation for x86 servers that need to deliver more scalability, resiliency, and efficiency to meet the requirements of the most demanding, data-intensive workload environments as well as large scale consolidation and virtualization. Balancing for Resiliency and Efficiency: Maximizing uptime is a key requirement for supporting virtualized, business critical workloads. The DL980 server with HP PREMA architecture delivers hardware-based resiliency with breakthrough cost-efficiency. Flexible, Scale-up Consolidation: Workloads from multiple legacy 8 to 32 core RISC servers can be virtualized and consolidated onto a single DL980. The scale-up architectures of the HP DL980 allows enterprise IT to offer improved resource availability combined with advanced virtualization and workload management. 3
High Performance Computing: The HP DL980 is an excellent platform for research and development applications that require large, globally addressable memory, as well as for applications with regular memory access that can use large caches and/or large numbers of registers. The DL980 and HP Converged Infrastructure HP Converged Infrastructure (CI) enables IT management to avoid costly data center expansion by delivering a services-oriented, dynamic workload infrastructure. The DL980 has the Intel Xeon Processor 7500 series, purpose-built to support virtualization and management capabilities. Combined with demonstrated excellent performance in benchmark tests, this makes an ideal platform for workload management when operating with the leading hypervisors such as VMware vsphere, Xen, and KVM with RHEL6 and SLES, and the Citrix Xen server. 2 HP CI provides operational efficiencies through common integrated management. The DL980 shares HP management tools such as HP Insight Control, HP Thermal Logic, and ilo 3 Remote Management with HP BladeSystem and Superdome 2 server families. Red Hat Enterprise Linux 6 Enhanced Feature Set Red Hat Enterprise Linux 6 (RHEL6), the latest release of Red Hat s trusted datacenter platform, delivers advances in high availability, application performance, virtualization, and scalability. With RHEL6, you can deploy physical, virtual, and cloud computing within your datacenter, reducing complexity, increasing efficiency, and minimizing administration overhead, while leveraging technical skills and operational know-how. Table 1: Features of Red Hat Enterprise Linux 6 Feature Function Benefits High Availability Virtualization Performance Leverages new hardware capabilities to offer a number of RAS improvements: Hot-add of devices and memory Enhanced error checking for PCIe devices via AER (Advanced Error Reporting) Support for MCA (Machine Check Architecture) capabilities allows the system to recover from hardware errors that would previously have caused complete system failure. ABRT (Automated Bug Reporting Tool) provides a more consistent way to identify and report system exception conditions like kernel failures and user space application crashes. Features fully integrated virtualization based on the KVM hypervisor. KVM benefits from a number of core kernel features such as the tickless kernel, cgroups (Control Groups) and the CFQ (Completely Fair Queuing) schedule. Red Hat's leadership in KVM development has fostered the development of sophisticated paravirtualization capabilities. The wide range of performance enhancements in RHEL6 reaches all components of the platform. For example LUN flush daemons that enable dramatic performance improvements for applications New features improve system availability by coupling advanced error recovery with enhanced logging/reporting of errors thereby increasing MTBF (Mean time between failures) and reducing MTTR (Mean time to repair). Supports MCA capabilities and comprehensive data integrity checking, previously only available only on proprietary systems. KVM Hypervisor enhancements make it practical to deploy large-scale enterprise applications in virtualized environments. SR- IOV-enabled performance enhancements allow I/O-bound applications to be virtualized, so IT departments can adopt a run anywhere capability. RHEL6 s new benchmark showcase leadership over competing UNIX, Microsoft, and VMware products across many applications. 2 http://h18004.www1.hp.com/products/servers/benchmarks/products.html#8p 4
Scalability Resource Management Provisioning and workload management deployed on large I/O subsystems. RHEL6 scales to the largest systems on the market today. For x86-64 systems, limits are now up to 4,096 CPUs, 33,000 IRQs. 64TB of memory, 4 million processes, and 32,000 threads per process. The new cgroups feature of RHEL6 offers a powerful way to allocate processor, memory, and I/O resources among applications and virtual guests. Many new features in the I/O subsystem cover interconnects (FCoE, iscsi, etc.) and hardware/software optimizations, support for thin provisioning, and simplified administration. The scalability features are not restricted to physical systems: RHEL6 also provides improved scalability of virtual guests. Physical, virtual, and cloud deployments can scale to meet business requirements. You will be able to deploy flexible virtual and cloud environments more easily, mixing critical enterprise applications with lowpriority background applications, while ensuring that the resources needed by both are properly allocated. Red Hat Network and Red Hat Network Satellite continue to provide management, provisioning, and monitoring for large deployments. SUSE Linux Enterprise Server SUSE Linux Enterprise Server is a highly reliable, interoperable and manageable server operating system built to power mission critical workloads in physical and virtual environments. SUSE Linux Enterprise Server is a modular, general-purpose operating system, runs on five processor architectures and is suitable for a wide range of workloads. And it is optimized to run on leading hypervisors and supports an unlimited number of virtual machine guests per physical system with a single subscription, making it the perfect guest for virtual computing. For more information http://www.novell.com/products/server/ Linux High Availability Clustering vs. Proprietary RISC/UNIX Clustering Solutions Excellent high availability has many times been the key selling feature of proprietary RISC/UNIX servers. However, according to recent IDC studies, this trend is changing rapidly. 3 In 2009, x86 servers captured more than 55 percent of all server revenue and accounted for more than 96 percent of server units shipped worldwide. Your enterprise should be considering standardsbased x86 servers such as the HP DL980 with Linux clustering to handle your most critical workloads. The DL980s also supports bare-metal, enterprise-grade virtualization platforms like VMware vsphere, unmodified, right out of the box. Whether you are using High Availability Cluster Multi-Processing (HACMP) for IBM servers or a Solaris cluster for your Sun SPARC servers, these solutions are proprietary and costly to purchase and support. Linux clustering for high availability is rapidly establishing itself in the corporate enterprise. Almost all core business applications ERP, CRM, sales force automation, inventory, and supply chain management can benefit from high availability Linux clustering. The adoption of Linux as the platform of choice for high-availability clusters is driven by a number of factors. First, many organizations find that when clustering with Linux, you can match or exceed the performance and availability of equivalent proprietary solutions, while dramatically reducing the total cost of ownership (TCO). Second, the open source character of Linux gives organizations the 3 http://www.serverwatch.com/hreviews/article.php/3868681/risc-to-x86-server-migration-buyers-guide.htm 5
flexibility to tailor a solution to fit their specific needs. Finally, by using open source software, organizations avoid the costly licensing and support fees associated single vendor solutions. High Availability Options with Linux The HP DL980 server runs well with several Linux-based high availability clustering solutions. Some of the options are described below. Red Hat Cluster Suite (RHCS) Low-cost high availability for enterprise applications For applications that require maximum uptime, a Red Hat Enterprise Linux cluster with Red Hat Cluster Suite (RHCS) is the answer. RHCS provides two distinct types of clustering: Application/Service Failover Create n+1 node server clusters for failover of key applications and services. IP Load Balancing Load balance incoming IP network requests across a farm of servers. RHCS is an integrated set of software components that can be deployed in a variety of configurations to suit your needs for performance, high-availability, load balancing, scalability, file sharing, and cost effectiveness. RHCS provides application and guest operating system failover capabilities. Applications can be moved between guest operating systems on the same or different physical systems, and entire guest operating system environments can be moved between physical systems. Applications can be monitored for correct behavior and the results used to trigger automated recovery procedures. For high-volume open source applications, such as NFS, Samba, and Apache, Red Hat Cluster Suite provides a complete ready-to-use failover solution. For most other applications, customers can create custom failover scripts using templates. RHCS includes the following key components: Cluster infrastructure Provides fundamental functions for nodes to work together as a cluster. High-availability Service Management Provides failover of services from one cluster node to another in case a node becomes inoperative. Cluster administration tools Configuration and management tools for setting up, configuring, and managing a Red Hat cluster. Red Hat Global File System (GFS) Provides a cluster file system for use with RHCS. GFS allows multiple nodes to share storage at a block level, as if the storage were connected locally to each cluster node. Cluster Logical Volume Manager (CLVM) Provides volume management of cluster storage. Novell SUSE Linux Enterprise High Availability Extension SUSE Linux Enterprise High Availability Extension is an integrated suite of open source clustering technologies that enable highly available physical and virtual Linux clusters. Used with SUSE Linux Enterprise Server, it enables business continuity capabilities that protect data and reduce unplanned downtime. It provides all of the essential monitoring, messaging, data replication, and cluster resource management functionality normally found in proprietary third-party solutions. The High Availability Extension is a flexible, policy driven clustering solution that uses OpenAIS, the leading standards-based communication protocol. It also supports Pacemaker, a new, scalable cluster resource manager with a flexible policy engine that supports n-node clusters. Using Pacemaker, you can continuously monitor the health of your resources, manage dependencies, and automatically stop and start services based on highly configurable rules and policies. The product also includes Oracle Cluster File System 2 (OCFS2), which is now a POSIX-compliant 6
generic cluster file system, and clvm2, a clustered logical volume manager. SUSE Linux Enterprise High Availability Extension enables continuous data replication with Distributed Remote Block Device (DRBD) version 8. Finally, the product supports multiple clustering scenarios, including hybrid physical and virtual clusters. Combining SUSE Linux Enterprise Server, integrated Xen hypervisor, and SUSE Linux Enterprise High Availability Extension together can provide support for complex virtualized environments. This gives customers unprecedented flexibility to improve service availability and resource utilization. 4 NOTE: For a limited time, HP is bundling SUSE Linux Enterprise Server High Availability (HA) Extension with SUSE Linux Enterprise Server subscription sold with HP support. This unbeatable value allows you to implement SUSE Linux Enterprise Server HA Extension in your SUSE Linux Enterprise Server environment free of charge for the duration of your subscription. 5 Table 2: Availability functions of popular Linux distributions Red Hat Enterprise Linux 6 SUSE Linux Enterprise Server 11 Online Reconfiguration OS Online Reconfiguration Hardware Can dynamically add and remove CPUs; dynamically add memory On some hardware, can dynamically add and remove CPUs; dynamically add memory Can dynamically add and remove CPUs; dynamically add memory On some hardware, can dynamically add and remove CPUs; dynamically add memory Reduced OS Update Downtime Yes Yes Error Handling Architecture Yes. Advanced Error Reporting (AER) and Machine Check Architecture (MCA) Yes HA Cluster and DR Options Red Hat Cluster Suite SUSE Linux Enterprise High Availability Extension Veritas Cluster Server Symantec s Veritas Cluster Server ensures availability of business critical applications and databases during planned and unplanned downtime by monitoring the status of the application and failing it over in the event of an outage or disaster. 6 Customers that have deployed it previously on Solaris can move to Linux with only minor modifications. Some key capabilities: Automates and accelerates failover of applications and databases within a data center or to multiple remote data centers. Supports heterogeneous physical and virtual operating system platforms with out-of-the-box solutions for all major database, application, and storage vendors. Provides comprehensive disaster recovery testing while applications remain on-line. VMware vsphere VMware HA delivers the availability needed by many applications running in virtual machines, independent of the operating system and application running in it. VMware HA provides uniform, 4 Source: http://www.novell.com/products/highavailability/ 5 Source: http://h18000.www1.hp.com/products/servers/linux/suse/index.html 6 Source: http://www.symantec.com/business/cluster-server 7
cost-effective failover protection against hardware and operating system failures within your virtualized IT environment. 7 Monitors virtual machines to detect operating system and hardware failures. Restarts virtual machines on other physical servers in the resource pool without manual intervention when server failure is detected. Protects applications from operating system failures by automatically restarting virtual machines when an operating system failure is detected. DL980 Technical Design Overview The following sections provide specifics on the enterprise-level capabilities that the DL980 offers relative to typical x86-based servers. HP ProLiant DL980 G7 Server PREMA Architecture HP PREMA Architecture provides resiliency features enabled by built-in application, infrastructure, I/O, memory, and processor capabilities. Figure 1: HP PREMA Architecture HP PREMA Architecture Resilient HP XNC Fabric Resiliency System Resiliency OS enhanced availability Error Isolation Link Level Retry Corruption Prevention Hot Plug RAID Advanced memory protection CPU Resiliency Machine Check Architecture Self-healing Resiliency 200% boost in server availability 8 The ProLiant DL980 G7 takes advantage of the enhancements that Intel has provided in its processor technology and builds on these to provide the resiliency demanded by mission-critical enterprise environments. Through tight collaboration with OS vendors HP ensures that the OS has awareness of the resiliency features that are enabled in the system. 7 Source: http://www.vmware.com/products/high-availability/ 8 Based on System crash rates comparison between the DL980 G7 to the DL785 G5. System crash rate is determined by availability features such as hot-swap components, redundant paths, ECC, and tolerant links such as QPI. 8
Resilient system fabric HP PREMA Architecture extends the advanced reliability of the Intel Xeon processor 7500/6500 series in the ProLiant DL980 G7 with a resilient XNC fabric. This interconnect fabric provides higher interconnect bandwidth to improve performance and scaling, and availability features consistent with the QPI fabric. The XNC fabric enables: Redundant data paths The XNC fabric s provision of 50% more interconnect links (six here versus four in most competitive 8-socket systems with no node controller) improves system performance by providing more bisection bandwidth and dynamically balancing the traffic on the links. The fabric redundancy also helps reduce unscheduled downtime. A complete failure in any one of the XNC links can be addressed with a system re-boot. The system will re-initialize and route around the failed link allowing a scheduled service event to be delayed until a convenient time (versus requiring immediate service to get the server backup). Rapid recovery Improved error logging and diagnostics information means that administrators can easily take corrective actions. If a fatal error occurs, the DL980 G7 captures the error log on the re-boot to assist in diagnosis. With the system running, the administrator can then use the log information to diagnose the error and rapidly determine which repair assemblies are needed. The PREMA architecture also reduces communication errors on overloaded systems, providing 50% more interconnect capacity and dynamic QPI (QuickPath Interconnect) routing. CPU Resiliency MCA Machine Check Architecture (MCA) enables detection and capturing of errors and faults for use by all of the system components. Link level retry The inter-processor (QPI) link level retry retransmits data to make certain data transfer is completed without loss of integrity. Memory resiliency Memory mirroring backup of main memory can be maintained on second DIMM and failover in the case of component failure Rank sparing data from a failing rank pair is copied to another rank pair after exceeding a threshold of correctable errors. Demand scrubbing when the system detects a correctable error via the data and ECC bits, it writes back the proper data and ECC bits to memory. Link level retry The memory (SMI) link level retry retransmits data to make certain the transmission is completed without loss of data integrity. I/O resiliency Corruption prevention HW will shut down the PCIe (I/O subsystem) to contain errors and prevent contamination of the sub-system or LAN. Link level retry The processor to I/O subsystem link retransmits data to make certain the transmission is completed without loss of data integrity. Hot plug redundant fans and power supplies. Application resiliency Operating system enhanced availability tight collaboration with OS distribution providers ensures there is awareness of system resiliency features. This includes corrective and recovery actions (ignore, correct, or crash the system to prevent data loss) based on MCA logs triggered via HP firmware. 9
Balanced Scaling Designed to eliminate system bottlenecks With PREMA Architecture HP addresses customer requirements that have exceeded the performance and capacities offered by 4-socket x86 systems. The architecture supports an appropriately balanced system with more processors, more memory, and more I/O than previous generation x86 systems have provided. However, simply adding processors, memory, and I/O slots is not sufficient to achieve the needed scalability and resiliency. When a system scales to a larger number of interconnected processors, the communication and coordination between the processors grow at an exponential rate, creating a system bottleneck. To solve this issue in our 8-socket x86 server, HP looked to the design of our higher-end mission-critical servers. At the core of the HP PREMA Architecture is a node controller, derived from technology powering the HP Integrity Superdome 2. The node controller enables two key functionalities: smart CPU caching and the resilient node controller (XNC) fabric. These features serve to reduce communication and coordination overhead and provide availability features consistent with Intel s Xeon 7500/6500 series processor. HP Smart CPU Caching technology improves utilization and performance, allowing processors to remain focused on the task with less interruption. HP Smart CPU Caching technology enables faster access to local memory and reduces interprocessor communication traffic. Breakthrough efficiencies In addition to the balanced scaling and self-healing resiliency enabled by the HP PREMA Architecture, the HP ProLiant DL980 G7 incorporates key ProLiant innovations to deliver breakthrough efficiency. Extend data center environmental capacity with HP Thermal Logic that reduces power consumption and can reclaim power capacity, allowing you to triple your data center capacity. HP integrated Lights-Out 3 (part of Insight Control) provides remote console performance up to eight times faster than the previous generation ilo 2 processor - and equal to the performance of KVM and software-based remote management solutions. Increased levels of application availability with HP Mission Critical Services. Superior Manageability with HP Insight Control for Linux HP Insight Control for Linux is an essential server management platform that integrates the best of open source and HP technologies on HP SIM (System Insight Manager). Insight Control is used for discovery, imaging, provisioning, server deployment, remote operations, virtualization (XEN and VMware), and power management. Open source tools, such as Nagios and XEN, are fully integrated at system startup and autoconfigured for immediate user productivity. Lifecycle management capabilities are complemented by multi-system scaling, power management, and direct-to-the-hardware control. Key Benefits on Insight Control on the DL980: Integrated Toolset for Linux Management Productivity Customizable Open Source and Commercial Integration Linux-managed Reference Platforms for Business-important infrastructures Protected with Global Support from Hewlett-Packard 10
According to IDC study, interviews with customers who had deployed Insight Control showed that on average operational expenses were reduced by $48,000 for every 100 users in their organization over a 3 year period when using HP management tools. 9 Red Hat Enterprise Virtualization on HP ProLiant DL980 G7 Servers The combination of Red Hat Enterprise Virtualization (RHEV) on the HP DL980 enables you to achieve large-scale consolidation projects RHEV with the DL785 provide you a platform to easily migrate legacy RHEL3, RHEL4-based applications. Benefits include: Reduces server sprawl by supporting fewer physical servers running hundreds of virtual machines. Lowers costs by using less energy, floor space and data center resources. Migrates legacy RHEL3, RHEL4 and Solaris applications to RHEL6 on energy efficient servers. Reduces risk with a secure, reliable OS-integrated hypervisor. Enterprise-level Capabilities of the Intel Xeon 7500 Processor Whether deploying a private cloud or implementing data center consolidation, the Intel Xeon Processor 7500 series, with Next-Generation Intel Microarchitecture (Nehalem), allows you to run the most demanding enterprise Linux applications and maintain peak workload responsiveness. Redundancy and dynamic routing capabilities are business critical characteristics designed into the DL980 by HP. HP PREMA Architecture provides Resilient System Fabric that can reduce the impact of application errors and improve system throughput. Features such as Cyclic Redundancy Checksum (CRC), link level retry, link width reduction (LWR), and link retrain work together to keep the system operational in the presence of link errors. Running Linux on Proprietary Hardware Platforms As Linux has become more widely accepted in the enterprise, proprietary hardware vendors have been working to adapt their legacy architectures to handle Linux as a secondary operating system. Linux distributions tailored to run on proprietary hardware must be funded by the vendor selling the hardware - not a long-term self-sustaining model. Any customer running Linux-based applications on RISC or other proprietary architectures should take a hard look at this for the future. In addition, Linux running on proprietary hardware platforms requires specialized expertise. There are also fewer applications available on specialized Linux distributions than on the releases that run on Linux/x86. IBM z-series Mainframe: Integrated Facility for Linux (IFL) IBM offers an IFL as a hardware module add-on to the mainframe. IBM wants you to believe that this is delivering the cost advantages of Linux and open source and the mission criticality of mainframe hardware. IBM positions the IFL as: A cost effective path for consolidating low utilization x86 servers Effective virtualization and workload management for 80% or higher utilization levels A method to extend your mainframe ecosystem Realities of the mainframe IFL as a target Linux platform: IBM s TCO claims assume that the mainframe is already purchased and paid for and that the x86 servers are several years old and not running any virtualization software. 9 IDC White Paper sponsored by HP, "Gaining Business Value and ROI with HP Insight Control," Doc # 210479, May 2009. Actual value of $48,480 over 3 years 11
Specialized expertise for managing applications on zlinux and the mainframe are still required The complexity and challenge of running multiple Linux applications in a virtualized zlinux environment should be investigated prior to implementation. IT staff to run Linux on an IFL need mainframe skills. Skills for Linux on x86 are much more common. Hardware maintenance costs are similar to the mainframe environment. They are high and are expected to rise. Unlike x86 servers, which are well suited to refresh cycles every 3 5 years, the mainframe is expensive and time consuming to upgrade, expensive to change, and requires significant specialized skills. Major mainframe performance improvements come incrementally, especially when compared to the constant advances in the x86. IBM claims that nearly 2,800 of the 5,000 mainframe applications are zlinux based, but this compares with the more than 14,000 applications available for Linux on ProLiant. Some key tools, utilities, and helper applications are unavailable on zlinux, which should be investigated if you are considering consolidating applications on an IFL. The HP DL980 enjoys the same economy of scale enjoyed by the market volume of x86 Linux servers. There is a reason why 46% of current mainframe owners are moving to more affordable and scalable industry-standard x86 solutions. 10 Running Linux on Proprietary POWER While IBM supports Linux on POWER, many Linux-base applications have yet to be written with full binary compatibility on the IBM POWER architecture. In contrast, HP's standards-based ProLiant platforms support the majority of existing Linux applications. Installing Linux even on the latest POWER systems still has hardware compatibility issues. 11 As well, Linux applications that directly access hardware devices require unique kernel modules when running on IBM POWER systems. 12 In most case, running Linux on POWER has shortcomings. For example, POWER s online CPU add/replace is not compatible with Linux, as it still needs to reboot. For dynamic LPAR changes on the fly (for example, adding CPU or memory, or removing a PCI adapter), some additional packages are needed for RSCT (Reliable Scalable Cluster Technology) communication with the HMC (Hardware Management Console). While a move to Linux brings cost savings, a decision to deploy Linux on POWER locks applications into a proprietary hardware cost structure. It is a lower cost, lower risk decision to run Linux on the HP DL980. 10 https://proliantsolutioncentral.com/enterprise/detail/linus_on_proliant_mfa_solution_brief_on_dl580_and_dl980 11 http://www14.software.ibm.com/webapp/set2/sas/f/lopdiags/info/linuxalerts.html 12 http://www.ibm.com/developerworks/linux/lx86/ 12
Capability Comparison Linux, Solaris and AIX The following table compares different operating system ecosystems against key metrics such as Availability, Virtualization, Management, and Provisioning all significant factors for running missioncritical enterprise applications. Table 3: Comparing key features between the OS ecosystems Linux Solaris AIX Availability Hardware: Hardware: Hardware: Runs on x86 and x86-64 servers, blades, clusters, and grids to POWER, SPARC, and IBM System z mainframes. The Linux Server provides flexibility to take advantage of current and future hardware. Software: With the Linux Server, more than 4,000 software applications are available and certified, including popular open source applications as well as proprietary applications from leading software vendors such as SAP and Oracle Solaris 10 is supported on SPARC and x86 servers sold by Oracle Sun and other vendors. Software: Solaris 10 on x86 has limited application availability to date. Many software vendors providing SPARC-based versions of their applications for Solaris have not ported those applications to Solaris on x86. HA Products: Solaris Cluster AIX runs only on any other platform that is non-ibm. Software: Limited Application level availability HA Products: PowerHA High Availability Cluster Multi- Processing (HACMP) EchoCluster HA Products: Red Hat Cluster Server SUSE Linux Enterprise High Availability Extension Veritas Cluster Server LifeKeeper Linux-HA Virtualization Mostly affordable standards-based tools like VMware ESX / Server, XEN, OpenVZ VirtualBox. VMware ESX Paravirtualized (modified) guest operating systems Fully virtualized (unmodified) guest operating systems with the necessary PV drivers on VT enabled hardware (Intel VT or AMD-V chipsets) Solaris Containers Solaris xvm Sun xvm VirtualBox Solaris 10 comes with built-in virtualization. In contrast, Xen provides isolation by running separate instances of one or more instances of various operating systems. E.g. SUSE Linux Enterprise Server, Microsoft Windows and NetWare. Sun xvm VirtualBox is x86 virtualization software installed on an existing host operating system such as Solaris, Windows, Linux, or Mac OS X. IBM MicroPartitions PowerVM LPAR IBM Power servers. AIX in combination with PowerVM 13
Management Converged Infrastructure HP Insight Dynamics-Virtual Server Environment Using SIM with Insight Rapid Deployment Pack. The Rapid Deployment Pack automates the process of deploying and provisioning server software, combining an off-the shelf version of Altiris Deployment Solution for Linux and the ProLiant Integration Module. Webmin - A web-based system configuration tool for UNIX-like systems. It is possible to configure operating system internals, such as users, disk quotas, services, or configuration files, as well as modify and control open source apps. ZENworks Suite of software products that supports multiple server platforms and multiple directory services. Sun Management center: Provides in-depth monitoring and diagnostics for systems, OS and services Provides Solaris Containers support: management, monitoring, reporting, resource utilization, and scheduling Manages advanced Solaris 10 features, including Solaris Container and DTrace IBM enterprise management: Manage workload partitions across multiple systems. Monitor and manage energy use. Configure and deploy virtual AIX images. Manage pools of virtual resources as a single system Provisioning LinMin's ASR Bare Metal Provisioning Bare Metal Imaging Provisioning Rollback Remote System Rescue JumpStartTM Enterprise Toolkit (JET): Install any of multiple versions of Solaris onto either SPARC or x86/x64 based clients. Deploy Flash archives. Utilize multiple boot methods: bootp, dhcp (including grub for x86/x64), and wanboot. Work across multiple subnets. Tivoli Provisioning Manager: Can create thousands of virtual machines pseries provisioning tool Migrating Proprietary Applications to Linux-based HP ProLiant DL980 servers: Managing and supporting legacy, proprietary servers like Sun SPARC or IBM POWER servers is expensive. Legacy RISC server environments typically have many under-utilized, non-virtualized servers sized for peak workloads. This increases the cost of application deployment and system management. The DL980 with clustered Linux can address this enterprise need with consolidation, high availability, virtualization, and simple workload management. Migrating from Solaris to Linux It is now time to migrate business critical Solaris workloads to Linux and the HP DL980. Migrating Web server farms, Web applications/portals, ERP systems and homegrown applications to Linux on HP ProLiant has been occurring for years. However, now it makes sense to move workloads to the HP DL980 requiring enterprise-level availability and functionality. There are advantages to moving your legacy enterprise Solaris workloads to Linux. Since both Linux and Solaris platforms use POSIX compliant APIs, porting time is reduced. In addition, the open-source software used with the Solaris platform is generally available on Linux, which speeds the migration process. HP also offers a Solaris-to-Linux porting kit, which can reduce porting time by 10x and the number of errors by 90%. 14
Case Study Migrating from Oracle Sun SPARC-based servers to Red Hat Enterprise Linux running on HP ProLiant Customer Situation: National City Corporation is part of PNC Financial Services Group, Inc., one of the nation's largest financial services enterprises and the fifth largest U.S. bank by deposits. Table 4: Migrating from Oracle Sun Solaris to Red Hat Enterprise Linux Business Challenge Solution Benefits Re-engineering the datacenter infrastructure from costly proprietary RISC machines running Sun Solaris and other UNIX distributions to x8-based blade machines running Red Hat Enterprise Linux. Reason for Migration: Cut costs and scale capacity to accommodate rapid growth of the business. The new datacenter runs Red Hat Enterprise Linux on a combination of physical and virtual HP ProLiant servers. The HP ProLiant servers support the bank s most business-critical applications, from Oracle financial software, to JBoss Enterprise Middleware based applications, online transaction processing (OLTP) systems, and customer- facing loan application systems. The price-performance of the Red Hat Enterprise Linux platform has been exceptional. IT operating costs were reduced to two cents per transaction, which has the potential to save millions over the life of the system. It also provided the ability to scale IT for business growth, and the- datacenter was re-engineered without unscheduled business interruptions. Source : http://rhcustomers.files.wordpress.com/2009/08/red-hat_national-city_case-study_1156464.pdf Migrating from AIX to Linux Migrating AIX applications to Linux on HP ProLiant servers is a straightforward process. As IBM offers Linux on POWER, there are tools that will allow you to evaluate what is needed to move your AIX workloads to Linux. As mentioned before, the advantage here is that moving AIX-based workloads to a lower cost operating environment and hardware platform will lower TCO. Case Study Migrating IBM AIX platform application to RHEL on HP ProLiant Servers Customer Situation: Wells Fargo is the fourth largest bank in the US by assets and the third largest bank by market capitalization. Table 5: Migrating IBM AIX platform application to RHEL on HP ProLiant Servers Migration Need Solution Results The Customer wanted to migrate their Teller application running on IBM AIX platform to RHEL/x86. The teller application is a business critical application for the bank and runs in over 100 branches of the bank HP migrated the teller application consisting of over 4 million lines of code, along with all the 3rd party and open source applications, from the AIX platform to RHEL running on HP ProLiant platform. Target OS was Red Hat Enterprise Linux AS release 4 (Nahant Update 4) The ported application went to production in about 8 months. The HP migration team worked closely with the customer and 3rd party ISVs throughout the engagement and completed the project in time. Source: http://enhanced2.sharepoint.hp.com/teams/presales/hq/warroom/hpmigrationcompetencycenter/shared%20documents/ migration-briefs.htm#_toc230593025 15
HP Migration Services HP offers a wide range of migration capabilities to help move you to Linux on the HP DL980: Cloud Consulting Services To help determine your cloud strategy and implementation roadmap. Converged Infrastructure Services Services to help you design, plan, implement, and support your next-generation IT structure. Data Center Transformation Services Consolidate facilities to cut operations costs, boost energy efficiency, enhance utilization. Consolidate servers, storage, and networks to increase capacity, simplify management, and improve market responsiveness. IT Consolidation Services Take complexity and cost out of your technology environment. Operating System Services Rely on HP experience and expertise spanning UNIX, Linux, and Microsoft Windows and other platforms. Virtualization Services Efficiently integrate virtualization technology into your environment. Manage virtual and physical resources with the same consistent processes. Specific HP capabilities include: Program management and training Turnkey migration solutions Application and database migration and porting The full support of HP Technology Services and HP Enterprise Services TCO Comparison: RISC Servers and the HP DL980 Comparison #1: IBM POWER 770 vs. HP ProLiant DL980 G7 Table 6: Compared Servers IBM POWER 770 vs. HP ProLiant DL980 G7 Server OS DBMS # of Servers # of sockets/cores Total cores IBM Power 770 POWER7 + (3.5 GHz) AIX Oracle 11g 1 8/48 48 DL980 G7 Intel Xeon X7560 (2.26GHz) Red Hat Oracle 11g 1 8/64 64 16
Table 7: 3-Year TCO comparison IBM POWER 770 vs. HP ProLiant DL980 G7 Solution A Solution B Difference TCO Comparison Cumulative 3-Year IBM P7 1 x 770/48c AIX HP ProLiant 1 x DL980 G7/64c (A B) Amount RHEL AP (A B) Percentage Server Hardware $412,531 $112,092 $300,439 72.8% Server Software (OS & DB) $1,812,048 $1,146,207 $665,841 36.7% Hardware and Software Support & Maintenance $1,286,877 $775,348 $531,529 41.3% IT Overhead Costs System Administration $43,620 $36,371 $12,249 28.1% Facilities (Power, Cooling & Floor space) Implementation costs (Planning & Deployment) $38,235 $24,084 $14,151 37.0% $0 $0 $0 0.0% Total IT Costs $3,593,311 $2,069,102 $1,524,209 42.4% Data Source: Ideas International Ltd. & Alinean Inc. (June 2010) TCO/ROI Summary: Overall savings of 42% over 3 years with HP ProLiant DL980 G7 Hardware acquisition cost savings of 73% Hardware and software support and maintenance cost savings of 41% 17
Comparison #2: Oracle SUN M9000 vs. HP ProLiant DL980 G7 Table 8: Compared Servers Oracle SUN M9000 vs. HP ProLiant DL980 G7 Server OS DBMS # of Servers # of sockets/cores Total cores M9000 SPARC64 VI (2.4 GHz) Solaris Oracle 11g 2 16/64 128 DL980 G7 Intel Xeon Processor X7560 (2.26GHz) Red Hat Oracle 11g 1 8/64 64 Table 9: 3-Year TCO Comparison Oracle SUN M9000 vs. HP ProLiant DL980 G7 Solution A Solution B Difference TCO Comparison Cumulative 3- Year TCO Comparison Sun SPARC 2 x M9000/64c Solaris 10 HP ProLiant 1 x DL980 G7/64c (A B) Amount RHEL AP (A B) Percentage Server Hardware $2,796,122 $112,092 $2,684,030 96.0% Server Software (OS & DB) $3,570,000 $1,146,207 $2,423,793 67.9% Hardware and Software Support & Maintenance $3,755,085 $775,348 $2,999,737 79.9% IT Overhead Costs System Administration $80,000 $31,371 $49,029 61.0% Facilities (Power, Cooling & Floor space) Implementation costs (Planning & Deployment) $470,022 $24,084 $445,938 94.9% $0 $0 $0 0.0% Total IT Costs $10,671,629 $2,069,102 $8,602,527 80.6% Data Source: Ideas International Ltd. & Alinean Inc. (June 2010) TCO/ROI Summary: Overall savings of 81% over 3 years with HP ProLiant DL980 G7 Hardware acquisition cost savings of 96% Hardware and software support and maintenance cost savings of 80% 18
The DL980 as the Smart Enterprise Choice As you consider your next enterprise platform, consider all the options. The HP DL980 is the culmination of many years of effort between HP and Intel. The result is a new solution, with the ability to scale and cluster Linux operating environments and applications on a lower-cost 64 core server. Even if your RISC environment is only 3 to 4 years old, the DL980 with Linux is an ideal platform for you to consider for migrating legacy UNIX workloads. Characteristics that make the DL980 an ideal platform for you to consider include: Integrated storage, networking, and server capabilities with HP Converged Infrastructure capability, allowing you to create pools of standards-based, managed resources. Enterprise resourced platform with up to 64 cores and 2 TB of memory that supports higher consolidation densities and larger virtual workloads. Single pane-of-glass management for simplifying operations with many legacy workloads on a small number of servers A 200% boost (compared with DL785) in availability to increase application uptime. Processor architecture that supports applications requiring large globally addressable memory and up to 64 cores of processing capacity. 16 PCI-e (x16, x8 and x4) slots for high performance I/O for accelerators, visualization, and high speed communications. HP has specifically designed the DL980 to meet your enterprise computing needs. It can operate at the core of your organization, replacing expensive, proprietary platforms and putting you on a path to free up budget and resources for other projects and initiatives 19
Links and Resources HP ProLiant DL980 G7 server Datasheet http://h20195.www2.hp.com/v2/getpdf.aspx/4aa1-5671enw.pdf HP ProLiant DL980 Server with HP PREMA Architecture http://h20195.www2.hp.com/v2/getpdf.aspx/4aa3-0643enw.pdf Intel Xeon Processor 7500 http://www.intel.com/itcenter/products/xeon/7500/index.htm?cid=cim:ggl xeon_us_7500 ks160 CA s Red Hat Enterprise Linux 6 Feature Function Benefit Summary http://www.redhat.com/f/pdf/rhel-6-features.pdf The Real Story about the IBM Mainframe http://h20338.www2.hp.com/enterprise/us/en/messaging/realstory-ibm-mainframe.html Linux on HP ProLiant servers http://h18004.www1.hp.com/products/servers/linux/index.html HP Insight Control for Linux http://h18004.www1.hp.com/products/servers/management/insightcontrol_linux2/index.html Migrating from Sun SPARC to HP ProLiant Servers with Intel Xeon processors http://h18000.www1.hp.com/products/servers/software/solaris/4aa2-4842enus.pdf Share with colleagues Copyright 2010 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. Intel and Intel Xeon are trademarks of Intel Corporation in the U.S. and other countries. Linux is a U.S. registered trademark of Linus Torvalds. Oracle is a registered trademark of Oracle Corporation and/or its affiliates. Microsoft and Windows are registered trademarks of Microsoft Corporation in the United States and other countries. UNIX is a registered trademark of The Open Group. AMD is a registered trademark of Advanced Micro Devices, Inc. 4AA3-1768ENW, Created November 2010; Updated November 2010, Rev. 1