IBM z/vm V4.4 Improves Virtualization Capabilities for Linux on zseries

Transcription

1 Software Announcement May 13, 2003 IBM z/vm V4.4 Improves Virtualization Capabilities for Linux on zseries Overview The infrastructure for e-business on demand environments can be deployed with visionary technology on z/vm, including advanced virtualization, support for Open Source software, and autonomic computing enhancements. With virtualization technology as its foundation, z/vm continues to provide new function and technology exploitation on the mainframe that enables you to virtualize processor, communication, memory, storage, I/O, and networking resources, with the potential to help reduce the need to plan for, purchase, and install hardware to support new workloads. z/vm supports Linux, one of the world s leading Open Source operating systems, on the mainframe. Within the VM environment, Linux images can share hardware resources and use internal high-speed communications. While benefiting from the reliability, availability, and serviceability of IBM zseries servers, z/vm V4 offers an ideal platform for consolidating select UNIX, Windows, and Linux workloads on a single physical zseries server, which allows you to run tens to hundreds of Linux images. z/vm V4 is priced on a per-engine basis and supports IBM Integrated Facility for Linux (IFL) processor features for Linux-based workloads, as well as standard engines for all other zseries and S/390 workloads. Enhancements that have been added to z/vm to enable and incorporate autonomic computing technology include: Self-configuring functions aid in the configuration and management of networks, along with the definition and dynamic reconfiguration of hardware I/O resources using a graphical interface. Additionally, APIs are provided to create and manage virtual images running under z/vm and the Virtual Machine Resource Manager (VMRM) allows dynamic changes to workload configurations. Self-optimizing capabilities have been added that incorporate I/O management facilities, enabling z/vm to prioritize guest and host I/O operations and improve z/vm scheduler performance. Self-management improvements using VMRM provide facilities that define and achieve performance goals of virtual machines. Self-protection enhancements to TCP/IP for z/vm provide better protection and management of the communication stack and cascaded FICON director support helps detect and report miscabling actions. z/vm integrates and continues to offer traditional capabilities to manage zseries operating systems, including Linux, on a single physical hardware server as guests of z/vm while benefiting from world-class VM virtualization technology. For the most current information on z/vm, refer to the z/vm Web site at: zvm/ IBM Global Financing provides financial packages to enable your company s smooth and affordable transition to the world of Linux on zseries and z/vm. Key Prerequisites Refer to the Hardware Requirements and Software Requirements sections for details. At a Glance z/vm V4.4 includes: Virtualization technology and Linux enablement: Potential to help reduce overhead and improve performance of virtual machines Higher efficiency when managing large numbers of virtual machines High-performance virtual FICON channel-to-channel adapters Network virtualization: Additional network-traffic configuration options using Virtual LANs (VLANs) External IP connectivity for Guest LANs through virtual switching Technology exploitation: IBM zseries 990 with: -- Improved logical-partitioning scalability due to logical channel subsystems -- Improved capacity planning and I/O performance measurement -- Twice the number of logical partitions (LPARs) Enhanced and simplified connectivity with cascaded FICON directors Systems management: Improved performance monitoring and reporting Better control, definition, and dynamic reconfiguration of hardware I/O Enhanced network security management Support for the new C/C++ for z/vm compiler For ordering, contact: Your IBM representative, an IBM Business Partner, or the Americas Call Centers at 800-IBM-CALL Reference: LE001 This announcement is provided for your information only. For additional information, contact your IBM representative, call 800-IBM-4YOU, or visit the IBM home page at: IBM United States IBM is a registered trademark of International Business Machines Corporation

2 May 13, 2003 Description Planned Availability Dates z990 compatibility support for: -- z/vm V3.1, V4.2, and V4.3 August 15, 2003 z/vm V4.4 including z990 exploitation support for: -- Multiple Logical Channel Subsystems (LCSS 0 and LCSS 1) -- Dynamic-I/O configuration only in LCSS 0 -- Extended-I/O measurement facilities -- Adapter interruptions for OSA-Express and FCP channels -- Performance assist for V=V guests (pageable guests) in Linux and VM environments October 31, 2003 z/vm V4.4 support for: -- Dynamic-I/O configuration in multiple LCSS environments LPARs -- Spanned channels -- IPL from FCP-attached disks when hardware function is available Virtualization Technology and Linux Enablement z/vm V4.4 extends its virtualization technology in support of Linux and other guests while providing some enhancements that enable z/vm to be self-optimized and self-managed. Enhanced Queued-Direct-I/O (QDIO) Performance: z/vm V4.4 and the new z990 server are designed to provide important performance improvements to the QDIO architecture initially introduced with the OSA-Express adapter. The use of the QDIO architecture was later extended to HiperSockets and the Fibre Channel Protocol (FCP) channels. The architecture itself was extended in HiperSockets to include a new type of high-performance I/O interruption known as an adapter interruption. The use of adapter interruptions has been extended to the OSA-Express and FCP channels on the z990 server. Thus, OSA-Express adapters for networking and FCP channels for Fibre Channel fabrics join HiperSockets in benefiting from the higher performance characteristics of these interruptions. In addition to the use of adapter interruptions by the OSA-Express and FCP channels, the z990 server is designed to include an important performance assist for the virtualization of adapter interruptions being given to guest operating systems. This hardware performance assist is available to V=V guests (pageable guests) that support QDIO on z/vm V4.4. This significant new IBM virtualization technology is designed to benefit all guest operating systems in environments where they can process adapter interruptions. This includes all users of HiperSockets, and guest operating systems that add adapter-interruption support for OSA-Express and FCP channels. With the enhancement of the TCP/IP stack in z/vm V4.4 to use adapter interruptions for OSA-Express channels, TCP/IP for VM can benefit from this performance assist for both HiperSockets and OSA-Express adapters. Linux Guest Capacity Improvements: Improvements to the z/vm Control Program (CP) scheduler increase the number of Linux and other guest virtual machines that can be managed concurrently. While this increase in capacity can be experienced on all zseries hardware configurations, it can be more dramatic on larger n-way servers. The scheduler lock is used to serialize scheduler activities, timer request block management, and processor local dispatch vector (PLDV) management. Because the contention on the lock is lowered by creating a new lock for timer request block management, CP overhead can be decreased. Systems Management APIs: The Systems Management APIs provide a basic set of functions that may be called by applications to allocate and manage resources for guests running in z/vm virtual machines (virtual images). Although these APIs are primarily intended for managing Linux virtual images, they can be used to manage any type of virtual image running in a z/vm virtual machine. The APIs are designed so that applications can be written by customers or Solution Providers to help administrators, especially those who lack in-depth VM knowledge, manage large numbers of virtual images running in a single z/vm system. Functions supported by the Systems Management APIs allow you to: Create and delete virtual images Add and remove resources in an image s static (directory) or dynamic (active) configuration Activate and deactivate virtual images (individual or lists) Manage connectivity between virtual images including exploitation of the new virtual switch support Query the status of long-running asynchronous operations initiated through the APIs Create, delete, replace, and query shared storage segments The Systems Management APIs require a directory manager. The IBM z/vm Directory Maintenance Facility (DirMaint ) feature has been enhanced with support for the APIs with PTFs for APARs VM63122 and VM Client applications invoke the Systems Management APIs through a set of IBM-supplied Remote Procedure Calls (RPCs) to a z/vm server virtual machine (either remotely or from within the z/vm system). The server then calls a set of Callable Service Routines, which perform the requested functions. The Callable Service Routines are replaceable to allow customized solutions for the Systems Management API functions. Exploitation of the Systems Management APIs: Solution Providers planning to exploit the new Systems Management APIs provided by z/vm V4.4 include but are not limited to: Levanta by Linuxcare is designed to configure, update, and replicate virtual Linux servers on IBM zseries under z/vm. Using Levanta, data centers can deploy Linux with best practices; Linux, PC, mainframe, networking, and other experts can more easily collaborate as a team; and system administrators can increase their effectiveness

3 For more information and availability, you can contact Linuxcare directly by sending an to calling , or visiting their Web site at: VMGUIOP by Sine Nomine Associates is designed to simplify the administrative tasks of managing VM operations through a graphical user interface that can help automate many of the most common VM administration operations in a menu-driven format. This graphic interface resembles the desktop environments many administrators commonly use on other systems, such as Linux and Windows. Sine Nomine Associates has announced that VMGUIOP is planned to be generally available in the third quarter of For more information and availability, you can contact Sine Nomine directly by sending an to info@sinenomine.net, calling , or visiting their Web site at: Virtual Machine Resource Manager (VMRM): VMRM has been enhanced to provide the infrastructure necessary to support more extensive workload and systems resource management features that may be delivered in future releases of z/vm. To accomplish this objective, the following have been added: Monitor data showing actual workload achievements An interface to dynamically change users in workloads, workload characteristics, and goals in the VMRM configuration file without manually stopping the server More flexibility when managing multiple users by supporting wild-card characters on user IDs in the configuration file Improvements in the reliability and performance of the VMRM service virtual machine s monitor data handling Serviceability enhancements, including improved messages, logfile entries, and new server options Secure Socket Layer (SSL) Upgrade for Linux: The upgraded SSL server shipped with TCP/IP for z/vm V4.4 provides appropriate Red Hat Package Manager (RPM) packages for the SuSE Linux SLES 7 and SLES 8 distributions. This upgrade helps avoid potential support issues with the backlevel Linux kernel and allows the SSL server to exploit the many performance improvements in the Linux 2.4 kernel. Guest IPL from Small Computer System Interface (SCSI) Disks: z/vm V4.4 allows initial program loading (IPLing) from FCP-attached SCSI disks for Linux and other guest operating systems that contain support for IPLing from FCP-attached SCSI disks, when z/vm is running on a z990, z900, z800, or equivalent server equipped with the IPL-from-FCP-attached-disk function. IBM intends to make this hardware function available in the future. Thus, when the hardware functionality is available, Linux guests can be started and run completely from FCP disks in your hardware configuration. z/vm continues to require ESCON - or FICON-attached disk or tape for its own IPL and other functions. Virtual FICON Channel-to-Channel Adapters (CTCAs): z/vm V4.4 provides virtual FICON CTCA devices for guest operating systems. Guests use virtual CTCAs to communicate among themselves within a single z/vm-system image, without the need for hardware CTCAs. Virtual CTCAs provide high-performance communication because z/vm simply moves data in system memory between the communicating virtual machines. Earlier releases of z/vm provided support for hardware FICON CTCAs, and this support carries forward into z/vm V4.4. Virtual FICON CTCA devices enhance previous virtual-ctca capabilities by adding the FICON protocol as an option for guest operating systems. Virtual CTCAs now support the 3088, ESCON, and FICON protocols. With the new FICON-protocol support, customers that use real FICON CTCAs in production can run their test configurations as guests on VM in the exact same way as will be done in production. Integrated 3270 Console: z/vm V4.4 adds support for real and virtual integrated 3270 console devices. Real support enables this device, which is provided through a Hardware Management Console (HMC), to be used as the z/vm system operator console. This removes the requirement to have an external 3270 device in order to install and service z/vm systems. The z/vm Stand-Alone Program Loader (SAPL) and stand-alone DASD Dump-Restore (DDR) program support the use of the integrated 3270 as a system operator console. Virtual support enables testing of guest operating systems and utilities that support the integrated 3270 console device. Guest Coupling Enhancement: The z/vm virtual Coupling Facility (CF) support, which was made available in VM/ESA V2.3, has been enhanced in z/vm V4.4 to allow VM/ESA or z/vm systems to run as second-level (or higher) guests while simulating complete OS/390 and/or z/os coupled sysplexes. This enhanced support loads the Coupling Facility Control Code (CFCC) from the Service Element of a zseries or S/390 server into Coupling Facility (CF) service virtual machines in a VM/ESA or z/vm system running as a second-level (or higher) guest of z/vm V4.4. This allows the testing of an OS/390 or z/os Parallel Sysplex environment at any guest level. Note: Each additional level of guest virtualization incurs a performance penalty that may make running a Parallel Sysplex impractical at higher guest levels. Network Virtualization IEEE Virtual Local Area Network (VLAN) Support: IEEE VLAN standard 802.1q describes a mechanism that is designed to enable groups of hosts to be logically connected into a single Local Area Network (LAN) even though they are connected to different switches in different physical locations. Network hosts can be organized into LAN segments which fit an organization s network traffic patterns, rather than being dependent on physical location. In addition, by connecting a router to multiple VLANs, savings can be realized by providing router connections to new LAN segments without having to add additional network interfaces to the routers. To support IEEE VLANs, z/vm V4.4 provides: Enhancements to TCP/IP for z/vm to enable membership in a VLAN for IBM zseries OSA-Express (QDIO) and HiperSocket adapters that support IEEE 802.1q Enhancements to z/vm virtual QDIO and HiperSocket network interface to support VLAN frame tagging as described by IEEE 802.1q Management and control of VLAN identifiers (VIDs) that can be used by guest virtual machines

4 zseries server support for IEEE VLANs in a Linux on zseries environment was delivered for the OSA-Express Fast Ethernet and Gigabit Ethernet features in QDIO mode in April For additional information on zseries IEEE VLAN support in a Linux environment, refer to Hardware Announcement , dated April 30, Network Consolidation using the Virtual Switch: z/vm V4.4 further enhances virtualization technology by introducing a virtual IP switch that is capable of bridging a z/vm Guest LAN to an associated real LAN connected by an OSA-Express adapter. The z/vm virtual switch is designed to help eliminate the need for virtual machines acting as routers to provide IPv4 connectivity to a physical LAN through an OSA-Express adapter. Further, it helps eliminate the need to define a separate routable subnet for the exclusive use of the members of a Guest LAN. Using the virtual switch, the convenience of a Guest LAN is maintained while allowing the guests to be assigned IP addresses in the real LAN subnet. Virtual routers consume valuable processor cycles to process incoming and outgoing packets requiring additional copying of the data being transported. The virtual switch helps alleviate this problem by moving the data directly between the real network adapter and the target or originating guest data buffers. Centralized network configuration and control of the virtual switch within CP helps allow the z/vm Guest LAN administrator to more easily grant and revoke access to the real network and to manage the configuration of Guest LAN VLAN segments. While the z/vm system can be a member of multiple VLANs, the z/vm Guest LAN administrator can control which guests belong to which real VLAN, without requiring additional network adapters or switch port configuration. If a guest does not support IEEE 802.1q, z/vm will transparently join the virtual network interface into the desired VLAN. The z/vm virtual switch employs transparent bridging to enable the switch to dynamically determine and maintain node connectivity so that the LAN administrator has less network maintenance to perform. Guest LAN Support of Internet Protocol Version 6 (IPv6): The z/vm Guest LAN support for OSA-Express simulation in QDIO mode is updated for IPv6. Virtual machines in the z/vm Guest LAN environment can define and use simulated OSA-Express devices that support both the Internet Protocol version 4 (IPv4) and IPv6 protocols. Extended HiperSockets Support: In addition to the VLAN support in TCP/IP for z/vm and Guest LAN simulation, TCP/IP broadcast support is provided for the HiperSockets environment when utilizing IPv4 with z/vm V4.4. Applications that use the broadcast function can now propagate broadcast frames to all TCP/IP applications when using either HiperSockets or an OSA-Express (QDIO) adapter. Technology Exploitation Exploitation of the New IBM eserver zseries 990 (z990): The z990 is the latest zseries enterprise server of the IBM brand, providing new and unique functions and features to meet today s e-business on demand challenges. For additional information, including hardware availability, and to understand the specific benefits of the z990 server, refer to Hardware Announcement , dated May 13, z/vm V4.4 plans to support the following functions of the z990 at the general availability of z/vm V4.4, planned for August 15, 2003: Logical Channel Subsystems A new logical channel subsystem (LCSS) structure is planned to be introduced with the z990 to break the barrier of 256 channels per server machine. This new facility is designed to allow the definition of more than one channel subsystem (CSS), providing channel-path and subchannel controls for configuring channel-subsystem images. Each channel-subsystem image can be configured with up to 256 channel paths, and each logical partition has access to one channel-subsystem image. Support for dynamic-i/o configuration on z/vm V4.4 has been extended to allow channel paths, control units, and devices to be dynamically added, changed, and deleted in multiple logical channel subsystems when z/vm V4.4 is running on a z990 server with the applicable hardware support. This hardware support (such as dynamic-i/o configuration in a multiple logical channel subsystem) is planned to be available in October When z/vm is running on a z990 server and is the controlling logical partition for dynamic-i/o configuration, z/vm is designed to handle all of the new elements of each CSS facility for changing your hardware I/O configuration. To dynamically change the I/O configuration, one of two methods can be employed: z/vm Control Program (CP) suite of interactive dynamic-i/o-configuration commands HCM/HCD new configuration-management tools Spanned Channels Transparent sharing of HiperSockets channels across multiple logical channel subsystems is planned to be supported on the z990 server in October 2003, extending the sharing capabilities now provided by the Multiple Image Facility (MIF). MIF allows sharing of channel resources across logical partitions. In October 2003, HiperSockets channels (IQD CHPID-type) can be defined as spanning channels, which can allow them to be configured as part of multiple logical channel subsystems and to be transparently shared among any subset or all of the configured logical partitions. With z/vm V4.4, spanned HiperSockets channels help enable inter-process communication (IPC) among Linux guests running on VM images in different logical partitions, when the VM images are using different logical channel subsystems. Other IPC configurations are also supported, such as a HiperSockets link between a stand-alone Linux logical partition using one logical channel subsystem and a Linux guest of a VM image in another logical partition using another logical channel subsystem. Extended Channel Measurement Data Support (ECMDS) Extended I/O-measurement facilities are planned to be provided in the z990 server. z/vm V4.4 supports this new z990 function, which provides an extended I/O-measurement block for each subchannel and an extended measurement word at the completion of each I/O operation. Each extended I/O-measurement block has its own 64-bit address, allowing the measurement blocks to be stored in noncontiguous real memory. The extended measurement word provides channel measurement data for each I/O operation when it completes, streamlining measurement processing in the operating system

5 Support for 30 Logical Partitions (LPARs) Prior to the z990, there was an LPAR limit of 15. This limit is planned to be increased for the z990 server in October z/vm V4.4 can handle I/O-configuration definition and dynamic-i/o configuration for up to 30 LPARs. In addition, the CP Monitor has been updated to allow performance data to be collected and recorded for z/vm systems running on servers with more than 15 LPARs. Compatibility Support for the zseries 990 (z990): Compatibility support allows you to run existing z/vm workloads while taking advantage of increased capacity and improved performance of the z990 server. Compatibility support for the z990 server is planned to be available on May 13, 2003, for z/vm V3.1, V4.2, and V4.3 and does not provide exploitation of the following new z990 features: Dynamic-I/O configuration in a multiple LCSS environment 30 LPARs Extended-I/O measurement facilities IPL from FCP-attached disks Adapter interruptions for OSA-Express and FCP channels Performance assist for V=V guests (pageable guests) in Linux and VM environments z/vm V4.4 will exploit these new features when applicable hardware is available. The compatibility support that is being provided for dynamic-i/o configuration is designed to function with a z990 server where only LCSS 0 is defined. Note: Refer to the z/vm subset of the 2084DEVICE Preventive Service Planning (PSP) bucket prior to installing the z990 server. Cascaded FICON Directors: z/vm V4.4 supports the zseries capability to cascade two FICON directors within a Fibre Channel fabric. z/vm and its guests can take advantage of this enhanced and simplified connectivity, which is particularly useful in disaster-recovery and business-continuity procedures. For additional information, including hardware availability, and to understand the specific benefits of a cascaded-ficon-director configuration, refer to Hardware Announcement , dated August 13, The z/vm support for cascaded FICON directors is embodied in support for two-byte Fibre Channel link addresses when defining and configuring control units within the fabric. Previously, only single-byte link addresses were needed in a single-switch fabric. Cascaded FICON directors require the use of two bytes of the standard three-byte Fibre Channel link address; the third byte, for arbitrated-loop support, is not used in FICON fabrics. When using the cascaded-ficon-director function, your hardware-i/o definitions need to be updated accordingly. IBM TotalStorage Enterprise Storage Server (ESS) FlashCopy Support: z/vm V4.4 provides support for the new ESS FlashCopy Version 2 (V2). FlashCopy V2 is designed to enable business continuance solutions with the delivery of new FlashCopy functions and enhancements. FlashCopy V2 is designed to help improve business efficiency, along with FlashCopy performance improvements designed to help minimize operational disruption and includes: Data Set FlashCopy offers a new level of granularity for zseries and S/390 environments. This removes the restriction where source cylinders must be copied to the same physical target cylinders. The removal of this restriction is especially helpful when copying one VM minidisk to another. It allows a non-fullpack source minidisk to be copied to a non-fullpack target minidisk when the minidisks exist on different physical cylinder extents of the same, or different, physical volumes. Multiple Relationship FlashCopy allows a source to have multiple targets. This function allows a guest operating system to copy one source volume to many target volumes. Elimination of Logical Storage Subsystem (LSS) constraint allows a source and target relationship to span across an LSS. This removes the restriction where both the source and target volumes must reside in the same logical control unit. Reduced establish time results from performance improvements designed to provide at least a 10X reduction in FlashCopy Establish Time. This can minimize operational delays and can potentially allow the benefits of FlashCopy to be extended to new and different environments. Detailed information on the IBM ESS V2 FlashCopy capabilities is described in Hardware Announcement , dated May 13, IBM ESS Long Distance Data Copy: z/vm V4.4 provides guest support for the IBM ESS Peer-to-Peer Remote Copy Extended Distance (PPRC-XD) function as described in Hardware Announcement , dated April 23, z/vm guests with DATAMOVER authority in the user directory can copy full volumes of data in non-synchronous mode, which extends the distance between the primary and secondary IBM ESS with a minimal effect on performance. PPRC-XD must be supported by the guest operating system and is suitable for data migration, backup, and disaster recovery procedures. Data can be copied at distances well beyond the 103 km supported for PPRC synchronous transmissions, with the distance typically limited only by the capabilities of the network and channel extension technologies. PPRC-XD support is also available for guest use on all supported levels of VM/ESA and z/vm with no PTFs required. Native VM support for PPRC-XD was previously provided with the availability of ICKDSF Release 17 and ICKDSF running in a CMS virtual machine. Detailed information on ICKDSF Release 17 capabilities is described in Software Announcement , dated November 12, z/vm V4.4 also supports PPRC Version 2 (V2) for guest operating systems, designed to offer an Asynchronous Cascading solution providing a complete, consistent, and coherent copy of data at a remote site. Asynchronous Cascading provides a two-volume synchronous PPRC implementation with a non-synchronous third volume serving as a backup device that can provide a multi-site, long distance solution. Detailed information on the IBM ESS PPRC V2 capabilities is described in Hardware Announcement , dated May 13, IBM ESS Model 800 Support: z/vm V4.4 provides support for the IBM ESS Model 800 described in Hardware Announcement , dated July 15, The

6 Model 800 provides storage consolidation to support your growing, performance-demanding applications. IBM TotalStorage Enterprise Tape Drive 3590 Model H Support: z/vm V4.4 provides support for the 3590 Model H tape drives described in Hardware Announcement , dated June 25, The 3590 Model H11 and H1A tape drives offer increased cartridge capacity to meet the growing needs of new and existing customers. PTFs for APARs VM63073 (CMS), VM63074 (CP), and VM63097 (DFSMS/VM ) to support these tape drives are available for z/vm V3.1 and all currently supported releases of z/vm V4. Systems Management Improvements Improvements in systems management help to provide self-configuring, self-managing, and self-optimization facilities. Hardware Configuration Definition (HCD) and Hardware Configuration Manager (HCM): z/vm V4.4 provides HCM and HCD support for the z/vm environment to create and manage your I/O configuration definition. This new support provides a comprehensive, easy-to-use I/O-configuration-management environment similar to that available with the z/os operating system. The HCM and HCD programs work together to help you easily create and manage the hardware and software aspects of your I/O configuration. HCM executes on a personal computer running Microsoft Windows and is connected to the z/vm system via a TCP/IP network. HCM provides an easy-to-use graphical user interface along with commands to help you configure your system. You provide the needed I/O-configuration information to HCM, which processes the information and passes it to HCD. HCD runs in a z/vm server virtual machine and performs the work of creating and changing the hardware aspects of your machine configuration and the software aspects of the z/vm I/O configuration. HCM provides the primary user interface to HCD, but HCD provides a backup user interface on your z/vm host for performing certain I/O configuration tasks when HCM is not available. HCM and HCD on z/vm can be used to create and manage the hardware aspects of the I/O configuration definition for all of the LPARs in your zseries or S/390 server. For the software aspects, HCM and HCD create and manage the I/O configuration definition only for the z/vm system on which they are running. The I/O configuration definitions for operating systems in other logical partitions are handled by those operating systems (which may also include instances of HCM and HCD). The new HCM and HCD support is an easy-to-use alternative to create and manage your I/O configuration definition compared to the existing z/vm method of using the I/O Configuration Program (IOCP) and CP s dynamic-i/o configuration CP commands. The two methods are not intended to be used together, so you should choose one or the other method for your installation. Performance Toolkit for VM: The Performance Toolkit for VM provides enhanced capabilities for a z/vm systems programmer, operator, or performance analyst to monitor and report performance data. The Performance Toolkit for VM is a priced, optional feature and is derived from the FCON/ESA program (5788-LGA), which was not available in all countries. The function provided by the Performance Toolkit for VM includes: Full-screen mode system console operation Management of multiple z/vm systems (local or remote) Post-processing of Performance Toolkit for VM history files and of VM monitor data captured by the MONWRITE utility Viewing of performance monitor data using either Web browsers or PC-based 3270 emulator graphics TCP/IP performance reporting In addition to analyzing VM performance data, the toolkit processes Linux performance data obtained from the Resource Measurement Facility (RMF ) Linux performance gatherer, rmfpms. The rmfpms application is available from the zseries Web site at: pmlin.htm The Linux performance data obtained from RMF can be viewed and printed similarly to the way VM data is presented. The RealTime Monitor (RTM) and Performance Reporting Facility (PRF) priced, optional features will continue to be supported in z/vm V4.4. Future performance management enhancements are planned to be made primarily to the Performance Toolkit for VM. It is planned that this will be the last release in which the RTM and PRF features will be available with z/vm. Both the RTM and PRF features are planned to be withdrawn from marketing in a future z/vm release. The Performance Toolkit for VM feature is considered a program update for the RTM and PRF features as defined in the z/vm Addendum for Support, Z Therefore, any existing z/vm V4 customers who have purchased a Subscription and Support (S&S) license for the RTM or PRF features are entitled to receive a no-charge upgrade to the Performance Toolkit for VM feature when ordering z/vm V4.4. Because the Performance Toolkit for VM feature was derived from the IBM FCON/ESA program (5788-LGA), any existing VM customers who have previously purchased FCON/ESA are entitled to receive a no-charge upgrade when ordering the Performance Toolkit for VM feature with z/vm V4.4. FCON/ESA is planned to be withdrawn from marketing for all releases of VM/ESA and z/vm at the availability of the Performance Toolkit for VM feature on z/vm V4.4. Service support for FCON/ESA will end at the announced end of service date of the VM operating system on which FCON/ESA is running or on December 31, 2004, whichever is earlier. The discontinuance of support for FCON/ESA is planned to occur as follows:

7 VM Release Announced Date of FCON/ESA or Version VM EOS Date Announcement EOS Date VM/ESA V2.4 June 30, 2003 February 5, 2002 June 30, 2003 z/vm V3.1 Not announced Not announced December 31, 2004 z/vm V4.1 June 30, 2003 May 29, 2001 June 30, 2003 z/vm V4.2 December 30, 2003 October 4, 2001 December 30, 2003 z/vm V4.3 December 31, 2004 May 13, 2003 December 31, 2004 IBM strongly encourages you to order an S&S license for the Performance Toolkit for VM feature. S&S provides an easy and effective way to gain access, during the coverage period, to eligible new versions and releases of the Performance Toolkit for VM feature. S&S also provides you with the level of support that was available when you paid a monthly license charge for RTM and PRF products with the IBM Customer Agreement software license. Automated Shutdown of the Shared File System (SFS): Utilizing the automated shutdown support added in z/vm V4.3, SFS now shuts down automatically when CP is shut down. Use of this function helps ease z/vm system administration and helps maintain the integrity of the SFS and its data. Installation and Service Improvements: The installation process continues to be improved in z/vm V4.4 by reducing the amount of system administrator input required to complete a z/vm installation. The improvements allow the service disks of the z/vm components to reside in the shared file system (SFS). Moving a product or component into SFS directories has been automated more completely. The current practice of maintaining a complete copy of z/vm for each LPAR wastes disk space. Utilizing SFS for installation allows sharing a single copy of source and service between LPARs and can help you reduce the DASD costs associated with installing and maintaining z/vm. New functions preinstalled in z/vm V4.4 include: HCD and HCM for z/vm V4.4 Performance Toolkit for VM Language Environment for z/vm is now an integrated component of z/vm V4.4 rather than a no-charge feature as it was with previous releases of z/vm. The service process has been updated to support removing local modifications. Full VMSES/E message log support has been added to the SERVICE and PUT2PROD EXECs so that message logs can be viewed using the VMFVIEW command. Network Performance and Security TCP/IP Stack Performance: The performance of the TCP/IP stack has been improved by optimizing additional high-use code paths and adding virtual multiprocessing capabilities. In z/vm V4.3, stack performance focused on environments in which routing and gateway functions were the primary functions being performed. The z/vm V4.4 enhancements concentrate on environments in which the z/vm TCP/IP stack and its clients and servers are the source of or final destination for network traffic. TCP/IP Stack Security: The overall security and auditability of the TCP/IP for z/vm stack and the integrity of the z/vm system have been improved to aid self-protection by providing better controls, monitoring, and defaults, which include: Logging of all TCP/IP administrative commands, including NETSTAT, OBEYFILE, and IFCONFIG, that alter, or attempt to alter, the active IP or Control Program (CP) configuration Changing the defaults of the AssortedParms statement within the TCP/IP configuration to: RestrictLowPorts, to prevent impersonation ( spoofing ) of well-known TCP/IP for z/vm applications by CMS users or guest operating systems VarSubnetting, to avoid creating unnecessary TCP/IP routing table entries IMAP User Authentication Exit: User authentication for the IMAP server can now be performed via an exit that makes use of a separate virtual machine (IMAPAUTH). This support removes prior user ID and password-length restrictions, and is designed to help eliminate the need for every IMAP client to have a VM user ID and password. Instead, authentication is handled by a user-written exit routine, providing greater flexibility for choosing authentication methods. Included in the exit support is the ability to pre-authenticate connections by IP address. Application Enablement C/C++ for z/vm Compiler: The new C/C++ for z/vm compiler (5654-A22), which executes on CMS, helps enable you to increase your programming productivity through state-of-the-art application development tools and to exploit the latest in software development technology. The new compiler allows C/C++ programs to be compiled and executed on CMS and creates portability between z/vm and z/os C/C++ programs. C/C++ source files can be read from a CMS minidisk, the SFS, or the Byte File System (BFS) and output can be written to any of these file systems. C/C++ can be ordered stand-alone or through the z/vm V4.4 SDO order. The C/C++ for z/vm compiler executes only on z/vm V4.4 and can only be licensed to operate on standard processor engines. Applications compiled with the C/C++ compiler will run on standard processor engines or Integrated Facility for Linux (IFL) processor features. For additional information on the C/C++ for z/vm compiler, refer to Software Announcement , dated May 13, In order to support the C/C++ for z/vm compiler and updated runtime libraries for C/C++, COBOL, and PL/I, the Language Environment component has been updated to the level shipped with z/os V1.4 and is integrated in z/vm V4.4. Engine-Based Software Pricing The engine-based pricing introduced in z/vm V4.1 may make it financially more attractive to add VM workload to an existing z990, z900, z800, or S/390 server. Engine-based pricing means customers will pay a one-time charge (OTC) for each processor engine. This can be for standard processor engines or IFL processor features. For additional information on the z/vm support for the Integrated Facility for Linux, refer to Software Announcement , dated May 29,

8 z/vm V4 includes engine-based pricing for the optional features RTM, PRF, DirMaint, RACF, and the Performance Toolkit for VM. These optional features are automatically installed with z/vm V4.4 but you must have a license to use them and they must be enabled by the customer. The Recommended Service Upgrade (RSU) tape must be applied for these features to support new functions of z/vm V4.4. When placing your order for z/vm V4, you must specify the quantity of engines required. When z/vm V4.4 or the optional features are ordered to operate on standard processor engines, the quantity of engines ordered must be equal to the number of standard processor engines on the zseries or S/390 server. When z/vm V4.4 or the optional features are ordered to operate on processor engines with the IFL feature, the quantity of engines ordered must be equal to the number of processor engines with the IFL processor feature on the zseries or S/390 server. System Delivery Offering (SDO) Ordering Process The z/vm V4 SDO (5739-A04) is the IBM program offering for z/vm that includes a single deliverable containing z/vm V4.4 and a group of fully serviced IBM licensed program products. The z/vm SDO is the only method that can be used to order z/vm V4.4 and ordering capability is planned for August 1, The z/vm SDO package contains: z/vm V4.4 System DDR z/vm SDO Optional Products Enabling Aid, formerly known as the SDO Enabling Feature z/vm SDO licensed products The key benefit of the z/vm SDO is the ability to dynamically update the SDO with new program products (versions and releases), with service when available, or to delete withdrawn products. This capability makes the SDO a more effective solution for the z/vm V4 customer. For a complete list of products available and serviced with the z/vm V4.4 SDO or for more information on the z/vm SDO, refer to the z/vm Web site at: The z/vm V3.1 SDO (5654-A20) will still be orderable after the availability of the z/vm V4.4 SDO. Solution Financing for zseries from IBM Global Financing Moving your company to the world of e-business presents significant opportunities, but with real challenges that include sourcing, integrating new applications, and establishing better ways to serve your customers. Today s highly competitive environment often requires creative financial solutions to execute timely IT projects that are essential to your future. IBM Global Financing (IGF) can help you perform a successful and cost-effective transition to the e-business marketplace by making your zseries solution more affordable. A Financing Solution with Important Benefits Costs for z/vm V4.4, including the OTC per processor engine as well as the charges for Subscription and Support (S&S), can be financed with payments designed to fit your budget. Payments for your entire zseries solution (hardware, OTC software, services, both IBM and non-ibm components) can be designed to match benefit flow. Flexibility is crucial because the financial rewards of Linux implementation are realized at project end, while costs are incurred at the start. Possible off-balance-sheet financing may help eliminate the need for additional equity by spreading up-front costs over the project life. You receive a single invoice for all financed elements of solution. Finance Some or All of Your Linux Solution zseries, S/390 Parallel Enterprise Server Generation 5 and 6, or Multiprise 3000 server acquisition, and any associated additional hardware Linux for zseries or S/390, z/vm, WebSphere, and other IBM initial license charge (ILC) software, and independent software vendor (ISV) OTC software Services, from either IBM, IBM Business Partners, or an ISV Extended maintenance offerings For help in designing the correct financing approach for your Linux for zseries and S/390 solution, contact your IGF representative or visit the IGF Web site at: Accessibility by People with Disabilities The following features support use by people with disabilities: Operation by keyboard alone Optional font enlargement and high-contrast display settings Screen readers and screen magnifiers tested for use by people with visual impairment Product Positioning z/vm V4 is ideally suited for deploying Linux workloads on the mainframe. Attractive per-engine pricing of z/vm V4 and support for Integrated Facility for Linux (IFL) processor engines enable customers to add processor capacity to existing zseries systems without necessarily increasing the cost of running existing workload on standard engines. z/vm V4 is the follow-on product for VM/ESA and z/vm V3. It provides additional support and exploitation opportunities for the thousands of users who have built enterprise-wide automation and infrastructure enhancements on the VM platform in support of their applications, database systems, and e-business solutions. z/vm V4 is a component of the zseries Offering for Linux and the IBM Integrated Platform for e-business on zseries. For information on the zseries Offering for Linux, refer to Hardware Announcement , dated January 29, For information on the IBM Integrated Platform for e-business on zseries, refer to Software Announcement , dated January 21, z/vm V4 is intended to address the following situations: Running more Linux server images on a single zseries server. Considerably more images than are currently supported by the LPAR mode of operation (30 on the z990 on October 31, 2003, and 15 on all other zseries

9 servers) can be achieved with z/vm guest support. These Linux images can be deployed on standard processor engines or IFL processor features with z/vm V4. Running multiple Linux images on an IFL-configured z/vm system may not increase the IBM software charges of your existing zseries environment. z/os, z/os.e, OS/390, z/vm, VM/ESA, TPF, VSE/ESA, or Linux customers can add z/vm V4 running on IFL engines to their environment and at a z/vm Version 4 price that corresponds to the number of IFL engines. Moving select Linux, Windows, and UNIX workloads to a single zseries server while maintaining distinct server images and current LAN topology. This ability can help reduce systems management complexity. Since the number of real hardware servers and associated physical LANs is reduced, cost savings may be realized by managing large server farms deployed on virtual servers instead of using multiple hardware servers. Deploying Linux workloads on z/vm V4 is particularly attractive if they interact with applications or data located on the same zseries server. Running multiple Linux server images in a logical partition or basic (native) mode. Basic mode requires the use of standard processor engines on a zseries 800, 900, Generation 5 and 6, and Multiprise IFL engines and the z990 are usable only in LPAR mode. The performance benefits of exploiting V=R or V=F for I/O assist will be lost when z/vm runs in a logical partition. VM systems running in LPAR mode are not eligible for I/O assist. Enhanced virtual networking. This provides external IP connectivity for Guest LANs through an OSA-Express adapter without requiring a VM or Linux router virtual machine. Executing 64-bit operating systems running on a zseries server. This helps provide real memory constraint relief by running z/vm V4 in 64-bit mode on a zseries server. This constraint relief is provided for ESA/390 guest operating systems. Running TPF or large numbers of MVS, OS/390, VSE/ESA, or Linux images may benefit from z/vm 64-bit support. Migrating from the S/390 Virtual Image Facility for Linux to z/vm V4. This helps enable additional flexibility offered in the virtual machine environment, which includes: Guest support for zseries operating systems, including OS/390, z/os, z/os.e, VSE/ESA, TPF, Linux, z/vm, and VM/ESA System administration tools for performance, accounting, auditing, and more Enhanced security features A wider variety of supported disk and tape storage devices A number of ease-of-use features The Virtual Image Facility is no longer serviced by IBM. Migrating from VM/ESA to z/vm V4. This helps enable: More memory to cache CMS minidisks by exploiting memory above 2 GB for minidisk cache Connectivity (TCP/IP) enhancements and additional device support Added security with SSL-protected TCP/IP network traffic, such as Telnet sessions and Web transactions Note: IBM will discontinue program services for VM/ESA V2.4 on June 30, Migrating from OS/390 to OS/390 R10, z/os, or z/os.e using z/vm V4. This provides added flexibility for migration, production, and testing. Enhanced guest Parallel Sysplex support in z/vm with the exploitation of the z/architecture (64-bit). This enables addressability of larger amounts of real and virtual memory, allowing the development and testing of 64-bit Parallel Sysplex applications in a guest environment. z/vm V4.2 and subsequent releases provide a duplexing capability for coupled guests in a Parallel Sysplex system environment. Statement of Direction IBM plans to take the following actions in the future: IBM intends to support greater than 16 processors in a single image of z/vm on future releases. This support is planned to increase the scale of a single z/vm image by taking advantage of the increased number of processors available on a z990 server. Future releases of z/vm will impose an Architectural Level Set (ALS) requiring z/architecture. Consequently, future releases of z/vm are planned to operate only on IBM zseries 990, 900, and 800, equivalent servers, and future zseries servers. IBM intends for z/vm V4.4 or later to provide support for up to 60 logical partitions (LPARs) with corresponding support on a z990 or future server. IBM intends for z/vm V4.4 or later to support four logical channel subsystems with a maximum of 256 channels supported per image with corresponding support on a z990 or future server. IBM intends to provide z/vm guest support of the PCIX Cryptographic Coprocessor (PCIXCC) adapter. It is planned that future performance management enhancements will be made primarily to the Performance Toolkit for VM. z/vm V4.4 is planned to be the last release in which the RTM and PRF features will be available. Both the RTM and PRF features are planned to be withdrawn from marketing in a future z/vm release. All statements regarding IBM s plans, directions, and intent are subject to change or withdrawal without notice. Any reliance on this Statement of Direction is at the relying party s sole risk and will not create any liability or obligation for IBM. Reference Information For information on the zseries 990, refer to: Hardware Announcement , dated May 13, 2003 For information on the zseries 800, refer to: Hardware Announcement , dated February 19, 2002 For information on the zseries Offering for Linux, refer to: Hardware Announcement , dated January 29,