Modernization Hub

HV - Hypervisor

Enhanced Definition

In the IBM mainframe context, a hypervisor is a layer of software or firmware that creates and runs virtual machines (VMs). The primary hypervisor on IBM zSystems is **PR/SM (Processor Resource/Systems Manager)**, which enables a single physical mainframe to be divided into multiple logically independent operating environments, known as Logical Partitions (LPARs). Each LPAR can run its own instance of an operating system like z/OS, z/VM, or Linux on Z.

Key Characteristics

    • Hardware-assisted Virtualization: PR/SM is deeply integrated with the zSystem hardware, leveraging specialized instructions and architectural features for highly efficient and secure virtualization.
    • Logical Partitioning (LPARs): It creates isolated LPARs, each with dedicated or shared access to CPU, memory, and I/O resources, providing strong workload separation.
    • Resource Management: PR/SM provides sophisticated mechanisms for allocating, sharing, and managing processor, memory, and I/O resources among LPARs, including weight-based dispatching and capping.
    • High Isolation and Security: LPARs are highly isolated from each other, preventing failures or security breaches in one LPAR from impacting others, which is crucial for multi-tenant and critical environments.
    • Dynamic Resource Changes: Resources (e.g., CPUs, memory) can often be dynamically added to or removed from LPARs without requiring a system reboot, enhancing flexibility and availability.
    • Firmware-based: PR/SM is primarily a firmware component of the zSystem hardware, conceptually similar to a "Type-0" or "bare-metal" hypervisor, offering minimal overhead.

Use Cases

    • Workload Consolidation: Running multiple z/OS instances, z/VM, or Linux on Z distributions on a single physical mainframe to reduce hardware footprint, power consumption, and operational costs.
    • Environment Isolation: Separating production, development, test, and disaster recovery environments into distinct LPARs to prevent interference and enhance system stability and security.
    • Multi-tenancy and Cloud: Hosting multiple independent clients, departments, or cloud instances on a single mainframe, each within its own secure and isolated LPAR.
    • Disaster Recovery and High Availability: Creating mirrored LPAR configurations across geographically separated mainframes for robust disaster recovery and continuous operations.
    • System Upgrades and Testing: Providing dedicated LPARs for testing new z/OS versions, application upgrades, or system configurations before deployment to production.

Related Concepts

The hypervisor (PR/SM) is foundational to the concept of Logical Partitions (LPARs), as it is the technology that creates and manages these isolated environments. It works in conjunction with z/VM, which itself is a Type-1 hypervisor that runs *within* an LPAR to further virtualize resources and host hundreds or thousands of guest operating systems (e.g., Linux on Z, z/OS development instances). This layered virtualization enables extreme flexibility, resource utilization, and consolidation capabilities on zSystems.

Best Practices:
  • Strategic Resource Allocation: Carefully plan CPU, memory, and I/O allocations for each LPAR based on workload requirements to ensure optimal performance and prevent resource contention.
  • LPAR Weighting and Capping: Effectively utilize PR/SM LPAR weights and capping to prioritize critical workloads and manage resource sharing among LPARs, ensuring service level agreements are met.
  • Performance Monitoring: Implement robust monitoring tools (e.g., RMF, SMF) to continuously track LPAR resource utilization, identify potential bottlenecks, and optimize resource distribution.
  • Security Configuration: Ensure proper security configurations for LPARs, including access controls, network segmentation, and adherence to security policies, to maintain isolation and protect sensitive data.
  • Dynamic Resource Management: Leverage dynamic LPAR capabilities to adjust resources (e.g., add or remove CPUs, memory) on the fly, responding to changing workload demands without service disruption.

Related Vendors

IBM

646 products

Trax Softworks

3 products

Related Categories

Operating System

154 products

Browse and Edit

64 products