Linux_on_Z_Uniforum.ppt

IBM Linux on zSeries

© 2008 IBM Corporation

Linux on System ZAn Introduction

Scott O. [email protected]


Linux is Everywhere © 2008 IBM Corporation

TrademarksThe following are trademarks of the International Business Machines Corporation in the United States and/or other countries.

The following are trademarks or registered trademarks of other companies.

* Registered trademarks of IBM Corporation

* All other products may be trademarks or registered trademarks of their respective companies.

Intel is a registered trademark of the Intel Corporation in the United States, other countries or both.

Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both.

Java and all Java-related trademarks and logos are trademarks of Sun Microsystems, Inc., in the United States and other countries.

UNIX is a registered trademark of The Open Group in the United States and other countries.

Microsoft, Windows and Windows NT are registered trademarks of Microsoft Corporation.

SET and Secure Electronic Transaction are trademarks owned by SET Secure Electronic Transaction LLC.

Notes:

Performance is in Internal Throughput Rate (ITR) ratio based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput that any user will experience will vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve throughput improvements equivalent to the performance ratios stated here.

IBM hardware products are manufactured from new parts, or new and serviceable used parts. Regardless, our warranty terms apply.

All customer examples cited or described in this presentation are presented as illustrations of the manner in which some customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics will vary depending on individual customer configurations and conditions.

This publication was produced in the United States. IBM may not offer the products, services or features discussed in this document in other countries, and the information may be subject to change without notice. Consult your local IBM business contact for information on the product or services available in your area.

All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.

Information about non-IBM products is obtained from the manufacturers of those products or their published announcements. IBM has not tested those products and cannot confirm the performance, compatibility, or any other claims related to non-IBM products. Questions on the capabilities of non-IBM products should be addressed to the suppliers of those products.

Prices subject to change without notice. Contact your IBM representative or Business Partner for the most current pricing in your geography.

DB2*DB2 ConnectDB2 Universal Databasee-business logoGDPS*Geographically Dispersed Parallel SysplexHyperSwapIBM*IBM eServerIBM logo*Parallel Sysplex*

System zTivoli*VM/ESA*WebSphere*z/OS*z/VM*zSeries*



Agenda

Introduction to Linux on System Z System Z Hardware Overview Hypervisor Comparisons Linux on System Z Value Proposition Platform Placement Guidelines

► Operating System

► Hardware Best Fit and Good Fit Examples



What is Linux

A fully-networked UNIX-like operating system

Multi-user, multitasking, multiprocessor Coexists with other operating systems Open Source software

►Community Development Network

►Fosters Fast Technical Innovation & Support

►Guarantees No Single Vendor Control

Runs on multiple platforms

PervasiveComputing

IntelligentNetworking

Supercomputing

Approximately 1% of Linux code is platform specific

Applications are not binary compatible across platforms



What is Linux on System Z

A native System Z operating environment► Pure Linux, an ASCII environment

► Exploits IBM S/390 hardware, including IEEE floating point

► Linux for S/390 - 32-bit

► Linux for zSeries - 64-bit

Not a unique version of Linux or other operating system► Developed and supported by the Open Source Community

► Distributed by SuSE, RedHat, and Others

Not a replacement for other IBM System Z operating systems Can be run under z/VM or in its own LPAR

► Hardware provides a pool of capacity



What System Z Hardware Brings to Linux The most reliable hardware platform available

► MTF measured in decades

► RAS features built into hardware Scalability

► Both Physical and Logical

► Non-disruptive capacity upgrade on demand Designed to support mixed work loads

► Complete work load isolation

► High speed inter-server connectivity

► High Internal Bandwidth, sophisticated cache nest

► Virtualization Hipersockets

► Virtual network between LPARs, at memory speed Integrated Facility for Linux (IFL)

► Lower priced hardware

► Protects against software costs



Linux on System Z Value Proposition

Virtualization -- Primary Value► Consolidation of many servers – “Lots of Little”

► Cost savings

– Software– People, particularly in volatile environment

► Lifecycle management

– Rapid provisioning– Reclamation and reuse

► Environmental issues

– Floorspace– Power and cooling



IBM System z10

IBM System z10



InternalBatteries(optional)

PowerSupplies

3x I/Ocages

Fiber Quick Connect (FQC) Feature

(optional)

Processor Books, Memory, MBA and

HCA cards

2 x CoolingUnits

InfiniBand I/O Interconnects

2 x SupportElements

FICON & ESCON FQC

Ethernet cables for internal System LAN connecting Flexible Service Processor

(FSP) cage controller cards

z10 EC – Under the covers (Model E56 or E64)



Quad-core chip

MCM

Memory DIMMs



z10 Configuration Options

1 – 4 books per machine► 5 different configurations

– Special high-end configuration is fifth

Each book has 12 – 17 general purpose processors► Processors are purchased and turned on individually

► Additional processors on MCM used for spares and SAPs

Maximum of 64 general purpose processors per machine



Non-disruptive Hardware Changes

Books can be added, and in some situations removed

Unused engines on MCMs can be turned on and off ► Disaster recovery

► Temporary capacity

► Upgrades

New resources immediately available to all Linux virtual machines



Simplified Design Differences Between Distributed and Z

Core Core Core Core

Core Core Core Core

Bus

Healthy CoresAdequate Bus

Great Performance*Good Throughput*

Core

Bus

Core Core Core

Core Core Core Core

Adequate CoresHealthy Bus

Good Performance*Great Throughput*



Comparison of 64-way MachinesSystem Z

Application Processors

Application Processors

System Assist

Processors

SparesCross check processors

Other Machines

Additional processors on I/O Cards



15

Hypervisor software/firmwareruns directly on server

Hypervisor software runs ona host operating system

System z LPAR and z/VM®

POWER™ HypervisorVMware ESX ServerXen Hypervisor

VMware ServerMicrosoft® Virtual ServerHP Integrity VMUser Mode Linux®

S/370™ SI-to-PP and PP-to-SI, Sun Domains, HP nPartitions

Logical partitioning

Physical partitioning

IBM eServer™ pSeries® LPAR,HP vPartitions

Adjustablepartitions

PartitionController

...

SMP Server

OS

Apps

OS

Apps

Hypervisor

SMP Server

...OS

Apps

OS

Apps

Host OS

SMP Server

Hypervisor

...OS

Apps

OS

Apps

Hardware Partitioning Bare-metal Hypervisor Hosted Hypervisor

Server is subdivided into fractionseach of which can run an OS

Hypervisor provides fine-grainedtimesharing of all resources

Hypervisor uses OS services todo timesharing of all resources

Characteristics:

• Bare-metal hypervisors offer high efficiency and availability• Hosted hypervisors are useful for clients where host OS integration is important• Hardware partitioning is less flexible than hypervisor-based solutions

Server Virtualization Approaches



16

Trap and Emulate

Hypervisor Calls (“Paravirtualization”) Direct Hardware Virtualization

Examples: CP-67, VM/370Benefits: Runs unmodified OSIssues: Substantial overhead

LoadAddStorePrivOpLoad...

Hypervisor PrivOpemulation code

• VM runs in user mode• All privileged instructions

cause traps

Trap

Examples: POWER Hypervisor, XenBenefits: High efficiencyIssues: OS must be modified to issue Hcalls

LoadAddStoreHcallLoad...

Hypervisor service

• VM runs in normal modes• OS in VM calls hypervisor

to access real resources LoadAddStorePrivOpLoad...

Hypervisor service

• VM runs in normal modes• Hardware does most of the

virtualization (SIE architecture)• Hypervisor provides control

Exit

Examples: System z LPAR, z/VMBenefits: High efficiency, runs unmodified OSIssues: Requires underlying hardware support

Virt Mach

Virt MachVirt Mach

Translate, Trap, and Emulate

LoadAddStoreTrapOpLoad...

Hypervisor PrivOpemulation code

• VM runs in user mode• Some IA-32 instructions must

be replaced with trap ops

Trap

Examples: VMware, Microsoft VS Benefits: Runs unmodified, translated OSIssues: Substantial overhead

Virt Mach

Call

Hypervisor Implementation Methods

Hypervisor callsalso supported



17

Supports a wide range of unmodified Windows and Linux versions Guest OS runs in User Mode; privileged instructions trap to Virtual Machine Monitor

(VMM)► “Trapping and mapping” is a significant source of performance overhead

Guest OS binary code is translated incrementally at load time► Instructions that behave differently in User Mode vis-à-vis Supervisor Mode

must be replaced with explicit trap instructions so the appropriate behavior will occur

Modified Linux device drivers run in the VMkernel Intel VT hardware feature is used only for 64-bit guests

Memory nicnicNICDiskCPU

ServiceConsole

OSVMkernel

Scheduler MemoryMgmt

SCSIDriver

EthernetDriver

VMM VMM VMM VMM

Application Application Application Application

Guest OS Guest OS Guest OS Guest OS

x86 SMPHardware

VMwareESX

Traps

Cannot add or remove VM resources on the fly

No isolationor protectionof VMs fromfailures inI/O adaptersor drivers

VMware ESX Server



Virtualize everything with up to 100% utilization rates► CPU, memory, network, I/O, cryptographic features, coupling facility, ...

Massively scale your workload on a single System z mainframe► The Linux-on-z/VM record is 97,943 virtual machines► Each virtual machine on z/VM can access up to 24,576 devices

Security for everything► Highest security classification for general purpose servers in the world ► System z LPAR technology is EAL 5 certified

Non-disruptively add anything► 54x CPU scalability per mainframe, 32x CPU scalability per z/VM LPAR ► z/VM is designed to support up to 8 TB of active virtual memory

IBM System z: The Ultimate Virtualization Platform

Consolidate all typesof workloads

Smart economics: start smalland grow big in the same box

Secure your virtualservers and reduce

business risk

Rapidly respond toworkload spikes



IFL1 IFL2 IFL3CP1 CP2 CP3 CP4 CP5

LPAR1

z/OS

LPAR2

z/OS

LPAR3

Linux

z/VM

Linux

LPAR4

z/VM

z/VM

Linux

Linux

IBM System z Server

RealCPUs

LogicalCPUs

RealCPUs

LogicalCPUs

VirtualCPUs

Virtual2

CPUs

IBM System z Server VirtualizationLinux Example

Linu

x

Linu

x

Linu

x

Linu

x

Linu

x

Linu

x

Linu

x

Linu

x

Linu

x



Virtualization with z/VM V5.4

LPAR

z/VM

Linux

Memory

I/O and Network

Linux

RealResources

CPU

VirtualResources

z/OSCMSLinux

Configure virtual machineswith z/VM-unique facilities

Up to 256 channel paths

Add virtual CPUsnon-disruptively (up to 64)

z/VM can provision virtual machines with a mix of real and virtual resources with exceptional levels of scalability, availability and security

Up to 256 GB*

Up to 32 CPUs*

Simulate resourcesnot in the LPAR

Optimize virtual servers withdedicated real resources

More than 1 TB* (in aggregate)

* z/VM V5.4 maximums



System z Virtualization Architecture Summary

IFL Processors

Memory

z/VM

Linux Linux Linux Linux

L P A R

Memory

z/OS

L P A R

Control Program

WebSphere

Memory

L P A R

Processor Resource / System Manager (PR/SM)

Traditional OLTP and

BatchApacheDB

Server

WebSphere

Test

HiperSockets & Virtual Networking and Switching

WLM WLM

I/O & Network

Intelligent Resource Director (IRD)

Processors

z/OS

z/VM

Linux

FTP

z/OS

Test

Memory

L P A R

Memory dedicated to an LPAR,can be reconfigured

I/O shared among entire machine



Managed by Linux

z/VM Technology: Disk SupportSupports Both Traditional Mainframe and FCP Disk

z/VM Minidisks► Used for user filesystems

► Partial or full pack

► Shared or exclusive

► Can use minidisk caching

Dedicated FC devices► Full FC LUNs

► Managed by Linux guest

Virtual Disk (VDISK)► In Memory disk,

► Ideal for Linux page spaces

Temporary Disks (TDISK)► Temporary work or utility disks. You can mix/match both traditional z/VM

(ECKD) and FC attachment for each guest

ECKD Storage DeviceManaged by z/VM CP

z/VM Memory

Linux

z/VM Control Program

Linux

VDISK

z/VM Minidisk Cache (ECKD Devices Only)

TDISK

TDISKShared

MinidiskMinidisk

Minidisk

MinidiskFC

Disk

Linux

VDISK VDISK

…



z/VM z/VM

System z and N_Port ID Virtualization (NPIV)

Linux1 Linux2 Linux3 Linux4Linux2Linux3Linux4

Linux1 Linux1 Linux1 Linux1Linux2 Linux2 Linux2 Linux2Linux3 Linux3 Linux3 Linux3Linux4 Linux4 Linux4 Linux4

Without N_Port ID Virtualization

FCP DiskWith N_Port ID Virtualization

= Virtual Worldwide Port Name (WWPN)

Linux1 Linux2

Linux4Linux3

Linux1 Linux2

Linux4Linux3



Linux on System Z Value Proposition

Virtualization Close Proximity to z/OS Data

► Hipersockets

Other zSeries Features► Disaster Recovery

► Availability

– Ability to create HA configuration with no additional hardware► Backup Capabilities

► Security

– The most secure network is no network► Technology Refresh Cycle

► Others



Everyone can virtualize

So what makes System Z different….



Memory

Disk C CC

Simplified Computer Architecture

P PP

C

Cache design can have a great impact on performance

and workload characteristics

SAP Front side bus



Memory

Processors

Cache

Private Cache C-C Bus Shared Cache

Reported BW Zero ZeroC-C Bus Speed

Effective BW C-M-C Speed C-C Bus Speed Infinite

Internal Bandwidth Example



What determines system capacity

Single system capacity is determined by:► Processor speed

► Memory hierarchy

► I/O structure

There's more to performance than just processing power

CPU Time Memory Time I/O Time

CPU Busy I/O Busy

Processor, memory, and I/O times vary greatly by application and by machine type



Relative single system capacity

There's more to performance than just processing power


CPU Busy


zSeries

Others


zSeriesCPU Time Memory Time I/O Time Others

Data Intensive Workloads

Compute Intensive Workloads

The relative difference between Data Intensive andCompute Intensive is an order of magnitude



Fragmented Whitespace

Growth

Headroom

Spike

Average

CapacityQuantum

A "Typical" IntelServer Utilization Profile

zLinux

Why Virtualization Works

VMWare



Sweet Spots of Various Platforms

WinTel ► Low hardware cost

► Dedicated servers

► Software that only runs on WinTel

System Z► Virtualization (lots of little)

► Data intensive

► Skewed or skewless OLTP

► Large or unpredictable working sets

► Applications with fine grained interactions

► Lots of different applications simultaneously

► High QoS Requirements

System P ► Virtualization (moderate amount of

medium)

► Compute intensive

► Parallel applications

► Skewless OLTP

► Controlled or distributable working sets

► Workload with common characteristics



Hardware selection for Linux applicationsTechnical Considerations Other Considerations

Quality of Service

Data intensity

Speed of deployment

Other

Architecture

Application availabilityCertification of solution on hardware/software platform

Workload ManagementManageability and scaling characteristics

Especially DB2 on z/OSProximity of data to applicationThe best network is an internal network!

Compute intensity

System Z

Other

Architecture

Other

Architecture

Other

Architecture

System Z

System Z

System Z



Where to deploy – z/OS or Linux on System Z?Technical Considerations Other Considerations

Quality of Service

Degree of portability

Speed of deployment

Linux z/OS Application availability

Workload Management function and granularity

File sharing across a Sysplex

Manageability and scaling characteristics

Availability of skill

Linux

Linux

z/OS

z/OS



z/OS and zLinux Comparison

WLM► Most sophisticated► Allows highest utilization

Tight Integration between Applications RACF

► Highest level of security RAS

► Best HW and SW RAS Shared Everything

► Requires careful testing Sysplex SW Pricing

► Generally MLC► Capacity based

Fast Context Switching Sophisticated HW Cache Hierarchy Incremental costs decrease with

increasing capacity

VMRM► Hard and soft caps► Utilization higher than distributed, lower than z/OS

Network Hop between Virtual Servers Linux Security

► Multiple options RAS

► zSeries HW, Linux SW State of the Art Virtualization

► Isolated yet shared Clustering done with SW SW Pricing

► OTC, distributed basis► Engine based

Fast Context Switching Sophisticated HW Cache Hierarchy Requires “Critical Mass” for cost savings

z/OS zLinux



What Makes Best Fit

Leverage classic strengths of the zSeries ► High availability► High I/O bandwidth capabilities ► Flexibility to run disparate workloads concurrently ► Requirement for excellent disaster recovery capabilities► Security

Shortening end-to-end path length for applications► Collocation of applications► Consolidation of applications from distributed servers► Reduction in network traffic► Simplification of support model



WebSphere MQSeries® DB2 Connect™ CICS® Transaction Gateway IMS Connect™ for Java™ Web Logic/WebSphere Application Servers and Java applications for

production and development Applications requiring top end disaster recovery model ComServer and Communications Controller for Linux LDAP security services IBI Web Focus

Best Fit



Evaluate server choices ► Correct application availability,

► Supporting applications,

► Total Cost of Ownership (TCO)

► Politics within the organization Issues that can influence architecture decision

► Shortening end-to-end path length for applications

► Collocation of applications

► Consolidation of applications from distributed servers

► Reduction in network traffic

► Simplification of support model

What Makes a Good Fit



SAP Application Servers DB2 UDB Oracle Database Apache Web serving SAMBA Network Infrastructure, FTP, NFS, DNS etc.., e-Mail solutions

Good Fit



Summary

If Linux makes sense for the Enterprise, then...

Consider Linux on System Z for that subset where

It is cheaper

Or where there are other special considerations and benefits

Linux_on_Z_Uniforum.ppt

Documents

hardware products

ibm system z operating

linux code

ibm eserver ibm logo

system z tivoli

ibm representative

unique version of linux

hardware scalability