Performance Analysis and Tuning for HPE NonStop Systems

Become familiar with the system load balance approach and performance tuning concepts for your HPE NonStop system. Learn how to capture and analyze performance data, then adapt the adjustments to maximize performance and increase system utilization. Topics include Measure, Measure entities, queuing theory, TPM, RPM, ViewSys, Web ViewPoint practical approaches to system tuning, and using performance tools. The course is 60 percent lecture and 40 percent hands-on labs using NonStop servers.

After you successfully complete this course, expect to be able to:

• Apply the basic tuning principle for NonStop 
• Analyze Measure’s key entities and use Reload Analyzer
• Apply several basic queuing theory formulas
• Analyze disk cache-hit ratios and fragmentation
• Calculate a volume’s true write cache-hit ratio
• Analyze process priorities and process memory consumption
• Identify processes with long $RECEIVE queues
• Evaluate TCP process and server class parameters for best performance
• Use Measure, SQLCI, and MXCI to analyze NonStop SQL/MP and NonStop  SQL/MX performance
• Identify positive and negative factors in application performance 

• System managers, technical support, and performance analysis personnel
• Systems and application designers and developers

• Concepts and Facilities for HPE NonStop Systems (U4147S) and
• NonStop NB-series Server Administration I (HG776S) and NonStop NB-series Server Administration II (HG777S) or
• Equivalent system administration courses or experience 

Performance Analysis Introduction

• Steps in tuning a NonStop system and basic tuning principle
• Two aspects of response times
• Service Level Agreement (SLA) options
• Understanding your system(s)
• Measure subsystem and its entities 

CPU Subsystem

• Measure’s CPU entity
• Processor queueing and its relationship to queueing theory
• Processor’s memory subsystem
• Physical and logical disk loss and CISC and EPIC processes
• Measure’s processor data in a matrix
• NSMA architecture shown in Measure 

Queueing Theory

• Queueing theory notation and basic queueing theory formulas
• Slow and fast device queueing 

Disk Subsystem

• Integrity NonStop NS-series disk architecture
• NB-series systems storage CLIMs
• DISC and SERVERNET entities and Measure’s disk data in a matrix
• Disk cache-hit ratios and buffering non-audited files
• Disk settings with SCF and disk fragmentation 

Disk Files

• Key-sequenced file structure and buffered/unbuffered files
• Calculating a volume’s true write cache-hit ratio
• Locating heavily-written and unbuffered files
• File partitioning, file fragmentation, and Reload Analyzer 

Processes

• TACL RUN command options and analyzing process priorities
• Mixed Workload Enhancement (MWE)
• Key Measure PROCESS counters
• Analyzing process memory consumption
• Processes with long $RECEIVE queues and process acceleration
• Measure’s PROCESS data in a matrix

Pathway Tuning

• TCP process parameters for best performance
• Server class parameters for best performance
• Pathway statistics and establishing links to server processes
• Application Cluster Services (ACS) features 

HPE SIM and Other Performance Tools

• HPE Systems Insight Manager (SIM) performance essentials
• ViewSys, Web ViewPoint, Peek, and Enform
• Guardian Performance Analyzer (GPA)
• Tandem Performance Data Collector (TPDC)
• Tandem Performance Management (TPM)
• Availability Stats and Performance (ASAP)
• Automatic Process Balancer (APB)
• Disk Prospector (Diskpro) and Real-time Process Monitor (RPM)
• Tandem Capacity Model (TCM)

Application Tuning

• Application performance—positive and negative factors
• Meascom output for DISCOPEN and FILE entities
• Meascom output for PROCESSH and USERDEF entities
• Meascom output for TMF entity
• Remote Database Facility (RDF) performance issues