IT Infrastructure Last Lecture(s)

• Performance Concepts
• Introduction
• Perceived Performance
• Performance During Infrastructure Design
– Benchmarking
Lecture # 8 – Vendor Experience
– User Profiling Infrastructure Load

• Quiz # 1
Dr. Muhammad Aamir Khan
Assistant Professor
Department of Informatics and Systems
School of Systems and Technology (SST)
University of Management and Technology

SST UMT Lahore IT Infrastructure ‐ Lecture 8 1 SST UMT Lahore IT Infrastructure ‐ Lecture 8 2

Outlines Reading Assignment # 2

• Performance of a Running System • Read Chapter # 4 of the book “IT Infrastructure Architecture:
– Managing Bottlenecks Infrastructure building blocks and concepts” by Sjaak Laan.
– Performance Testing
• Performance Testing – Breakpoint
• Performance Patterns
– Increasing Performance on Upper Layers
– Caching
• Disk Caching

SST UMT Lahore IT Infrastructure ‐ Lecture 8 3 SST UMT Lahore IT Infrastructure ‐ Lecture 8 4

Managing Bottlenecks
• The performance of a system is based on:
– The performance of all its components

Performance of a Running •
– The interoperability of various components
For instance, building an infrastructure with really fast networking
components has little benefits when the used hard disks are slow.
System • A component causing the system to reach some limit is referred to as the
bottleneck of the system because the performance or capacity of the entire
system is limited by a single component, slowing down the system as a
whole.
• Every system has at least one bottleneck that limits its performance.
• Knowing where in the system the bottleneck occurs can improve performance
by removing that bottleneck. However, usually another bottleneck arises and
there will always be a bottleneck somewhere in the system.
• If the bottleneck does not negatively influence performance of the complete
system under the highest expected load, it is OK

SST UMT Lahore IT Infrastructure ‐ Lecture 8 5 SST UMT Lahore IT Infrastructure ‐ Lecture 8 6

School of Systems and Technology

IT Infrastructure ‐ Lecture 8 1
 Performance Testing Performance Testing ‐ Breakpoint
• Benchmarking is a way to measure individual components, while system • Ramp up the load
performance tests measure the system as a whole. There are three major – Start with a small number of virtual users
types of performance tests for testing complete systems: – Increase the number over a period of time
• The test result shows how the performance varies with the load, given as
• Load testing ‐ shows how a system performs under the expected load number of users versus response time.
– It is a check to see if the system performs well under normal
circumstances.
• Stress testing ‐ shows how a system reacts when it is under extreme load
– Goal is to see at what point the system "breaks" (the breakpoint) and
where it breaks (the bottleneck).
• Endurance testing ‐ shows how a system behaves when it is used at the
expected load for a long period of time
– Typical issues that arise are memory leaks, expanding database tables,
or filling disks, leading to performance degradation.
SST UMT Lahore IT Infrastructure ‐ Lecture 8 7 SST UMT Lahore IT Infrastructure ‐ Lecture 8 8

Performance Testing Performance Testing

• Performance testing software typically uses: • Performance testing should be done in a production‐like environment
– One or more servers to act as injectors – Performance tests in a development environment usually lead to
• Each emulating a number of users results that are highly unreliable
• Each running a sequence of interactions – Even when underpowered test systems perform well enough to get
– Recorded as a script, or as a series of scripts to emulate good test results, the faster production system could show
different types of user interaction. performance issues that did not occur in the tests
– A test conductor (separate server) • To reduce cost:
• Coordinating tasks – Use a temporary (hired) test environment
• Gathering metrics from each of the injectors
• Collecting performance data for reporting purposes

SST UMT Lahore IT Infrastructure ‐ Lecture 8 9 SST UMT Lahore IT Infrastructure ‐ Lecture 8 10

Increasing Performance on Upper

Layers
• 80% of the performance issues are due to badly behaving applications
• Application performance can benefit from:
– Database and application tuning
Performance Patterns – Prioritizing tasks
– Working from memory as much as possible (as opposed to working
with data on disk)
– Making good use of queues and schedulers
• Typically more effective than adding compute power

SST UMT Lahore IT Infrastructure ‐ Lecture 8 11 SST UMT Lahore IT Infrastructure ‐ Lecture 8 12

School of Systems and Technology

IT Infrastructure ‐ Lecture 8 2
 Disk Caching Caching
• Disks are mechanical devices that are slow by nature
Time it takes to fetch 1 MB of
• Caching can be implemented in: Component
data (ms)
– Disks
Network, 1 Gbit/s 675
– Disk controllers
Hard disk, 15k rpm, 4 KB disk blocks 105
– Operating system
Main memory DDR3 RAM 0.2
• All non‐used memory in operating systems is used for disk cache
CPU L1 cache 0.016
• Over time, all memory gets filled with previously stored disk
requests and prefetched disk blocks, speeding up applications.
• Cache memory:
– Stores all data recently read from disk
– Stores some of the disk blocks following the recently read disk blocks

SST UMT Lahore IT Infrastructure ‐ Lecture 8 13 SST UMT Lahore IT Infrastructure ‐ Lecture 8 14

Next Lecture Questions?

• In the next lecture(s) we will start discussing about IT Infrastructure Non‐
Functional Attributes and Performance Concepts.

SST UMT Lahore IT Infrastructure ‐ Lecture 8 15 SST UMT Lahore IT Infrastructure ‐ Lecture 8 16

School of Systems and Technology

IT Infrastructure ‐ Lecture 8 3