Supplementary Class Materials

Copyright © Performance Dynamics Company. All Rights Reserved.

1 Guerrilla Aphorisms
2 GDAT Reading List
3 USL Analyzer
4 PDQ Analyzer
5 Performance Management Standards
6 Modeling Internet/Web Traffic
7 Statistical Methods for Capacity Planning
8 Virtualization
9 Professional Organizations
10 Benchmarks
11 Performance Jobs

1 Guerrilla Aphorisms

This is the extended online version of the original Guerrilla Manual that was included with the first print run of the GCAP book.

2 GDAT Reading List

Although there is no separate Guerilla book for the GDAT course, this reading list includes books that are referred to during the class.

3 USL Analyzer

See How to Quantify Scalability with the Universal Scalability Law (USL) for a more complete summary.

Example EXCEL spreadsheet: USLcalc.xls
Example EXCEL solver for 3-parameter USL model: usl3-solver.xls
Example R script for 3-parameter US model: USL3-analyzer.r

R package USL by Stefan Möding

	install.packages("usl")
	vignette("usl")

	library(usl)
	data(specsdm91)
	usl.model <- usl(throughput ~ load, specsdm91)
	summary(usl.model)
	peak.scalability(usl.model)
	plot(specsdm91, pch=16)
	plot(usl.model, col="red", add=TRUE)

R package SATK by Paul Puglia
The Scaling Analysis Tool Kit (SATK) is an R package that contains super-classes of
R's standard nls class specialized for fitting parameteric scaling models like
Amdahl's Law and Gunther's Universal Scaling Law.
```
	install.packages("SATK", repos="http://R-Forge.R-project.org")

	library(SATK)
	data(USLcalc)
	uslcalc.zones <- zones(USLcalc)
	plot(uslcalc.zones)   
	
```

4 PDQ Analyzer

DownloadPDQ (Pretty Damn Quick).
Modeling workload priorities with a "shadow" server.
1. PDQ model with virtual server subroutine.
2. PDQ Report1 without priority.
3. PDQ Report2 with priorities.

5 Performance Management Standards

ARM (Application Response Measurement)
InfiniBand interconnect architecture for servers and storage.
SNMP (Simple Network Management Protocol)
SPC (Storage Performance Council)
UMA (Universal Measurement Architecture)
Implementing Performance Flight Recorders in a Distributed Computing Environment with A+UMA
by Neil Gunther and Leon Traister, IEEE TCOS, (7)3, 1995. (Overview of the UMA implementation)

6 Modeling Internet/Web Traffic

The limitations of Poissonian assumptions for modeling Internet traffic are discussed in:

On the Self-Similar Nature of Ethernet Traffic (Original Bellcore draft 1993)
A Storage Model with Self-Similar Input (Associated router buffer-sizing paper 1994)
Wide-Area Traffic: The Failure of Poisson Modeling (LBL paper 1995)
Explaining World Wide Web Traffic Self-Similarity (Boston Univ. 1997)
Traffic Generation Concepts (Notes by Guerrilla alumus Jim Brady, 2004)
Open Versus Closed: A Cautionary Tale (Usenix NSDI conference 2006)
Load Testing Large User Population Web Applications (CMG paper by Guerrilla alumus Jim Brady, 2012)

7 Statistical Methods for Capacity Planning

What Do You Mean? Revisiting Statistics for Web Response Time Measurements (CMG paper by Guerrilla alumus David Ciemiewicz, 2001)
Making Statistical Sense out of Time Series Data with Home Grown Perl Scripts (CMG MeasureIT article by Guerrilla alumus Jim Brady, 2010)
Transforming Time Series Data into Capacity Planning Information (CMG paper by Guerrilla alumus Jim Brady, 2013)
Harmonic Averaging of Time Series Rate Data (Monitorama slides by NJG, 2013)

8 Virtualization

Virtualization and CPU Wait Times in a Linux Guest Environment (by Guerrilla alumnus Jim Brady, 2005)