Capacity Planning Free Solution

The Performance People
Performance Management with
Free and Bundled Tools
Adrian Cockcroft
Netflix Inc.
acockcroft@netflix.com
(Co-authored with Mario Jauvin
MFJ Associates
mario@mfjassociates.net)
14 May 2014

May 14, 2014 Adrian Cockcroft and Mario Jauvin
Agenda
 Overview of Capacity Planning Requirements
and Data Sources
 Performance Data Collection
 Free Network Monitoring Tools
 Free System Monitoring Tools
 Free Load Generation and Modelling Tools
 Licences and References

What are we talking about?
Network
monitoring with
WireShark, MRTG,
BigSister, Cacti,
Nagios, OpenNMS,
Zenoss, Openxtra,
ntop
Database Tier monitoring
With SEtoolkit, Orca,
XEtoolkit
Application Tier
monitoring with Orca,
Cacti, BigSister, Ganglia,
XEtoolkit
QA Load generation with
Grinder or SLAMD,
modelling with PDQ and R

Capacity Planning
Requirements and Data
Sources

Definitions
 Capacity
– Resource utilization and headroom
 Planning
– Predicting future needs by analyzing historical data
and modeling future scenarios
 Performance Monitoring
– Collecting and reporting on performance data
 Free Tools
– Bundled with the OS or available for no $$$

Capacity Planning Requirements
 We care about CPU, Memory, Network and Disk
resources, and Application response times
 We need to know how much of each resource we
are using now, and will use in the future
 We need to know how much headroom we have to
handle higher loads
 We want to understand how headroom varies, and
how it relates to application response times and
throughput

CPU Capacity Measurements
 CPU Capacity is defined by CPU type and
clock rate, or a benchmark rating like
SPECrateInt2000
 CPU utilization is defined as busy time
divided by elapsed time for each CPU
 CPU load average measures the average
number of jobs running and ready to run

Memory Capacity Measurements
 Physical Memory Capacity Utilization and Limits
– Kernel memory
– Shared Memory segment
– Executable code, stack and heap
– File system cache usage
– Unused free memory
 Virtual Memory Capacity - Swap Space
 Memory Throughput
– Page in and page out rates

Network Capacity Measurements
 Network Interface Throughput
– Byte and packet rates input and output
 TCP Protocol Specific Throughput
– TCP connection count and connection rates
– TCP byte rates input and output
 NFS/SMB Protocol Specific Throughput
– Byte rates read and write
– NFS/SMB service response times
 HTTP Protocol Specific Throughput
– HTTP operation rates
– Get and put payload byte rates and size distribution

Disk Capacity Measurements
 Detailed metrics vary by platform
 Easy for the simple disk cases
 Hard for cached RAID subsystems
 Almost Impossible for shared disk
subsystems and SANs
– Another system or volume can be sharing a
backend spindle, when it gets busy your own
volume can saturate, even though you did not
change your own workload

Capacity Planning Challenges
 Constantly changing infrastructure
 Limited attention span from staff
 Horizontally scaled commodity systems
 Per node software licencing costs too much
 Too many tools, too many agents per node
 Too much data, not enough analysis
 Non-linear and non-intuitive scalability
 Lack of tools and metrics for virtualized resources

Observability
 Four different viewpoints
– Management
– Engineering
– QA Testing
– Operations
 Each needs very different information
 Ideal would be different views of the same
performance database
 Reality is a mess of disjoint tools

Management Viewpoint
 Daily summary of status and problems
 Business oriented metrics
 Future scenario planning
 Marketing and management input
 Concise report with dashboard style status
indicators
 Free tools: R, Spreadsheet and Web based
displays, no good summarization tools

Engineering Viewpoint
 Large volumes of detailed data at several different
time scales
 Input to tuning, reconfiguring and future product
development
 Low level problem diagnosis
 Detailed reports with drill down and correlation
analysis
 Free tools: XE/SE Toolkit, Orca, Ganglia, Cacti, R

QA Test Viewpoint
 Workload specification tools
 Load generation frameworks
 Testing for functionality and performance
 Regression tools to compare releases
 Modelling difference between test configuration
and production configuration
 Free Tools: The Grinder, SLAMD, R, PDQ

Operations Viewpoint
 Immediate timeframe
 Real time display, updated in seconds
 Alert based monitoring
 High level problem diagnosis
 Simple high level graphs and views
 Free tools: BigSister, Nagios, OpenNMS,
MRTG, Cacti, Ganglia, WireShark, ntop

Measurement Data Interfaces
 Several generic raw access methods
– Read the kernel directly (not a good idea)
– Structured system data (Solaris kstat, Linux /proc)
– Process data
– Network data
– Accounting data
– Application data
 Command based data interfaces
– Scrape data from vmstat, iostat, netstat, sar, ps
– Higher overhead, lower resolution, missing metrics
 Data available is platform specific either way
 Much more detail on this topic in the Solaris/Linux Performance
Measurement and Tuning Class

Free Network
Monitoring Tools

SNMP
 Simple network management protocol
 UDP protocol based on port 161
 Client/server like
– Client is called management application entity
– Server is called an agent entity
 Agent entity is designed to be implemented
on network hardware, router, switches, etc

SNMP – MIBs
 Management information base
 Defines the structure and the semantic of the
information that can be reported on
 Most commonly used is MIB-II which defines a set of
standard networking attributes
– Interface tables
– System level information
– Routing tables
 Specified using ASN.1 (abstract syntax notation 1)

SNMP – commands
 Called PDU (protocol data units)
 GET
 GETNEXT
 GETBULK
 SET
 Encoded using BER (basic encoding rules)

Versions
 Version 1, original version done in May 1991
 Version 2, around 1993. Failed because the
IETF credo of “rough consensus and running
code” could not be met on securing SNMP
 Turned into V2c for community string security
(like V1)
 Version 3, added security and complexity in
1998

SNMP tools
 Too numerous to name all but…
 OpenNMS
 Nagios
 Cacti
 MRTG
 Net-snmp
– See www.snmplink.org

SNMP tools
 Snmpwalk – will report all data in a specified
MIB
 getIf – will report data about interfaces and
includes built-in MIB browser
 Snmptable – will report tabular data from MIB
tables

OpenNMS
 Well…. it’s not that portable
– 95% java is not 100% java
– Requires about 20-30 different platform specific
packages (PostgreSQL, Perl, RRD tool, Tomcat 4
etc…)
– Difficult to install
– Easy auto discovery
– Web-based interface

OpenNMS
 Main screen shot

OpenNMS
 Node screen shot

Nagios
 Easy to build/compile (on Solaris 10)
 Easy to install
 Quick response from CGI
 Configuration is manual and a pain
– 13 configuration files with all kinds of interrelated
entries
– Tedious and error prone
 Requires plugins to do anything

Nagios
 Main screen shot

Nagios
 Host detail screen shot

ntop
 Similar to familiar UNIX top tool for
processes but used for network
 Provide huge selection of real-time data
 Can be found at http://www.openxtra.co.uk/

ntop – Active Sessions

ntop Hosts

ntop Network Load

ntop_Network_Thruput

ntop Port Dist

ntop_Protocol_Dist

ntop Protocols

Zenoss
 Open source monitoring and management of
IT infrastructure
 Zenoss core is free
 Other editions are for a fee
 Get it from http://www.zenoss.com/download/

zenoss Architecture

zenoss Dash Config

zenoss Google

zenoss Google Alerts

Zenoss Graphs

zenoss Topology

MRTG
 Really simple to install and configure
 Require manual config file creation
 Only for MIB-II interface plotting out of the
box
 Graphing not flexible, axis, time etc

MRTG
 Interface screen shot

MRTG
 Other CPU screen shot

RRD tool
 Software to store, retrieve and graph
numerical time series data
 Use a round robin algorithm
 Data files are a fixed size
– Don’t grow
– Don’t require maintenance

RRD tool
 Compiles on most platforms
 Used by many SNMP based tools
– OpenNMS
– Cacti
– BigSister
– WeatherMap4RRD
– MailGraph

RRD tool
 14all CGI script that plots data similar to
MRTG
 Configurable to collect data at different
interval (unlike MRTG)
 Flexible and variable in what data can be
collected

RRD tool
 Sample screen shot

RRD tool
 Screen shot

RRD tool
 Create a RRD database
rrdtool create test.rrd
--start 920804400
DS:speed:COUNTER:600:U:U
RRA:AVERAGE:0.5:1:24
RRA:AVERAGE:0.5:6:10

RRD tool
 Create a graph
rrdtool graph speed.png
--start 920804400 --end 920808000
DEF:myspeed=test.rrd:speed:AVERAGE
LINE2:myspeed#FF0000

Free Performance Data
Collection and Rules
Toolkits

SE toolkit Example Tools
 A free performance toolkit for rapidly creating custom data sources
 Makes all the very extensive Solaris metrics easily available
 Very system specific and not enough metrics exist to port to Linux
 Written by Rich Pettit with contributions from Adrian Cockcroft
 Get SE3.4 from http://sourceforge.net/projects/setoolkit/
 Open source with support for SPARC & x86 Solaris 8, 9, 10
Function Example SE Programs
Rule Monitors cpg.se monlog.se mon_cm.se live_test.se percollator.se
zoom.se virtual_adrian.se virtual_adrian_lite.se
Disk Monitors siostat.se xio.se xiostat.se iomonitor.se iost.se xit.se disks.se
CPU Monitors cpu_meter.se vmmonitor.se mpvmstat.se
Process Monitors msacct.se pea.se ps-ax.se ps-p.se pwatch.se pw.se
Network Monitors net.se tcp_monitor.se netmonitor.se netstatx.se nfsmonitor.se nx.se
Clones iostat.se uname.se vmstat.se nfsstat-m.se perfmeter.se xload.se
Data browsers aw.se infotool.se multi_meter.se
Contributed Code anasa dfstats kview systune watch orcollator.se
Test Programs syslog.se cpus.se pure_test.se collisions.se uptime.se dumpkstats.se
net_example nproc.se kvmname.se

SE language features
 SE is a 64bit interpreted dialect of C
– Not a new language to learn from scratch!
– Standard C /usr/ccs/bin/cpp used at runtime to preprocess SE scripts
– Main omissions - pointer types and goto
– Main additions - classes and “string” type
– powerful ways to handle dynamically allocated data
– built-in fast balanced tree routines for storing key indexed data
 Dynamic linking to all existing C libraries
– Built-in classes access kernel data
– Supplied class code hides details, provides the data you want
 Example scripts improve on basic utilities e.g. siostat.se, nx.se, pea.se
 Example rule based monitors e.g. virtual_adrian.se, orcallator.se

Creating Rules
 Based on real experiences of all the things that go
wrong
 Capture an approximation to intuition
 Test and calibrate rules on as many systems as
possible
 Easy??

Configuring Rules
 Thresholds should be configured
 Very application dependent
 Capture the operating envelope
– Measure the underlying values
– Measure peaks in normal operation
– Note values during problems
– Set thresholds to capture the difference
 This applies to any tool
– SE Toolkit, Cacti, Ganglia, Nagios, OpenNMS etc.

Rules as Objects
 Define only the input and output information
 Hide implementation details
 Make high level rule objects trivial to use and
reuse
 SE Toolkit does it in three lines of code:
– #include <rules file>
– Declare rule object as a typed variable
– Read and use or print object status

"virtual adrian" rules summary
 Disk Rule for all disks at once
– Looks for slow disks and unbalanced usage
 Network Rule for all networks at once
– Looks for slow nets and unbalanced usage
 Swap Rule - Looks for lack of available swap space
 RAM Rule - Looks for short page residence times
 CPU Power Rule
– Scales on MP systems
– Looks for long run queue delays
 Mutex Rule - Looks for kernel lock contention and high sys CPU time
 TCP Rule
– Looks for listen queue problems
– Reports on connection attempt failures

XE Toolkit - www.xetoolkit.com
 Complete re-write of SE Toolkit by Rich Pettit
– Extensible Java collector, customize with jar files
– Release 1.2 available April 2008
– Multi-platform support Solaris, Linux/x86, Windows, BSD,
OSX, HP-UX, AIX, Linux/s390, Linux/Power
 Licencing
– Free GPL version for standard use and shared derivations
– Open source, hosted at http://sourceforge.net/projects/xe-toolkit/
– Commercial support available if needed
– Commercial product license for custom in-house derivations
 Addresses all the issues people had with SE toolkit !

Captive Metrics / XE Toolkit
Architecture

Free System Monitoring
Tools

Collated Performance Data - Orca
 Problems with time sync when collecting data from multiple tools
– No timestamp at all for vmstat, netstat, df...
– No timestamp by default for iostat and ps...
– No way to collect realtime stats from an http logfile
 Use SE Toolkit to generate one timestamped row containing all the data
– First version of percollator.se written by Adrian Cockcroft in 1996
– Extended orcallator.se written by Blair Zajac a few years later
– Graphs generated by orca batch job feeding rrdtool based web pages
– Active community developing tool at http://www.orcaware.com
– Extended to collect much more data, including process workloads
– Basic data collection ported to Linux, HP-UX and Windows
 Orca is basically MRTG for System metrics rather than Network
 See http://www.orcaware.com/orca/docs/Orca_Understanding_Performance_Data.ppt

Orca data collections
 Collected using “procollator” reading info from /proc on Linux
[Uptime] [Average # Processes in Run Queue (Load Average)] [CPU Usage]
[New Process Spawn Rate] [Number of System & Running Processes]
[Context Switches & Interrupts Rate] [Interface Input Bits Per Second]
[Interface Output Bits Per Second] [Interface Input Packets Per Second]
[Interface Output Packets Per Second] [Interface Input Errors Per Second]
[Interface Output Errors Per Second] [Interface Input Dropped Per Second]
[Interface Output Dropped Per Second] [Interface Output Collisions]
[Interface Output Carrier Losses] [TCP Current Connections] [IP Statistics]
[TCP Statistics] [ICMP Statistics] [UDP Statistics]
[Disk System Wide Reads/Writes Per Second] [Disk System Wide Transfer Rate]
[Disk Reads/Writes Per Second] [Disk Transfer Rate] [Disk Space Percent
Usage] [Physical Memory Usage] [Swap Usage] [Page Ins & Outs Rate]
[Swap Ins & Outs Rate]
 Orca on Solaris collects many more metrics than shown above
 Strength of Orca is lots of detailed metrics with low overhead for collection
 Easily customized to add more system metrics or application metrics
 Orca can already track HTTP traffic and parse log files

All metrics are stored in
“round robin database” format
using RRDtool to generate
displays over different time
spans
Web page is simple collection
of plots with drill down by
metric or by time
Suitable for monitoring a
relatively small number of
systems in great detail, e.g.
backend database servers

Cacti – www.cacti.net
 Web based user interface based on RRDtool
 More sophisticated GUI than Orca or MRTG
 Less sophisticated system metric collection,
but more coverage of networking
 Better management of groups of systems
and devices than Orca, useful for tens to
hundreds of nodes
 Access control and personalization for users

Ganglia – www.ganglia.info
 Web based RRDtool GUI somewhat similar to Cacti
 Better management of clusters of systems and
devices than Cacti, useful for hundreds to thousands
of nodes in a hierarchy of clusters
 Provides many summary statistic plots at cluster
level and collects detailed configuration data
 XML based data representation
 Uses low overhead network protocol
 In common use at hundreds of large HPC Grid sites,
less visibly in use at some large commercial sites

BigBrother and BigSister
 Network and system dashboard alert monitor
 Widely used at internet sites
 Bigbrother is at http://www.bb4.com
 Bigsister is at http://bigsister.graeff.com
 Bigsister seems to have more features, alert
logging, better portability and more efficient
data collection. Compatible update to BB4.

Free QA Test and
Modelling Tools

QA Test Requirements
 Generate test workload
– SLAMD, Grinder
 Collect performance metrics
– Any of the tools already mentioned
 Report regression against baseline
 Predict capacity needed for production system
– Use spreadsheets for simple linear prediction
– Use modelling tools such as PDQ for queuing models

Grinder 3 - Powerful New Features
 100% Pure Java - works on any hardware platform and any
operating system that supports J2SE 1.3 and above.
 Java and Jython based load testing framework
– Web Browsers: simulate web browsers using HTTP, and HTTPS.
– Web Services: test interfaces using SOAP and XML-RPC.
– Database: test databases using JDBC.
– Middleware: RPC and MOM based systems using IIOP, RMI/IIOP,
RMI/JRMP, and JMS.
– Other Internet protocols: POP3, SMTP, FTP, and LDAP.
 See http://grinder.sourceforge.net/g3/features.html
 J2EE Performance Testing with BEA WebLogic Server by Peter
Zadrozny, Philip Aston and Ted Osborne, originally published
by Expert Press and now by APress uses Grinder 2 throughout.

SLAMD
 Load generation framework, written in Java
 Originally built to test LDAP servers by Sun
 Extended to be very generic and published
as open source. Actively being developed.
 Sophisticated functions and user interface
 See http://www.slamd.com
 Latest Release 2.0 has better usability focus

PDQ Modelling Tool
 Dr Neil Gunther’s toolkit at
http://www.perfdynamics.com
 Library used from C or Perl provides MVA queueing
models
 Use to calibrate in QA and predict in production
 PDQ modelling tool details:
– The Practical Performance Analyst Dr. Neil Gunther -
McGraw-Hill, 1998 ISBN 0-07-912946-3
– Analyzing Computer System Performance with Perl:PDQ
2004, ISBN 3-54-020865-8

References and
Conclusion

Licences for Free Tools
 Open Source Initiative
– “OSI Approved licences”
– http://opensource.org/licenses/category
 Comparisons of Common Licences
– http://zooko.com/license_quick_ref.html

Web Pages and Books
 Adrian’s Performance and other topics blog
– http://perfcap.blogspot.com
 MFJ Associates performance tools link page
– http://www.mfjassociates.net/perf_links.html
 More free tools compiled by John Sellens
– http://www.generalconcepts.com/resources/monitoring/
 More tools compiled by Openxtra
– http://www.openxtra.co.uk/resource-center/open_source_network_monitor_tools.php
 SE toolkit info: Sun Performance and Tuning - Java and the Internet - Adrian
Cockcroft and Richard Pettit - Sun Press/Prentice Hall, 2nd
Edition, 1998 ISBN 0-13-
095249-4
 Solaris 8 and Linux: System Performance Tuning 2nd
Edition – Gian-Paolo Musumeci,
O’Reilly 2002 ISBN: 0-596-00284-X
 Solaris Internals http://www.solarisinternals.com
– Richard McDougall and James Mauro - new 2nd edition and new performance book by
Richard McDougall and Brendan Gregg

Concluding Remarks
 Many large installations depend on free tools
 A full suite of functionality is available
 Several tools are needed to cover the bases
 Tradeoff between function and ease of use
 Support may be available, but typically
Google is the best support tool
 Functionality is increasing….

Questions?
acockcroft@netflix.com
mario@mfjassociates.net

Capacity Planning Free Solution

More Related Content

Similar to Capacity Planning Free Solution (20)

More from luanrjesus (7)

Recently uploaded (20)

Capacity Planning Free Solution

Editor's Notes