Application Performance Management

1
Application Performance Management

• Klaus Fellner, Christian Plonka

2
Goals of this Session

• Understand general industry information and
  trends
• Learn SilkCentral Performance Managers
  capabilities and positioning
• Be able to articulate why SilkCentral Performance
  Manager is a superior solution

3
Industry Information and Trends

• Application Performance Management

4
The IT Quality Tools Market Quadrants
Pre-Production
Production
Application Performance Management (APM)
AutomatedSoftware Quality(ASQ)
Clientlevel
Application
Codelevel
Network Systems Management(NSM)
Device Quality
Network System Level
5
The Many Objectives of APM

• Health assessment about the application
• Application performance monitoring from the
  end-user perspective
• Service-level monitoring from the end-user
  perspective
• Error detection
• Root cause analysis of errors and performance
  problems that occur to the end-user
• Capacity planning
• Application error prediction and prevention

6
Reuse Of RD Test Assets Enables Efficient
Application Performance Management
IT aloneis not able to effectively do APM,
because they lackthe requiredapplication
knowledge

• Pre-production
• Unit tests
• Functional tests
• Load Tests

• Production
• Active monitoring
• Passive monitoring

7
Convergence of APM NSM
APMNSM will converge within 18 months 451Group

• What is Segues response?
• Need for reuse of test assets from RD/QA for APM
• Need for passive monitoring (also pointed out by
  451Group)
• Need for doing APM beyond Web applications
• Integration with the big NSM vendors

8
SilkCentralPerformance ManagerApplication
performance management for mission-critical
applications from theend-user perspective
9
APM With SC Performance Manager
Target Application
Real Users
SilkCentral Performance Manager
10
What Are the Benefits of This Architecture?

• Advantages
• Accurate assessment of application performance
  and service level status based on true end-user
  experience from various locations
• Support for monitoring service levels from the
  end-user persepective
• Reuse of application knowledge of RD/QA for
  efficient APM
• Enable application and IT engineers to work
  together to solve performance problems that hit
  the end-user
• Run many business transactions on one or more
  agent machines
• Propagate events to NSM systems and client-side
  health values to CA Unicenter
• Support for 24 x 7 x 365 operation
• Benefits
• Reduce risk of missed application failures or
  service-level violations
• Reduce resource demand (staff equipment)
• Reduce error turnaround time
• Leverage existing IT investments

11
Timed Savings and Reduced Risk Through Advanced
Automated Analytics

• Features
• Health status report
• SLA status report
• Health history report
• Health snapshot report
• Advantages
• Provides a quick overview of the current health
  and SLA status, as well as their trends per
  project, for application and IT executives
• Provides history data required for capacity
  planning as well as error prediction and
  prevention
• Benefits
• Save time otherwise spent performing manual
  analysis
• Reduce risk of unforeseen application outages or
  service-level violations
• Lower ongoing maintenance costs

12
Quickly Assess Current Health and SLA Status
Including Trends Per Project
Birds-eye view of application and system
status eases analysis
Monitor service levels of all business
transactions
Configurable to your needs
Predict and prevent problems
13
Quickly Assess if a Problem Exists Now and Where
it is Located
Shows currentapplication, location system
status
14
Reduce Error Turnaround Time Through Powerful
Diagnostic Tools

• Features
• Health drill-down
• Visual front-end diagnostics
• Server-side performance diagnostics
• Internet performance diagnostics
• Advantages
• Quickly classify problems
• Find out exactly what happened from the end-user
  perspective
• Easily nail down the root cause in the server or
  network tier
• Benefits
• Reduce error turnaround time

15
Quickly Classify Error Situations Using Health
Drill-down
Conclusion Not an Internet, but a back-end
problem
Problem detected
Problem classified as an accuracy error
Without accuracy monitoring, problem would
have been missed
Error was visible to all locations
16
TrueLog Shows what has Really Happened From the
End-User Perspective
Visual log of the complete user session
Exact application error message that the user
received
Enablesfurther analyis viathe server logs
17
Find the Root Cause in the Server Tier Using
Server-side Performance Diagnostics
Automatedroot-cause analysis through
statistical correlation
Detected accuracy problem
18
Find the Root Cause In The Network Tier Using
Internet Performance Diagnostics
ProblemPeak in application response time
Root causeOutage of local DNS server
19
Nail Down the Web Page Object Causing the
Performance Threshold Violation
Is there a performance problem with the HTML
document or with the embedded objects?
Shows objecttimings for anindividual call
Does caching work?
Is the object too large?
Which individual object took up the most time?
20
Support for All Major Enterprise Applications and
Technologies

• Active monitoring
• Supports the same applications that are supported
  by test engines
• SilkPerformer
• SilkTest
• Passive monitoring
• HTTP(S)
• Infrastructure monitoring
• Supports monitoring the same infrastructure
  components as SilkPerformer

Use a single toolfor APM of all your different
enterpriseapplications
21
FeaturesAdvantagesBenefits superior to the
competition
22
Why Is SC Performance Manager Better?

• More accurate measurements
• Superior front-end diagnostics
• Better integration with QA
• Easier export of results to custom 3rd party
  applications

23
Why Is SCPM More Accurate?

• Features
• Shows true end-user experience- including
  accuracy (functional integrity)
• Real-user error detection
• Advantages
• Detection of application failures in addition
  to system failures
• Immediately detect all errors that occur for
  real-users, which are not covered by active
  synthetic business transaction monitors
• Benefits
• More productive end-users
• Reduce risk of unforeseen application outages
• Reduce error turnaround time

24
Reliable Error Detection Through Monitoring the
Applications Functional Integrity
Graceful application error - invisble to
system-level diagnostics
Target Application Environment
Application error detected only trough monitoring
accuracy
25
Passive Monitoring Reliably Detects Errors
Drastically Reduces Error Turnaround Time
Deployed
Found
Error Occurs
Diagnosed
Fixed

• Existence unknown
• Awareness depends on user complaints

• Root-cause analysis
• Assign priority
• Involve development

• Engineer

• Validate

Eliminate the occurrence/ awareness delay
Passive Monitoring
Error turn-around time
Quicker diagnosis through advanced root cause
analysis
Quicker validation through closed loop automated
testing
26
SCPM Provides Superior Front-end Diagnostics

• Features
• TrueLog On Error
• Advantages
• Illustration of user click-path unveils error
  message of the application
• Web page breakdown points out the objects and
  their network timings that have caused the
  performance threshold violation
• Benefits
• Reduce error turnaround time

27
Closed-Loop Testing Provides Better Integration
with QA
Closed-looptesting

• Advantages
• Support RD/QA with realistic test cases through
  recording erroneous real-user sessions
• Benefits
• Higher application quality after QA phase

28
Web Service Interface Provides Better Integration
with Custom Management Applications

• Generic Web service interface
• Access to the results repository
• Controls monitor executions
• Advantages
• Get access to application performance data from
  the end-user perspective from custom 3rd party
  applications
• Benefits
• Leverage existing IT investments in custom
  management applications

29
Competitive Tools

• Mercury Availability Center (previously Topaz)
• Compuware Vantage
• Empirix OneSight/FarSight
• Keynote (service only)
• Gomez (service only)

30
ROI Framework

• A look at what is under construction.

31
SilkCentral Performance Manager - Connecting
Features to Customer Business Objectives
Reduce Operational Cost
Improve Productivity
Meet Market Requirements
OBJECTIVES
2. Lower ongoing maintenance costs
4. Reduce defect repair cost
3. Efficient use of available resources
1. Reduce risk of missed or unforeseen
application problems
5. Save license costs for quality tools
STRATEGY
1.1 Monitor all mission critical business
transactions for your application in
production -Test assets reuse from
pre-production testing, in particular
SilkPerformer and SilkTest (business transaction
scripts, infrastructure monitors, application
performance knowledge)
3.4 Optimal use of monitoring infrastructure -Rem
ote POP technology -Scalable architecture where
one POP can contain multiple execution servers,
and one execution server can run many monitors
1.4 Predict and prevent error situations -Health
history reports -Health and SLA status trends
3.1 Reuse of application knowledge of RD/QA for
efficient APM -Test assets reuse from
pre-production load testing (business transaction
scripts, infrastructure monitors, application
performance knowledge)
4.2 Identify and repair defects faster -Passive
Monitoring eliminates the error
occurrence./awareness delay Powerful diagnostic
tools -Health-drill down -TrueLog On
Error -Server-side performance diagnostics -Intern
et performance diagnostics (Client-side network
time breakdown per Web page, Client-side network
time breakdown per Web component, TrueRoute)
1.5 Monitor your applications 24x7x365 -Automatic
failover
TACTICS Product Feature Services
3.2 Reuse results from monitoring the production
environment for pre-production load
testing -Closed-loop testing via passive
monitoring
3.5/5.1 Standardize on a single tool for all your
applications -Support for all major enterprise
applications and technologies
1.2 Accurate assessment of application
performance based on true end-user experience -3
dimensional health monitors including accuracy
(functional integrity) -Real-user error detection
through passive monitoring
2.1 Optimal capacity planning (hardware
resources) -Health history reports -Health and
SLA status trends -Server-side performance
diagnostic tools
4.1 Improve collaboration between RD and IT to
solve performance problems that hit the
end-user -Standard HTML Web-GUI, which
simplifies access to reports and diagnostics
3.3 Leverage existing IT investments Open
Interfaces -Generic Web service
interface -Generic SNMP interface -Specific
integration with CA Unicenter
2.2/3.2 Save time of manual analysis (human
resources) Advanced automated analytics -Health
status report -SLA status report -Health snapshot
report
1.3 Monitoring service levels from the end-user
perspective -Active business transaction
monitoring
NOTE Underlined features are competitive
advantages compared to Mercury Availability
Center (Topaz)
METRICS next step
32
Thank You!