LUKE DURKAN • GRADUATE DEVELOPER • ATLASSIAN
Knowledge is Power:
Visualising JIRA’s
Performance Data
At Atlassian we ask
the tough
questions…
Can we graph that?
A picture is worth a
1000 words
A picture is worth a
1000 words
A picture is worth a
1000 words
graph
A picture is worth a
1000 words
graph
A picture is worth a
1000 words
graph
numbers
WHAT YOU CAN SEE
WHAT IS PERF. MONITORING?
WHAT THE DATA CAN TELL YOU
GETTING THE BIG PICTURE
Agenda
ABOUT ME
About Me
Project AreasCurrent Role
About Me
- Log Analysis Tool
- Performance Monitoring
- Data Center
Developer on the JIRA
Software Enterprise
team
Why I’m
giving this
talk
Learnings From Project
First project I worked on at Atlassian
revolved around perf. monitoring
Work on Log Analyser
Driver for upgrades to the JIRA Log
Analyser tool
First Project
How useful perf. monitoring can be
for customers with the right strategies
Why I’m
giving this
talk
Learnings From Project
First project I worked on at Atlassian
revolved around perf. monitoring
Work on Log Analyser
Driver for upgrades to the JIRA Log
Analyser tool
First Project
How useful perf. monitoring can be
for customers with the right strategies
Why I’m
giving this
talk
Learnings From Project
First project I worked on at Atlassian
revolved around perf. monitoring
Work on Log Analyser
Driver for upgrades to the JIRA Log
Analyser tool
First Project
How useful perf. monitoring can be
for customers with the right strategies
Why I’m
giving this
talk
Learnings From Project
First project I worked on at Atlassian
revolved around perf. monitoring
Work on Log Analyser
Driver for upgrades to the JIRA Log
Analyser tool
First Project
How useful perf. monitoring can be
for customers with the right strategies
It is an immutable law in business that words
are words, explanations are explanations,
promises are promises-but only performance
is reality.
HAROLD GENEEN, FORMER PRESIDENT OF ITT
“
”
Why I’m
giving this
talk
Learnings From Project
First project I worked on at Atlassian
revolved around perf. monitoring
Work on Log Analyser
Driver for upgrades to the JIRA Log
Analyser tool
First Project
How useful perf. monitoring can be
for customers with the right strategies
What is Performance
Monitoring?
What is Performance
Monitoring?
What isn’t Performance
Monitoring?
5 Secrets to JIRA no has
told you about
5 Secrets to JIRA no has
told you about
7 Things they don’t want
you to know about JIRA
Performance
7 Things they don’t want
you to know about JIRA
Performance
Performance
Monitoring
Not Perf. Testing
This allows for quick reaction to
potential performance problems
Log File Analysis
Review logs around the time reported
problems happen to understand why
Real-Time Analysis
Performance testing is the use of
fake data to stress test applications
Performance
monitoring helps to
understand, why!
What you can see
Spoiler: You can see
a lot!
• Response Times
• Average Response Times
• Response Times by Category
• Total Requests
• Requests per min
• Concurrent Requests
• Requests per min by Category
• Percentile Response Times
• Peak Response Times
Some of the data we can see
• Response time by Node
• Average Response Time by Node
• Requests per Node
…
Average Response Time
Always keep track
of what’s being
measured
Request Count (req/min)
Average Response Time by Nodes (ms)
Write your ownNew RelicLog Analyser Tool
Examples of products you can use
What the data can tell you
Average Response Time (ms)
Request Count (req/min)
Request Count (req/min)
Average Response Time by Nodes (ms)
Average Response Time by Nodes (ms)
Getting the big picture
Each graph is one
piece of a Jigsaw
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Don’t AssumeMatch SymptomsFind a Start Point
General Strategies
Recent Events
If we’re investigating
problems matching the
symptoms to the graph
they appear on is really
helpful
Depending on the time
frame of your data,
there can be a lot to
look through
Don’t always assume
that there has to be a
problem even if it’s
been reported
Always be mindful of
things that have
happened recently, i.e.
plugin upgrades,
application upgrades
Key
Takeaways
Try Perf. Monitoring
Perf. Monitoring allows you to
understand why things happen
Have a Strategy
Analysing data without a strategy is
like looking for a needle in a haystack
Understand Why
go.atlassian.com/jiraloganalyzer to
use our log analysis tool for free.
Thank you!
LUKE DURKAN • GRADUATE DEVELOPER • ATLASSIAN

Knowledge is Power: Visualizing JIRA's Performance Data

  • 1.
    LUKE DURKAN •GRADUATE DEVELOPER • ATLASSIAN Knowledge is Power: Visualising JIRA’s Performance Data
  • 2.
    At Atlassian weask the tough questions…
  • 3.
  • 4.
    A picture isworth a 1000 words
  • 5.
    A picture isworth a 1000 words
  • 6.
    A picture isworth a 1000 words graph
  • 7.
    A picture isworth a 1000 words graph
  • 8.
    A picture isworth a 1000 words graph numbers
  • 9.
    WHAT YOU CANSEE WHAT IS PERF. MONITORING? WHAT THE DATA CAN TELL YOU GETTING THE BIG PICTURE Agenda ABOUT ME
  • 10.
  • 11.
    Project AreasCurrent Role AboutMe - Log Analysis Tool - Performance Monitoring - Data Center Developer on the JIRA Software Enterprise team
  • 12.
    Why I’m giving this talk LearningsFrom Project First project I worked on at Atlassian revolved around perf. monitoring Work on Log Analyser Driver for upgrades to the JIRA Log Analyser tool First Project How useful perf. monitoring can be for customers with the right strategies
  • 13.
    Why I’m giving this talk LearningsFrom Project First project I worked on at Atlassian revolved around perf. monitoring Work on Log Analyser Driver for upgrades to the JIRA Log Analyser tool First Project How useful perf. monitoring can be for customers with the right strategies
  • 14.
    Why I’m giving this talk LearningsFrom Project First project I worked on at Atlassian revolved around perf. monitoring Work on Log Analyser Driver for upgrades to the JIRA Log Analyser tool First Project How useful perf. monitoring can be for customers with the right strategies
  • 15.
    Why I’m giving this talk LearningsFrom Project First project I worked on at Atlassian revolved around perf. monitoring Work on Log Analyser Driver for upgrades to the JIRA Log Analyser tool First Project How useful perf. monitoring can be for customers with the right strategies
  • 16.
    It is animmutable law in business that words are words, explanations are explanations, promises are promises-but only performance is reality. HAROLD GENEEN, FORMER PRESIDENT OF ITT “ ”
  • 17.
    Why I’m giving this talk LearningsFrom Project First project I worked on at Atlassian revolved around perf. monitoring Work on Log Analyser Driver for upgrades to the JIRA Log Analyser tool First Project How useful perf. monitoring can be for customers with the right strategies
  • 18.
  • 19.
  • 20.
  • 21.
    5 Secrets toJIRA no has told you about
  • 22.
    5 Secrets toJIRA no has told you about
  • 23.
    7 Things theydon’t want you to know about JIRA Performance
  • 24.
    7 Things theydon’t want you to know about JIRA Performance
  • 25.
    Performance Monitoring Not Perf. Testing Thisallows for quick reaction to potential performance problems Log File Analysis Review logs around the time reported problems happen to understand why Real-Time Analysis Performance testing is the use of fake data to stress test applications
  • 26.
  • 27.
  • 28.
    Spoiler: You cansee a lot!
  • 29.
    • Response Times •Average Response Times • Response Times by Category • Total Requests • Requests per min • Concurrent Requests • Requests per min by Category • Percentile Response Times • Peak Response Times Some of the data we can see • Response time by Node • Average Response Time by Node • Requests per Node …
  • 30.
  • 31.
    Always keep track ofwhat’s being measured
  • 32.
  • 33.
    Average Response Timeby Nodes (ms)
  • 34.
    Write your ownNewRelicLog Analyser Tool Examples of products you can use
  • 35.
    What the datacan tell you
  • 36.
  • 37.
  • 38.
  • 39.
    Average Response Timeby Nodes (ms)
  • 40.
    Average Response Timeby Nodes (ms)
  • 41.
  • 42.
    Each graph isone piece of a Jigsaw
  • 43.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 44.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 47.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 48.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 52.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 53.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 56.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 57.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 59.
    Don’t AssumeMatch SymptomsFinda Start Point General Strategies Recent Events If we’re investigating problems matching the symptoms to the graph they appear on is really helpful Depending on the time frame of your data, there can be a lot to look through Don’t always assume that there has to be a problem even if it’s been reported Always be mindful of things that have happened recently, i.e. plugin upgrades, application upgrades
  • 60.
    Key Takeaways Try Perf. Monitoring Perf.Monitoring allows you to understand why things happen Have a Strategy Analysing data without a strategy is like looking for a needle in a haystack Understand Why go.atlassian.com/jiraloganalyzer to use our log analysis tool for free.
  • 61.
    Thank you! LUKE DURKAN• GRADUATE DEVELOPER • ATLASSIAN