AppSphere 15 - Smoke Jumping with AppDynamics

Smoke Jumping With AppDynamics
Jim Waldron, IHS
Principal Global Application Support Engineer

My Career
-  25 years in the industry working with .com’s and Fortune 500 Companies.
-  I have done everything from C, C++, Java programming to Systems Administration
work.
-  My Career found me I don’t believe I found it.
-  Degree is in History not IT.
-  Always have found myself in a Smoke Jumper role.
-  Drop in - fix it - get the heck out!
-  No better way to learn than Baptism by Fire!
-  There is nothing in life I would be rather be doing that what I am doing now.

About IHS
IHS offers a unique combination of information, analytics, and expertise. Our solutions
and capabilities are augmented by a professional team of subject matter experts, analysts,
and consultants.
These thought leaders provide you with actionable intelligence for expedited and improved
decision making. Customers, prospects and the media outlets alike rely upon IHS thought leaders
for analysis, forecasts, and perspectives on topics, events, and issues that impact the global
business landscape.

Industries IHS Provides Services Too
-  Academic and Education
-  Aerospace and Defense
-  Agriculture
-  Automotive
-  Chemicals
-  Electronics
-  Energy Oil and Gas
-  Power and Utility
-  As well as many others
-  Maritime

IT TAKES A BRIGADE NOT AN
INDIVIDUAL TO PUT OUT A FIRE

Initial Jump Into the Fire
-  First day I realized I had dropped into an OH S*&T moment
-  I guess I missed in the interview the gravity of sheer amount of applications that I was
going to support
-  Function of the AppSupport team was pretty primitive
-  History had dictated trust issues between the teams
-  Proponent for change had to build trust between the Development and AppSupport
teams
-  Root Cause Analysis took hours sometimes days to determine

The main function of the AppSupport Team was to jump in a fix the fire nothing
more nothing less!
Calls would come from the International Operations Center and routed to our
team. Time to resolve could take hours to days to figure out
Not only was it time consuming it was not always received well.
Multi-Environment issues made it difficult to determine where the underlying
application was causing the issue
Used to take Days sometime weeks to dig through log files in order to
hopefully determine if the incident was Code, Application or Infrastructure
related – at our customers expense
Copyright © 2015 AppDynamics. All rights reserved. 9

FROM SMOKE JUMPER TO FIRE
BRIGADE

Pulling together a Brigade
-  Battle to gain trust.
- Dev Team = Deployments
- Become a proponent for change.
-  Went from the fire being 10% contained to 75% in a relatively short amount of time.
-  Can Tools Help.

Core Requirements to Choosing an APM
-  Time to Value – Deep visibility out of the box.
-  Ease of use.
-  App to App correlation for multi-application interaction.
-  We needed a partner, not just a vendor.
-  We need a tool that delivers data not emotion.

What AppDynamics Provided that Dynatrace Didn’t
-  Immediate results right out of the box.
-  Detailed Information for All.
-  Finally the good fight was starting to pay off
-  DevOps was starting to become a reality not just a word being thrown around
-  DevOps only works if all teams buy into it not just individuals
-  Functionality to see the application as a whole.
-  Collaboration was set in motion.
-  Resolution was finally on track.

From Lone Jumper to a Brigade
After a few months of gaining trust we decided we would take a look at a
few APM’s to help us to gain a better insight into our outages.
We pitted Dynatrace against AppDynamics over the period of a few months.
We were looking for a partner in this battle not just a solution. We achieved
this with AppDynamics and not so much with Dynatrace.
Determined root cause whether Dev or Ops and have visible proof that
there were definitive - not just speculative.
With visible proof and data, we could go to the RIGHT team and it was easy
to pull together the brigade and put out the fire.

FINALLY A BRIGADE TO FIGHT THE FIRES

What we Learned In Order to Become a Brigade
-  Better insight and communication were key.
-  Finger pointing stopped and collaboration began.
-  Solid Bridges were built.
-  Time to resolve was reduced by 72%. What took days now only takes minutes or
hours
-  With valid results and effective communication application, infrastructure or database issues
are resolved with releases.
-  Complete turn around from 2 years ago.
-  The word DevOps now has meaning.

“AppDynamics is showing issues unfold before synthetics!”
“Is helping us understand our peak times and how it impacts
our customers”
“Could be the difference between an outage or not”
“Much more detail on issues, less confusion”
“The ability to understand where calls are coming from, and
which applications are being affected”
“We are actually working as a brigade and not just smoke
jumpers!” Copyright © 2015 AppDynamics. All rights reserved. 17

AppDynamics Application Impact
Oil and Gas Discovery Tools
Baseline (Pre-AppDynamics) BVA Projected Benefit BVR Realized Benefit
•  2-3 complex
performance issues
per App per month
•  Avg MTTR = 5 days
•  Root cause discovered
10% of time
•  Avg 5 FTEs to
troubleshoot
•  Reduce Root Cause
Analysis by 65%
•  Reduce MTTR by 65%
•  75% MTTR reduction
•  Avg 1 FTEs rather
than 5
•  AND, additional
benefit of improving
Root Cause to ensure
‘less repeat’ issues

AFTER THE FIRE WHAT WAS LEARNED

“We now have a view of all integration points. Much easier to
pinpoint issues and understand complexity. We can also see
opportunities for consolidation.”
– Dan Hauser Principal Software Engineer US Land Team.

Cross Application View (Impossible before AppDynamics)

Multi-Environment Testing
We have several Applications that use our Access Control
Environment.
All Applications have potential to impact our front end.
Determining which Application is causing the issue was difficult and
multiple teams were engaged prior to implementing AppDynamics.
With AppDynamics we are able to determine which Application is
having the issues to engage the proper team to resolve the issue
quickly.
Our Development Teams are now using AppDynamics to test their
code in their Stage Environments to research and determine coding
issues before releasing to production.

Looking Forward
-  By Mid January 2016 we are looking to mature our usage of AppDynamics.
-  We will be implementing Deeper and more thorough use of AppDynamics by the PDD Team
for upstream, pre production tuning and optimization for code and applications.
-  This will allow us to flip the 80/20 balance of reactive/proactive post production Operational
troubleshooting to be 80/20 proactive vs. reactive.
-  This will ensure tighter collaboration between PDD and Application Support during load/
stress testing and pre-production testing to ensure applications are optimal and consistent
between pre and post-production releases.
-  Proactive analysis and trending against baselines to detect and alert on deviations with
sufficient time to collaboratively engage and analyze the cause prior to any customer impact.

Thank You
Jim Waldron
Principal Global Applications Engineer
Jim.waldron@ihs.com
www.ihs.com

Pulling together a Brigade
-  At first it seemed like an uphill battle to gain trust.
-  Started with a simple task in regards to the Dev Team – Deployments.
-  Never gave up and kept fighting the good fight – Became a proponent for change.
-  Went from the fire being 10% contained to 75% in a relatively short amount of time.
-  Started to gain trust and build a team of jumpers to fight the fire.
-  After awhile we started to look into tools to help fight the good fight.

Collaborate, Collaborate, Collaborate!
Prior to AppDynamics multiple excuses were given for the root cause of the issue
We heard that is an known issue and has never really caused any impact before so
why now?
We will fix that issue in the next release and the issue would arise again
My favorite was how do you know that is an issue after searching logs for hours and
sometimes days to determine root cause.
Now, the data provided to the Dev Teams is undisputable and has allowed us to
collaborate with them to detect, investigate and reveal known and potential issues
with their Applications to work towards better coding which leads to less down time as
well as time to resolve.

Production – Stage – Test – Dev All In One Place

Initial Dashboard View Into the Application

Custom Dashboards
Key Dashboards
•  Utilizing the Application Support
Dashboard
•  Providing a dashboard for
Development
•  Merging the two for full coverage
•  SysOps dashboard
Key Metrics
•  Node CPU and Memory usage
•  Call load and slow calls per node
•  Node response time
•  Disk activity per node
•  Slow, very slow, stall CPM EPM

Why These Metrics?
Application versus Infrastructure Insight
Determine responsibility
Visible proof removes any doubt on responsibility

Give Dev Their Own Dashboard!
Development metrics != Ops metrics
Business Transactions are most important to
Developers
Lightning fast resolution to root cause in code

AppSphere 15 - Smoke Jumping with AppDynamics

More Related Content

What's hot (20)

Viewers also liked (11)

Similar to AppSphere 15 - Smoke Jumping with AppDynamics (20)

More from AppDynamics (20)

AppSphere 15 - Smoke Jumping with AppDynamics