insights into enterprise application management : how do we make them faster?

Post on 10-May-2015

1.425 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Hugh Brien did this presentation at Surge

TRANSCRIPT

APM Insights into Enterprise Application Management : How do we make them faster?

Hugh P. Brien

AppDynamics Proprietary & Confidential – All Rights Reserved

Copyright © 2010 AppDynamics. All rights reserved.

Who We Are

The leading provider of application management for

modern application architectures in both the

cloud and the data center

Our Heritage Company Facts

• Founded in 2008 by Jyoti Bansal• Based in San Francisco• 40,000 users of AppDynamics Lite

Delivering Measurable Results

Cloud AppMgmt

Needed to Maintain QoS with 40%

growth

Managing > 3000 Amazon EC2

nodes

<1% Overhead in Production with 100% visibility

New Distributed Architecture

$1B in eCommerce

revenue

End-to-end visibility into transactions

Ease-of-Use a key factor for broad

adoption

MTTR Reduction

Reduced MTTR from 6 hours to

10 minutes

Production Monitoring with Low Overhead

Deep Visibility, “Painless”

Deployment

Agile Dev & Operations

Agile – 10 Releases per

Month

Needed Visibility into Custom App

Reduced MTTR for slow

response & memory leaks

•Byte Code injection instrumentation

•Business transaction data collection

•Machine statistics data collection

•Distributed automation extension

AppDynamics Architecture

Machine Agent

Application Server Agent

(JVM/CLR)

One way HTTPS connections

•Central Management Server•SaaS or On-Premise deployment options

AppDynamics Controller

Copyright © 2010 AppDynamics. All rights reserved.

Continuously Discover and Auto-Instrument all tiers of your Application for 10x Visibility

AppDynamics Pro – How it Works

Self-Learn and Baseline the “Normal” Response Time of Business Transactions

Monitor Application Health in Production 24/7

Troubleshoot & Diagnose root-cause 90% faster

Respond via Policy-Driven automation

Copyright © 2010 AppDynamics. All rights reserved.

Monitor Application in Production 24/7

AppDynamics Pro – How it Works

Scalability – Very Very Very Large

Policy Engine And Automation

Big Data Visibility

Policy Driven Automation

Application Management Challenges

1. Production downtime and/or recurring performance problems

2. Poor end-user experiences may result in lost revenue or brand damage

• Inability to determine root cause and resolve problems in timely manner

Three Key Customer Challenges

Application Health

Copyright © 2010 AppDynamics. All rights reserved.

Tuning Your Enterprise Applications

What’s Normal

Copyright © 2010 AppDynamics. All rights reserved.

If You Can Measure It You Can’t .....

How Do You Know?

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

What Are “Your” KPIs?

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Major Garbage Collections

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Major Garbage Collections

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Who Didn’t Read the Directions? - Server

Configuration

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Request Overload

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Hibernate Abuse

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

Web Services

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

SynchronousTransactions

Copyright © 2010 AppDynamics. All rights reserved.

Knowledge Is Power

And By the Way It’s Not the Database It’s Probably Your

Code or Indices

Copyright © 2010 AppDynamics. All rights reserved.

Primary Application Bottlenecks

- Major Garbage Collections- Who Didn’t Read the Directions? - Server Configuration- Request Overload- Hibernate Abuse- Web Services- Synchronous Transactions- And By the Way It’s Not the Database It’s Probably Your Code or Indices

Copyright © 2010 AppDynamics. All rights reserved.

How is our Approach Disruptive?

OLD WAY NEW WAY SO WHAT?

Monitor Component (server, JVM/CLR,

method) Health

Manage Application health by Business

Transaction

Right Context for Distributed Apps

Aligned with Business

Manually select what code to instrument

Self-learned Instrumentation

Easy, Easy, EasyEnables Agile

Operations

Detection based on manually set static

thresholdsDetection based on self-

learned baselinesAdaptive

No False Alarms

•Byte Code injection instrumentation

•Business transaction data collection

•Machine statistics data collection

•Distributed automation extension

AppDynamics Architecture

Machine Agent

Application Server Agent

(JVM/CLR)

One way HTTPS connections

•Central Management Server•SaaS or On-Premise deployment options

AppDynamics Controller

Product Demonstration

Copyright © 2010 AppDynamics. All rights reserved.

APM Requirements for Modern Apps

• Fit for Production Environment = Low Overhead• Ease of Use

• Can I deploy it myself in less than an hour?• Is the instrumentation self-learned or is it manual/arduous?• Are “normal” performance baselines self-learned or are thresholds manually set?

• Application Visibility• Does the tool discover, visualize and monitor distributed business transactions?• Does the tool adapt automatically in agile environment?• Does the tool adapt automatically in a hybrid cloud/virtual environment?

• MTTR Reduction – Rapid Troubleshooting & Root Cause Analysis• When a problem occurs, is diagnostic data automatically captured or is it manual?• Can the tool identify problems between the tiers/services in a SOA environment?• Are there code-level diagnostic captured for every bad session?• Does the tool find root cause of memory leaks or just identify them?

• Preferred Unit of Management is the Business Transaction• Versus Server health vs Method-only vs Log files

• Ability to Automate Action• Can the tool take action based upon what it detects?

• NetFlix, Orbitz, Exact Target, Family Search, IG Index

• Todd Radar • Orbitz• Exact Target• IG Index• StubHub• Family Search

• Real Time Live• Automation• Big Data

Copyright © 2010 AppDynamics. All rights reserved.

Common Application Performance Issues

Problem Type Frequency Difficulty to Diagnose

Slow SQLVery High Medium

Slow Response / Errors due to malformed code

Very High High

SOA Remoting IssuesHigh High

Thread & Synchronization Issues

Medium Very High

Memory Leaks & ThrashMedium Very High

Copyright © 2010 AppDynamics. All rights reserved.

What’s the New Idea?

New World: Distributed applications running constantly Changing code on Dynamic infrastructure

Key Question:

How can you assure application availability when the code and the topology are constantly changing?

New idea: Monitor the “Business Transaction”• The Business Transaction = the collection of all processing

across a distributed infrastructure that fulfills a user request• The Business Transaction is a constant in a world of change

top related