Transcript All Java processes on One Screen

Dashboard
All Java processes on One Screen
52 Java Process
monitoring
Dashboard
True- Real Time Data
Every 1 sec
Just finished
Active Service Monitoring
Active service monitoring
• All symptoms of resource related performance
problem is presented in increase in active service.
• JENNIFER updates active service count every 1
second.
• User can see the detailed information of active
services upon request even if the service is not
yet finished.
Active Service Monitoring
What services are queued?
Just Click
Fetch Count
Instance Id
Service Call
Time
Thread Id
Service Name
Response Time
Status
Client IP
SQL Count
Active Service Monitoring
Detail information for a active service
Service Name
Click!
Last exec SQL
Active Stack
Active Profile
Select a transaction
Client IP
Transaction Trace & X-View
Transaction Trace and X-View
• Profiling data is collected for individual transaction.
• Select a transaction requiring analysis to see its
performance and profile.
• Performance and profile is displayed immediately after a
transaction is finished.
• Profile not only shows method-level response time but
also show CPU time, SQL text, parameters & execute
time, FILE/SOCKET open, and TP-CALL.
Transaction Trace & X-View
Response Time Scatter Graph
Transaction Response Time
X-View
High Response
Time – Tune It!
Error Transactions
(red)
Need Analysis
Normal
Transactions
Transaction End Time
Transaction Trace & X-View
X-View Patterns
Water drop Pattern
Waterfall Pattern
Matrix Pattern
Layered Cake Pattern
Transaction Trace & X-View
Review Transaction Profile
Click & Drag
Click
Focus-in on the
abnormal transactions!
Transaction Trace & X-View
Transaction Summary & Profile
• Transaction Performance Summary
• Service Name
• Client IP
• SQL Time, TX Time, Fetch Time, CPU Time
• Response Time
• JDBC Resource Usage
• Connection Open / Close
• Set Auto Commit / Commit / Rollback
• Statement / PreparedStatement / CallableStatment Class
• SQL and SQL Parameters
• External Service Call(ex TP-CALL)
• Method Response Time
• Method Parameter/Return
• File / Socket Open
• New thread init
• CPU Usage
• Current thread Name
• Customizable (with adapter)
JENNIFER Features
•
Service Monitoring
•
•
•
•
Peek Load Control
Transaction Profiling and X-View
•
•
•
•
•
Concurrent Users, Visit Users, Think Time
JDBC & SQL Monitoring
•
•
•
•
Response time scatter graph(X-View)
Transaction base profiling
Dynamic profiling
Dynamic stack trace
User Monitoring
•
•
Powerful service naming (URL, Http Parameter, Method Name or Parameter/Return)
TPS, Active Services, Hit per hour, Average response time, Client IP
JDBC Connections, JDBC status, JDBC object leakage
SQLs & SQL parameters
Uncommit, rollback trace
Resource Monitoring
•
•
Memory, Heap Memory, File/Socket
CPU, CPU core
JENNIFER Feature cont…
•
Performance Statistics
•
•
•
•
•
Utilities
•
•
Non App Server, Remon(Shell,SQL,Class), Extra Agent, Profiling customization
Alert
•
•
Multi JENNIFER server integration, manage by task or domain
Extensions
•
•
JDBC object leakage, Collection trace, Live object counting, Http session dump
Integrations
•
•
Java Env, Retrieve loaded classes, Jar check
Leakage Trace
•
•
Daily resource usage
Daily service performance
Relation between Applications and DB Tables (CRUD Matrix)
Relation between Applications and (SQLs or External services)
44+ Predefined Alert or Exception and customer alerts can be defined
Reporting
•
•
Daily, Weekly, Monthly performance and error statistic summary
Other commonly used and user-defined template based reporting
JENNIFER’s Overhead
Though many system experts agree that use of APM solution is necessary, they often
denote excessive overhead as a major problem for its disuse.
JENNIFER has been designed with overhead in mind~
According to survey that we conducted amongst our variable customers, JENNIFER’s
overhead is estimated to about 3~5% average.
JENNIFER’s Overhead
How does JENNIFER keep its overhead so low?
• Efficient use of data ports
– JENNIFER uses variety of UDP, TCP ports and data traffic efficiency
algorithm to minimize the overhead.
• Dynamic Profiling
– Turn on/off profiling level right from the dashboard without restarting the
application or server*
• Efficient data collection process
– Amount of data collected is the main culprit of overhead. Do not trade
overhead for needless statistic data. JENNIFER collects just enough data to
resolve all performance and establish SLA
Installing and deploying JENNIFER
Many business may not realize their need for APM until its too late.
JENNIFER
Wily
Installation Time
(1~4 servers) / (5~10
servers)
5 ~ 8 hours
2 ~ 3 days
2 ~ 4 days
6 ~ 10 days
Basic Configuration
(1~4 servers) / (5~10
servers)
1~2 hours
3~5 hours
1~2 days
3~5 days
Immediate
Long learning curve
How soon can provide
help after install?
JENNIFER is quickly installed and configured, thus providing immediate
help with resolving customer’s application performance problem.
JENNIFER Structure
JENNIFER is composed of three parts: JENNIFER Agent, JENNIFER Server, and JENNIFER Client.
Java Application
Java Application
Server
UDP
Java VM
OS
TCP
TCP
Java Plug-in
WEB Browser
UDP
Java VM
OS
(JENNIFER Structure Diagram)
16
JENNIFER Server / Client Requirement
JENNIFER requires following H/W and S/W specification for proper operation.
JENNIFER Server
JENNIFER Client
•
Platform – Not affected by OS
•
OS - Windows XP, Vista
•
System requirement may vary depending on the
•
Web Browser – Firefox 3.x, MS IE 6.0/7.0/8.0
quantify of Agent and data collected from the target
•
MS IE 6.x will not be supported on JENNIFER 4.1
system (Agent Data, JENNIFER DB, etc…)
and above.
•
- Number of Agents and TPS monitored by JENNIFER
Server (50 Agents or 500 TPS)
- ex. 1 – 200 or more Agents
- ex. 2 – 20 or more Agents, 500 TPS or more
* Hardware, Application Size (APPLS, SQLS) may also
have effect.
Java Applet Plug-in - JDK 1.6.0_10 and above.
JENNIFER Agent Supported Platform
JENNIFER supports most of the OS, JAVA (Sun, IBM, BEA, etc…) , and Java Application Server
commonly found in current IT environments.
Operating System (OS)
Java Application Server (Java Application Server)
• AIX 4.3.3, 5.x 32bit, 64bit
• BEA WebLogic 5.1, 6.x, 8.x, 9.x, 10.x
• HP-UX 11.x 32bit, 64bit, Itanium 64bit
• IBM WebSphere Application Server3.5, 4.x, 5.x, 6.x
• Sun Solaris 2.8, 2.9, 10 32bit, 64bit, x86
• Tmaxsoft JEUS 3.x, 4.x , 5.x, 6.x
• Intel Linux 32bit, Redhat Itanium 64bit
• Oracle Application Server 9iAS, 10gAS, OC4J, ERP
• Compaq Tru64 UNIX OSF1
• SUN Application Server 7.x, 8.x,9.x
• Microsoft Windows 2000, XP, 2003, Vista
• Fujitsu Interstage 5.x, 6.x, 7.x
• IBM iSeries(AS400) for WebSphere
• Hitachi Cosminexus 7
• IBM z/OS for WebSphere, zLinux
• Sybase EAServer 4.x, 5.x
• Macromedia JRun 4.x
• Apache Jakarta Tomcat 3.x, 4.x, 5.x,6.x
• Caucho Technology Resin 2.x, 3.x
• RedHat JBoss Application Server 3.x, 4.x
• Apache Jserv
• GlassFish
JENNIFER’s Paradigm
• Easy to deploy, lite on overhead.
• Service oriented approach to performance problem
(Active Service, X-View).
• APM must be real-time, immediate recognition of
problem, immediate analysis of root-cause.
Contact Information
• US Location : 360 Fairview Way
Milpitas CA, 95035
• US Office Phone : +1-408-946-5508
Fax
: +1-408-946-5509
• Sales Inquiry: [email protected]
 Java Application Serveron Hong, Business Development Manager
Java Application [email protected]
• Partnership Inquiry: [email protected]
 Java Application Serveron Hong: +1-408-464-8321
Java Application [email protected]
 Andy Lee: +1-408-946=5508
[email protected]