Introduction
Today applications are complicated…
• Today applications' architecture is complex and processing is distributed.
SOA (Service Oriented Architecture) uses a constant interaction among
systems/applications and technologies:
• Customer portal (IBMWebsphere Portal, Jboss Portal, Oracle Portal, etc.)
• Proxy layer – Enterprise Service Bus (IBMWESB, Oracle ESB, Mule, etc.)
• Process Management layer (IBM Process Server, Oracle BPM, JBPM, etc.)
• “Back-end” (Oracle Weblogic Server, Jboss, Tomcat, IBM Websphere, etc.)
• In such systems it is difficult to answer the following questions:
• Do my customers have a satisfactory level of service?
• What customers have problems? What kind of problems and where?
• What is a current performance and availability of business processes?
• Do I satisfy SLA and if not, why?
• Do I solve my customers problems efficiently?
• Do I have a deep insight into my SOA components and IT infrastructure?
• Can I solve a problem quick enough before my customer will know?
…and problems are difficult
• Selective problems classification
• Performance problems
• Transactions consistency problems (etc. missing messages)
• Availability problems (hanging JVMs)
• Performance problems can be solved in many ways:
• Logs analysis
• Raising log levels or adding more logging
• Making use of some randomly selected applications from the Internet.
• Calling for help from some consulting companies
• Calling for support from software vendors
• Engaging our own resources.
• However, none of the mentioned ways guarantees quick and efficient
problem solution.
40% of administrator work time
takes production problems analysis
The question is:
Is it possible to save this 40% work time, which is spent on repeatable and boring
tasks, and spend it something more interesting?
•5
•6
try to analyze this or find root cause...
Traditional monitoring - dashboards
What is Flopsar Suite?
Flopsar Suite is a tool to manage application performance in extremely
busy production environments based on Java. Flopsar presents a
completely new and innovative approach to monitoring.
Key features:
• The fault detection or non-standard behavior of the application. Real-
time
• Full diagnostics with an indication of the causes of the problems
• 24/7/365 - Full history of processing along with performance analysis
• Intuitive graphical interface. Zero cost training
• Absolute confidence. No compromises. Best Performance APM.
• Zero cost of implementation and maintenance
7
Flopsar (Flop Search And Rescue)
Most important feature
Shorten analysis time incidents and accidents by more than 99%.
On the basis of behavioral studies we have created a unique panel showing all requests from monitored
application servers. Because this panel usually shows thousands of requests, we called it Galaxy. With its help, the
operator with a single click is able to isolate the cause of failure or unexpected behavior of the application to the
component or resource responsible for the problem. The data presented in Galaxy dots are shown with a
maximum delay of 5 seconds after end of transaction.
Picture from http://variety.com/2015/film/news/john-nash-beautiful-
mind-dies-1201504401/
The x-axis is divided into 15 slots. On the right side
are visualized currently completed requests, far left
- requests completed 15 minutes ago.
The Y-axis shows the response
time of a single request.
Dot = single request. After selecting
a set of dots, you can make the
analysis
Technical details
0.5 sec Galaxy view 500ms scale
5 sec galaxy view 2500ms scale
15 sec Galaxy view 15000ms scale
One minute view… and anomaly around 30 sec
Root cause
Details. Problem – poor SQL
One more problem…
And immediate root cause…
Technical details
•18
• Matrix Pattern • Streaming Pattern
19
Galaxy – automatic failure detection
• Unique, acting on the real-time data spectrum analysis panel, showing
operators violated SLAs for business functions, including the causes of faults,
with possibility of deep tracing
20
Automatic slowdown reporting, and their causes
at selected business functions
• Flopsar can Report any application server
engine parameters – for example:
• Number of messages in JMS queues
• Status of connection pooling
• Thread’s Status
• We also report efficiency of the Java Virtual
machine and Garbage Collector’s
mechanism, with potential memory leaks.
The state of the application server / JVM
21
• Independent of the technology, agent allows rapid integration of Flopsar
with components that may affect on the performance, but are not written in
java.
• Using simple scripts the independent agent can collect data from the
operating system, filesystems, networks, databases, MQ, Tuxedo, or any
other components.
• Data is stored in a database and is correlated with other performance data.
22
Independent agent
23
• We do not loose information due to data aggregation. We save all traffic.
• In Flopsar Suite any report and can be easily created. In addition to the
standard set of built-in analytical reports for presenting historical data and
the export standard, we give users powerful capabilities - every piece of
data stored in the database Flopsar Suite can be obtained using the
supplied driver.
Data analysis and reporting
• A security - from the moment of their generation – until write to a database.
• Cluster proxies components, queue, multithreading, separate processes for
handling GUI / API
• Infrastructure code written in C. Performance and quality at every stage of
data processing is under our control.
• The incredibly high rate of transactions per second (TPS) – 20.000 metric /
second.
• This is the load we are able to handle on a single database instance, without
increasing message queues
24
We don’t sell garbage
Best quality and performance
• Only 30 minutes should be spent on installation and configuration of Flopsar
Suite.
• In the thirty-fifth minute, a team maintaining applications and can draw
their first conclusions on causes of invalid operations or slowdowns. The tool
does not need maintenance and it can be reconfigured within minutes.
25
Zero cost of maitenance and traning
• In the manufacturing phase of the system and its development
• Monitoring system
• Performance Analysis
• Analysis of production problems
• Capacity analysis
• Monitoring users
• During Go-Life or during the stabilization period
• During testing, the system load and "stress" tests
• Detecting Memory Leaks
• Verification of the results of performance tests
• Verification of integration tests
• Managing application code
• Profiling the application code
• Verification of functional tests
26
Where Flopsar can be of help?
Contacto: www.tesacom.net
Marcelo Marinangeli
mmarinangeli@tesacom.net

Flopsar tesacom-technical-introduction v1a-eng

  • 1.
  • 2.
    Today applications arecomplicated… • Today applications' architecture is complex and processing is distributed. SOA (Service Oriented Architecture) uses a constant interaction among systems/applications and technologies: • Customer portal (IBMWebsphere Portal, Jboss Portal, Oracle Portal, etc.) • Proxy layer – Enterprise Service Bus (IBMWESB, Oracle ESB, Mule, etc.) • Process Management layer (IBM Process Server, Oracle BPM, JBPM, etc.) • “Back-end” (Oracle Weblogic Server, Jboss, Tomcat, IBM Websphere, etc.) • In such systems it is difficult to answer the following questions: • Do my customers have a satisfactory level of service? • What customers have problems? What kind of problems and where? • What is a current performance and availability of business processes? • Do I satisfy SLA and if not, why? • Do I solve my customers problems efficiently? • Do I have a deep insight into my SOA components and IT infrastructure? • Can I solve a problem quick enough before my customer will know?
  • 3.
    …and problems aredifficult • Selective problems classification • Performance problems • Transactions consistency problems (etc. missing messages) • Availability problems (hanging JVMs) • Performance problems can be solved in many ways: • Logs analysis • Raising log levels or adding more logging • Making use of some randomly selected applications from the Internet. • Calling for help from some consulting companies • Calling for support from software vendors • Engaging our own resources. • However, none of the mentioned ways guarantees quick and efficient problem solution.
  • 4.
    40% of administratorwork time takes production problems analysis
  • 5.
    The question is: Isit possible to save this 40% work time, which is spent on repeatable and boring tasks, and spend it something more interesting? •5
  • 6.
    •6 try to analyzethis or find root cause... Traditional monitoring - dashboards
  • 7.
    What is FlopsarSuite? Flopsar Suite is a tool to manage application performance in extremely busy production environments based on Java. Flopsar presents a completely new and innovative approach to monitoring. Key features: • The fault detection or non-standard behavior of the application. Real- time • Full diagnostics with an indication of the causes of the problems • 24/7/365 - Full history of processing along with performance analysis • Intuitive graphical interface. Zero cost training • Absolute confidence. No compromises. Best Performance APM. • Zero cost of implementation and maintenance 7 Flopsar (Flop Search And Rescue)
  • 8.
    Most important feature Shortenanalysis time incidents and accidents by more than 99%. On the basis of behavioral studies we have created a unique panel showing all requests from monitored application servers. Because this panel usually shows thousands of requests, we called it Galaxy. With its help, the operator with a single click is able to isolate the cause of failure or unexpected behavior of the application to the component or resource responsible for the problem. The data presented in Galaxy dots are shown with a maximum delay of 5 seconds after end of transaction. Picture from http://variety.com/2015/film/news/john-nash-beautiful- mind-dies-1201504401/
  • 9.
    The x-axis isdivided into 15 slots. On the right side are visualized currently completed requests, far left - requests completed 15 minutes ago. The Y-axis shows the response time of a single request. Dot = single request. After selecting a set of dots, you can make the analysis Technical details
  • 10.
    0.5 sec Galaxyview 500ms scale
  • 11.
    5 sec galaxyview 2500ms scale
  • 12.
    15 sec Galaxyview 15000ms scale
  • 13.
    One minute view…and anomaly around 30 sec
  • 14.
  • 15.
  • 16.
  • 17.
  • 18.
  • 19.
    • Matrix Pattern• Streaming Pattern 19 Galaxy – automatic failure detection
  • 20.
    • Unique, actingon the real-time data spectrum analysis panel, showing operators violated SLAs for business functions, including the causes of faults, with possibility of deep tracing 20 Automatic slowdown reporting, and their causes at selected business functions
  • 21.
    • Flopsar canReport any application server engine parameters – for example: • Number of messages in JMS queues • Status of connection pooling • Thread’s Status • We also report efficiency of the Java Virtual machine and Garbage Collector’s mechanism, with potential memory leaks. The state of the application server / JVM 21
  • 22.
    • Independent ofthe technology, agent allows rapid integration of Flopsar with components that may affect on the performance, but are not written in java. • Using simple scripts the independent agent can collect data from the operating system, filesystems, networks, databases, MQ, Tuxedo, or any other components. • Data is stored in a database and is correlated with other performance data. 22 Independent agent
  • 23.
    23 • We donot loose information due to data aggregation. We save all traffic. • In Flopsar Suite any report and can be easily created. In addition to the standard set of built-in analytical reports for presenting historical data and the export standard, we give users powerful capabilities - every piece of data stored in the database Flopsar Suite can be obtained using the supplied driver. Data analysis and reporting
  • 24.
    • A security- from the moment of their generation – until write to a database. • Cluster proxies components, queue, multithreading, separate processes for handling GUI / API • Infrastructure code written in C. Performance and quality at every stage of data processing is under our control. • The incredibly high rate of transactions per second (TPS) – 20.000 metric / second. • This is the load we are able to handle on a single database instance, without increasing message queues 24 We don’t sell garbage Best quality and performance
  • 25.
    • Only 30minutes should be spent on installation and configuration of Flopsar Suite. • In the thirty-fifth minute, a team maintaining applications and can draw their first conclusions on causes of invalid operations or slowdowns. The tool does not need maintenance and it can be reconfigured within minutes. 25 Zero cost of maitenance and traning
  • 26.
    • In themanufacturing phase of the system and its development • Monitoring system • Performance Analysis • Analysis of production problems • Capacity analysis • Monitoring users • During Go-Life or during the stabilization period • During testing, the system load and "stress" tests • Detecting Memory Leaks • Verification of the results of performance tests • Verification of integration tests • Managing application code • Profiling the application code • Verification of functional tests 26 Where Flopsar can be of help?
  • 28.