Measuring performance of EJBs

Hi,
I am interested in measuring the performance of my beans. I realize that "performance" by itself is rather generic. To be very specific, I want to measure the time taken by my EJB to perform a database activity. For example, how does the update of my CMP bean compare with a JDBC call from a BMP bean ?
Is the capture of such timing data possible from the affected bean itself ? I thought of using AOP for this but ran into issues which caused me to dump that approach.
Are there any tools available which can be used to capture such data - even from a client perspective ?
Thanks

try:
http://www.jamonapi.com it can measure JSP, SErvlets, EJB and more
but just for EJB it is maybe better write own tool, it can not be so difficult.
System.currentMillis(); at the begginig and end might be easiest and maybe best
performance tool.
Point not to slow donw process be measuring tool itself.
Marian

Similar Messages

How to improve the Performance of Ejb ?

I have many ejb deployed but i was not happy with performance of them as they take more time to execeute
is there any ohter way to increase the performance of ejb
my application server in this case is weblogic

hanu007,
The J2EE Pattern Catalog includes numerous patterns that describe performance issues with respect to designing systems with Enterprise Beans. I'll reiterate a few points and encourage you to have a look at the beta documentation online (http://developer.java.sun.com/developer/technicalArticles/J2EE/patterns/) or the updated, full documentation in the book for further detail.
Value Object
Entity beans are shared, transactional objects representing persistent data. Since every invocation on an Entity bean is potentially a remote invocation, incurring container overhead, it is important to remember to expose attributes of the bean in a course-grained fashion. In other words, getting a reference to an Entity bean and then doing lots of fine-grained access (ie: lots of sets and gets) will prove quite expensive from a performance standpoint. This pattern suggests returning multiple values as a single atomic unit, so that the number of bean invocations is minimized.
[Just FYI, related to the Value Object pattern is:
Value Object Assembler
The value object assembler builds composite value objects from different data sources. The data sources are typically Session beans or Entity beans, which provide value objects to the assembler. The assembler then creates a composite object from these value objects.]
Session Facade
Using a session bean as a uniform, course-grained point of access into underlying business objects can also improve performance by cutting down on unnecessary remote invocations. If the system is designed with one session facade per use case, for example, or with an overly tight one-to-one mapping of Session beans to Entity beans, then network latency can become an issue and performance can suffer. This pattern describes the preferred mappings and may be worth a more detailed look.
Composite Entity
This pattern addresses the granularity of the Entity beans in the system. Too often Entity beans are mapped one-to-one to rows in a relational database, which can cause major performance problems and also creates a tight dependency of these objects in your system (and their clients) to your database schema. Entity beans are better suited to being course-grained components and this pattern describes using an Entity bean to manage a graph of interrelated persistent objects. This pattern can be implemented using either BMP or CMP, and refactoring a system that has too many fine-grained Entity beans to use this pattern can certainly have an impact on performance
I've tried to touch on a few of our patterns that address performance issues in the business tier, and hope this helps as a starting point for your researching your problems.
Thanks,
Dan Malks

Use SE30 to measure performance in background

Hi,
I want to measure performance of an existing program by running it in background because it takes a really long time to run.
I m trying to use the "Schedule Measurements for user service" option from the menu. But when I click on the new icon after going to Schedule Measurements for user service menu, I get a short dump with message
MOVE_NOT_SUPPORTED.
on the following line of code of standard SAP program SAPMS38T
> convert time stamp l_ts time zone sy-zonlo.
My question is
1. Am I using the right option to measure the performance in bacground or is there any other way (except changing code to put log statements)
2. How can I fix the above problem.
Will give points to the right answer. Thanks for reading.

I generally use ST05 to measure performance.
Rob

DBA Staff KPI(s) - Measuring Performance

DBA Staff KPI(s) - Measuring Performance



Hi,
I am an IT manager, managing two solutions: "SCM" & "WMS (Warehouse Management System)", and there is a team for each solution.
DBA staff are involved, almost, with everything (database administration, handling system/application performance & stability, Application support, creating & automating reports for all levels, monitoring interfaces issues, troubleshooting, handling investigations for operations/management/business/research/system issues ...etc.)
Its not easy to identify their KPI(s) and the additional responsibilities.

If you know some common KPI(s) that have been used before, I really appreciate if you can share them with me, so I can evaluate my team properly.




Thanks ......
Regards,
Ala' 


Just a comment that while your Subject says "DBA", the job responsibilities include many other things which are outside the definition of a DBA role, eg Application support, creating "reports" for "all" levels, troubleshooting "business" issues, etc
So what does "all" levels mean? Does that mean reports from the Application? or just the technical aspects of database administration like I/O, memory, latches etc. What does troubleshooting "business" issues mean? Does that mean identifying why the accounting books don't balance?
I understand that while the DBA role can be clearly defined in theory (Oracle Database Administrator's Guide - Task of a DBA), in practice many organizations, especially the smaller ones, are closer to what you have described. In such cases, the approach I'm familiar with has been to define as much as practicable the primary and secondary responsibilities of each position, and define the KPIs for the primary and make rather vague statements for the secondary role(s).
Thank you that was valuable,
So what does "all" levels mean?Answer: Application Reports go to all level of management, from department managers to senior mgmt and directors.
It includes,
- Productivity measurements.
- Identifying some of business requirements.
- Creating/Building estimates for any CR.
- Should be involved with warehouse expansions plans.
- Reporting critical operations & inventory activities.
- Building interactive reports.
- Cost allocation, ... etc.
Then the normal DBA comes which includes I/O, memory, DB performance, application & DB users ... etc.
You are right: I can take their primary responsibilities to define their KPIs and leave secondary responsibilities on vague statements for the secondary role(s), so, I can evaluate them according to the primary part and leave the secondary part for them to compete on it.
Thanks again

Measuring Performance of Storage Systems

Hi there
I know, measuring performance of storage systems with benchmarking tools is nothing compared to realworld scenarios, but they give a nice overview in test environments.
Under Windows there is a de-facto standard called IOMeter, which contains the traffic generator (dynamo) and the graphical test suite IOMeter.
It seems that the dynamo part is available for Mac OS X, but only for the PPC architecture (funny, as IOMeter was originally developed by Intel), and the graphical part is available for windows only.
I've seen sites on the Apple HP where they said, they have tested the xraid with IOMeter on a Intel XEON. Does anyone know how they did this, as IOMeter is not really available for Mac OS X on Intel ....
Are there any comparable tools, i.e. a traffic pattern generator and an analzer? Maybe one can use dtrace / Instruments fur the analyzer part, but what to use for the generator?
I need stuff more specific than the activity monitor, etc. as I need to define different block sizes, sequential / random rads / writes and their balance, etc. namely, evenrything IOMeter can do. I can't believe there is nothing out there running on Mac OS X Intel ......
Thanks a lot in advance

It seems like I can get it to compile correctly. However, when I try to start the IOManager I get the following errors...
./IOmanagerosx.cpp: line 52: /Applications: is a directory
./IOmanagerosx.cpp: line 53: /Applications: is a directory
./IOmanagerosx.cpp: line 65: //: is a directory
./IOmanagerosx.cpp: line 66: //: is a directory
./IOmanagerosx.cpp: line 67: //: is a directory
./IOmanagerosx.cpp: line 68: //: is a directory
./IOmanagerosx.cpp: line 69: //: is a directory
./IOmanagerosx.cpp: line 70: syntax error near unexpected token `('
./IOmanagerosx.cpp: line 70: `int Manager::ReportDisks(TargetSpec * disk_spec)'
Any idea? Has anyone gotten IOMETER to work on OSX 10.5.x or 10.4.x?

Tool to measure performance of the web application

Among these tool which one is most authentic to show correct data & is widely used to measure performance (both for on-premise and online):
Fiddler
DotTrace
HttpWatch
Developer Dashboard
F12 in IE browser etc.

among the tool i have mentioned which one is dependable to measurejquery coding?
Fiddler
DotTrace
HttpWatch
Developer Dashboard
F12 in IE browser etc.
Please share your experience only among given choice. There are many tools but people says they don't show correct results so I am not interested with other 3rd party tools.

Measuring Performance of Purchase department

Is it possible to measure performance of purcjasing department, based on minimizing the inventory.
Regards
Mahesh

Hi,
Yes you can measure the performance of the Purchase department.
Purchase department is mainly linked with the Purchaseing groups.
You can take the report based on the Purchasing groups.
Check with MC01 transaction --- Logisistics information library
Here goto Purchasing -- Select the option Purchasing group
You will get lot of reports based on the Purchasing group.
1) Purchasing values (purchasing group view)
2) Purchasing activities (purchasing group view)
3) Overview of purchase orders (purchasing group view)
4) Overview of scheduling agreements (purchasing group view)
5) Overview of RFQs (purchasing group view)
6) Overview of contracts (purchasing group view)
Some other standard reports are
1) How many POs are created in a Month per Purchasing groups ( ME2N etc)
2) How many POs are open in a month ( ME2N etc)
3) Price comparision with the vendors ( ME49)
4) Analysis of Net order values ( ME81N)
other wise you can consolidate all the reports & take a help of ABAPer to develop the Z report which can taken when ever you required about the performance of the Purchasing department.
Hope this will help you
rgds
gsc

How to measure performance of supplier when using scheduling agreement ?

Hello all,
My client has an absolute need to be able to measure the performance of its suppliers based on delivery dates and delivered quantities. That is to say he needs to be able to compare what dates and quantities were asked to what has been really delivered.
Most of the procurement processes used are based on scheduling agreements : schedule lines are generated by MRP and forecast is sent to supplier while firm requirements are sent through JIT calls.
It seems that when doing GR in MIGO, it is done against the outline agreement number, and not against the call. Therefore, we have no way to compare dates and quantity with what was expected (in the JIT call).
Do you know if SAP proposes a standard solution to this, and what could be a solution to this issue ?
Thanks for your help
E. Vallez

Hi,
My client faced the same problem and we ended up developing an own analysis in LIS. Since the GR is not linked to specific schedule line (SAP does some kind of apportioning, but it doesn't have to correlate to the correct match), one needs to do assumptions. Our assumption was the closest schedule line, i.e. each GR is related to the schedule line with the closest date. Then all GR the same day are totaled together before the quantity reliability is calculated, since the very same shipment can be reported through several GR transactions in SAP (one per pallet).
If anybody has info about what SAP has to offer in this question (or is developing), please tell us!
BR
Raf

How do you measure performance of an item renderer?

I'm creating an ItemRenderer in Flex 4.6 and I want to know how to measure total time to create, view and render an item renderer and how long it takes to view and render that item renderer when it's being reused.
I just watched the video, Performance Tips and Tricks for Flex and Flash Development and it describes the creation time, validation time and render time and also the reset time. This is described at 36:43 and 40:25.
I'm looking for a way to get numbers in milliseconds for total item renderer render time and reset time (what is being done in the video).

To answer your first question, in this video Ryan Frishberg recommends measuring and tuning your code. I'm trying to follow his example for my own item renderers.
I've taken some key slides out to show you.

How to measure performance?

Hi all,
I have a scenario where in i need to check performance of design being used.
I have one infocube in which data is on Per calender day basis. I have loaded that data in other cube on fiscal year basis with only specific characteristics and key figures i wanted from first cube.
How do i check performance if data is fetched from firtst cube and compare it with time taken to fetch data from other cube.
can i measure it if i am fetching data in Function module using ABAP?

Hi..........
If u want that then..............create query on both the cubes...........then take query statistics............
For this u can use tcode........... RSRT ........it shows the raw time and not percentage of the time that the query spent in each area............
For the percentages, you can either calculate them yourself or use transaction ST03 (expert mode). this will show the breakdown by %.........
Or u can schedule the following chains....order to load BI Statistics data to the Technical Content:
Master Data
System Master Data - 0TCT_MD_S_FULL_P01
This loads text for objects like u2018Process Statusu2019, u2018BI Object typeu2019, u2018Process Typeu2019
Content Master Data - 0TCT_MD_C_FULL_P01
This loads attributes & text for objects like u2018Process Variantsu2019, u2018Process Chainu2019
Initialization Loads
Query Runtime Statistics - Init - 0TCT_C0_INIT_P01
Data Load Statistics - Init - 0TCT_C2_INIT_P01
These process chains need to run only once (Immediate scheduling).
Delta Loads
Query Runtime Statistics - Delta 0TCT_C0_DELTA_P01
Data Load Statistics - Delta 0TCT_C2_DELTA_P01
These process chains can be scheduled for periodic execution
Delta Loads
Query Runtime Statistics - Delta 0TCT_C0_DELTA_P01
Data Load Statistics - Delta 0TCT_C2_DELTA_P01
These process chains can be scheduled for periodic execution
I hav already given u a link.......check that........
Hope this helps.......
Regards,
Debjani........
Edited by: Debjani Mukherjee on Nov 17, 2008 2:05 PM

Measure Performance

I have an IDOC to XI to Vendor scenario.
There is no need of a BPM at this point, because it is an Asynchronous.
But what I feel is introducing a BPM will help lots of development cost in the future. Because, in future if the scope changes to Build the IDOC or if they are looking for the Technical response, then we need to go for BPM.
But bringing the BPM will definitely hit the performance.
I want to study the two cases and want to submit a report on that. How can I do a STRESS testing and can measure the performance.
Thanks

Hi,
one of the ways would be to get a stress tool
like loadrunner (http://www.mercury.com/us/products/loadrunner/)
it comes with a free trial and do a stress test with that
then you can just compare the result in the XI
with such tools you can just record a transaction in
the sap or an http call or almost anything
and then reuse it many times
but you can also write your own scripts if you wish
Regards,
michal
<a href="/people/michal.krawczyk2/blog/2005/06/28/xipi-faq-frequently-asked-questions">XI / PI FAQ - Frequently Asked Questions</a>

Bad Font-Measuring Performance under Windows 8.1

Why does the following little loop (in c#) perform so bad using current version of WPF (.NET 4.5.2) if i change the font-family from "Segoe UI" to "Arial" (or something else...tried "Times New Roman" and "Courier New"
- Same problem)?
var tb = new TextBlock {Text = "Testtext", FontFamily = new FontFamily("Arial")};
for(int i = 0; i < 100000; i++)
tb.InvalidateMeasure();
tb.Measure(new Size(double.MaxValue, double.MaxValue));
With font-family set to Arial this block of code takes about 7.6s on my machine. Font-family set to "Segoe UI" takes about 1.9s. Why do (most) other fonts than Segoe UI perform so bad during measurement? Is there any tweak around, that does avoid
this enormous loss in performance?
As i found found out, "Tahoma", "Lucida Sans" and "Microsoft Sans Serif" are being measured really fast, also. Is this some "System-internal font"-thing?
Yes, i know, this is really constructed and broken down to a minimal reproducable example. The whole component is a custom datagrid with complex ui- and data-virtualization - much too big to post here. If i set the font of my
grid to, let's say "Arial", the scrolling-performance gets really bad. Using Visual Studio's Profiler i tracked the problem down to the measurement of my single grid-cells, which basically measure single textblocks and so i wrote the little test-code
above. Please keep in mind: My problem is NOT, that the code above is slow. (I know, this loop is totally senseless...it's for demonstration purposes only) My problem IS, that changing the font impacts measurement in such an enormous amount...
What is the difference between fonts like "Segoe UI", "Tahoma", "Lucida Sans" or "Microsoft Sans Serif" and fonts like "Arial", "Times New Roman" or "Courier New" that causes this huge
impact in measurement?
Btw.: This problem not only arises within my own grid component, it can also be reproduced with WPF's internal datagrid. Scrolling performance degrades dramatically when using "Arial" as the font-family.

Hi Max
I realise that WPF doesn't use Win32, but I suspected something similar (i.e. the font is being loaded and unloaded everytime a measure takes place rather than using a cached font). I just looked up TextBlock in ILSpy and can see a MeasureOverride method
which does quite a lot of work with the font before using it to measure text and then discards all of this information when it's finished. That's obviously where the bottleneck is. I don't program in WPF so I have no idea how to fix it.
protected sealed override Size MeasureOverride(Size constraint)
this.VerifyReentrancy();
this._textBlockCache = null;
this.EnsureTextBlockCache();
Follow the path of EnsureTextBlockCache() to see how much information is being processed and then dumped every time a measure operation is processed.
Mick Doherty
http://dotnetrix.co.uk
http://glassui.codeplex.com

How to Measure performance in Interaction centre

Dear Experts
Pls tell me whether can we measure IC agent performance based on calls attended by them in interaction center.
If yes, how?
Regards
Rajat

Dear Gurinder
There are various ways of findout out IC agent's performance in a call center.
There are some standard reports which are given by BI through which the performance can be measures
Some of these reports are
Abandonment Rate
Average Time to Abandonment
Volume of Connections
Transfers
Service Level
Average Response Time
Average Handling Time
The above reports can help find out out not only the agent's performance but also the performance of the entire process.
Let me give you an example of how these reports can figure out Agent's performance
- Let's take an excample of Handling time. Handling time in a generic way is defined as the time taken to talk to a customer.
Handling time is more often then not one of the most important metrics of running a call center.
Let's assume that the avegare handling time of a call center is 5 minutes but the average handling time of a particular agent is 8 minutes then this gives you an idea of the performance of this agent.
Another metric could be First call resoloution. This means how ofetn the call is resolved in first interaction itself.
There is a particular way to feed data to BI to get the analysis of the above mentioned reports. This way is that data come from the CTI provider gets loaded in CRM in some tables through a function module and then extracted by BI for this analysis.
Another way of conducting analysis is Blended analytics. Please read about it.
Thanks
Tarang

Are there any JCOP functionality to measure performance within card ?

Hello,
We are trying to measure some functions performances. Bu we dont have any specific tool for that.
We made some tests with PC applications, but it is not precise(as you know there are some scard library delays).
Is there any internal card(jcop) functionality to estimate the time interval for our card functionalities, or do you you have any other suggestions ?
Thank you in advance.

Hi,
Given a card does not have a real time clock it is not really aware of time in this way and adding a measurement would affect the performance as well (note there is some timing in the crypto chip to help protect against certain attacks, but this is not exposed to the JC API).
The best you can do it measure the time before and after transmitting an APDU at the lowest level possible. It is safe enough to assume that the PC/SC overhead is going to be consistent for the same amount of data. You could even benchmark just sending data to the card without doing anything with it to measure the time for transmitting your data. This is not generally needed as you are interested in the performance as perceived by a client application or a user and this transmission overhead is a part of that as well.
Just my $0.02 anyway.
Cheers,
Shane

Measuring performance in OSB

I would like to know if their is a way to get the performance metrics in OSB 11g (I dont have the entire SOA suite !). Is their a way to measure how much time is spent on a) proxy b) business services and also if possible in running the Transformation? Any other general performance measurement strategy would also help.

A very distinctive feature of OSB is its capability for service monitoring ..OSB can provide different levels of service monitoring stats..Plrase go through the below link:
http://download-llnw.oracle.com/docs/cd/E13159_01/osb/docs10gr3/operations/monitoring.html
You can access the statistical information for a service through the Oracle Service Bus Console or directly by using the JMX monitoring APIs

Measuring performance of EJBs

Similar Messages

Maybe you are looking for