IPlanet Portal V3 - response time anomaly

Hi all - I'm doing some stress & volume testing of iPlanet Portal 3. Clients are extranet (https) - mix of netlet terminal sessions & web users. When number of portal sessions gets to around 430 I'm seeing a consistent & repeatable increase in response times. I'm trying to find the cause. Portal & gateway boxes exhibit minimal CPU activity. I am seeing increases in number of threads (aka lightweight processes as reports by prstat - this is Solaris), but I wouldn't think this should have such a large impact (portal login resp time goes from 3 to 6 secs). JVM sizing seems to be ok - no serious growth, so I'm guessing garbage collection is not cause either. Any thoughts?? Thanks. Steve

If you are looking for a Response Time Monitoring tool there is a free one you can use. [Monitoring Tool|http://www.real-user-monitoring.com]

Similar Messages

Portal Page Response time

Hi All,
Enter the portal page url in IE,
I observed portal page response time using the HTTPWatch, it was taking 6 to 7 mintes to get response .
What are the things to do improve the response time.
Please suggest me possible solutions.
Thnaks,
Venkata Sarvabatla

Too big a topic :). Involves download a profiler (commercial ones like JProbe or OptimizeIt or YourKit or ClearApp or free ones available on a google search) which can profile CPU execution and installing and configuring your JVM to send data to it. Then running the problematic part of your application, the profiler will tell you how long each method call took, how many calls were made, the call graphs etc. Once you know which part of your application is taking time , you would then inspect the code and see what the problem is. Im assuming you have already done your rule of thumb tuning (on VM, OS, DB etc).
Google will give you plenty of examples, the tool you choose will also be important , whether you already have a commercial profiler available to your project , or whether youll need to configure some open source one.
This is my writeup for TPTP (free open source)
http://theworkaholic.blogspot.com/2009/10/profiling-bea-weblogic-portal-apps.html
If you havent profiled/tuned before , i suggest you find someone on your project who has and watch them while they do it. Setting up profilers on app servers and writing tests is somewhat painful the first time around and you can learn much more from someone else.

CCMS Monitoring for Average HTTP response time

Hi guys ,
Is it possible to monitor Av. response time for HTTP(this is based on ST03) using CCMS ?
Thanks in advance
George

hi
As per my understanding you tried to view the responsetime comp as same like in sto3 by ccms, the way as below, RZ20-> SAP CCMS montior templates- >Entiresystem-><sid>->application server -><server>R3 services -> DIalog you can get all the response time as same in the ST03.
else in otherway if you looking for monitoring Java and webdynpro, you can do this by GRMG.chk this for to knw http://help.sap.com/saphelp_nwpi71/helpdata/en/45/4ee37fea5917eae10000000a155369/content.htm|http://help.sap.com/saphelp_nwpi71/helpdata/en/45/4ee37fea5917eae10000000a155369/content.htm]
for configuration,[GRMG-config guide|http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/106a4d88-2a36-2c10-549b-dc549d00b8b9?quicklink=index&overridelayout=true]
chk and update
jansi

Crawling iplanet portal server secured content.

Hi, All,
I am new on the iplanet portal server. Try to come up a solution to crawling
the secured content with a valid user name and password. What this the
authentication mechanism of iplanet portal server keep the user's session?
is iPlanet Portal server using cookie to store the session id or pass it
back and forth as a parameter? Where can I find more information about this?
Any response is appreciated!
Hao Huang

currently there is no testing tool available as a part of the product.

Slow response time for JSP pages under iAS 6.0 SP4

Hi,
I got an application deployed on iplanet app server 6.0 SP4 on solaris
2.8. Using a single kjs engine and lite sessions. kjs memory size is
min 256 and max 256 megs. but verbose:gc shows memory is 98% free.
when i restart the app server, all JSP pages are really rendered fast.
After a while (1 or 2 days), the time to service the same request to
JSP pages is getting much longer (even with JSP pages having only
static content in them). CPU is idle ... It just takes time. KXS log
shows requet is taking like 2-4 seconds instead of about 150 milli
secs when the engine is just restarted.
Now if i call a servlet (which do not dispath to a JSP), the response
time is ok! Memory is ok. It looks like its related to JSP pages
only.
Anyone having an idea what the problem could be? One conig param is
the JSP page cache in iASAT. Default value is 10. What is a correct
setting for production? I have 4 different web app deployed in the
same server instance.
Tanks a lot for your input
Andre Paradis

Andre,
I have found the answer to my problem and perhaps yours. It seems that I18N (internationalization) in SP4 may have a performance bug in it.
My soak tests show that with i18N checked in the iAS Admin Tool, testing the fortune cookie sample application with light load (1 request / sec) resulted in a kxs response time initially of 15ms, however this response time increased by roughly 1% per request (i.e after 100 requests the response time had more than doubled).
Switching I18N off yielded a steady 7ms kxs response time from the fortune cookie application.
I would add that I turned I18N on AFTER the installation procedure.
Is this a known issue in SP4? Is there a patch?
regards,
Owen

Request Summary: Average Response time showing value 0 after every 5 minute

Greeting Experts,
We are facing a strange problem on our portal, we have been assigned with the task of monitoring system performance of portal every one hour. But at the end of every hour, the values it was giving is 0 in the request summary tab(System Administration> Monitoring> Request Summary). After monitoring we found out that the monitoring is resetting after every 5 minutes which means that every 5 minutes the values are showing 0. We need to increase the reset time in request summary so as to evaluate the real performance of the system.
Please advice whether it is a normal case in portal, and whether the threshold can be increased in the portal.
Thanks and Regards,
Sarthak syal

Increase "HungServerRecoverSecs"(default 300 seconds) in your iisproxy.ini.
Jong
"TejoKiran" <[email protected]> wrote:
>
Hi ,
Iam using an Applet (under JRE 1.3.0).
Iam trying to send a Http Request using URL Connection.
It is working fine . However, if the response time of the server is more
than
3 minutes ,
the Response is not coming back to the the Applet and it is in a loop
trying to
read from the Server.
I have observed that the request is sent to the server and it is processed
successfully.
Then , I tried to call this sample JSP from the Applet.
<CODE : delay.jsp>
<%@ page language="java" import="java.util.*"%>
<%
System.out.println("<START>");
out.println((new Date()).toString());
System.out.println((new Date()).toString());
long cTime;
cTime = System.currentTimeMillis();
long dTime = cTime;
while(true) {
long dif = System.currentTimeMillis() - cTime;
if(dif > (15*60*1000))
break;
if((System.currentTimeMillis() - dTime) > (2*60*1000)) {
out.println("2 Minutes");
out.flush();
System.out.println("2 Minutes");
dTime = System.currentTimeMillis();
System.out.println((new Date()).toString());
out.println((new Date()).toString());
System.out.println("<END>");
%>
</CODE>
Because , I was flushing the data after every 2 minutes , it worked fine
This one worked with WebLogic 5.1 when accessing the Weblogic server
directly
. Like
http://IP:7001/ccls/test/applet.jsp (This JSP downloads the Applet under
JRE1.3)

Dramatically increased response times when connected to network

I suddenly noticed that the response time for opening a folder item in my local Portal installation increased when I connected to the network. Why? I want to understand this, since I'm supposed to help a customer with their performance problems.
Further details:
The response time increased from about 2 seconds to about 10 seconds when I connected to the network. In this case, I connected via an ISDN line to my normal dial-up number at work. However, when I connect to my private ISP account (Telenor Online), the response times are unchanged, i.e. still 2 seconds.
One difference between the two access points is the use of a proxy in the first case.
What's the explanation?
Thanks, Erik Hagen
null

Hi,
To your issue, the following blogs would be helpful:
Outlook Performance Troubleshooting including Office 365
How To Troubleshoot Microsoft Exchange Server Latency or Connection
Issues
Thanks,
Jessie

IPlanet portal gateway throws HTMLTranslator caught UnsupportedEncoding exception

iPlanet portal gateway throws HTMLTranslator caught UnsupportedEncoding exception when translating java.io.UnsupportedEncodingException: ibm-1047, exception. This happens when one of the web server that I try to access sends 401 Basic Authentication response. The web server is IBM HTTP Server/V5R3M0.
HTTP/1.0 401 Unauthorized
Accept-Ranges: bytes
Date: Thu, 14 Feb 2002 21:48:06 GMT
Content-Length: 285
Content-Type: text/html; charset=IBM-1047
Expires: Thu, 14 Feb 2002 21:48:06 GMT
Cache-Control: no-cache
Connection: keep-alive
Server: IBM HTTP Server/V5R3M0
Last-Modified: Thu, 14 Feb 2002 21:48:50 GMT
Pragma: no-cache
WWW-Authenticate: Basic realm="ODTSO"
Via: 1.1 casoddc0-2 (NetCache NetApp/5.1R2D9)
2/14/02 1:48:50 PM PST: Thread[Thread-173,5,main]
HTMLTranslator: Translate /cgi-bin/db2www/restricted/slsmkt/ffms/DriverEntry.d2w/input
2/14/02 1:48:50 PM PST: Thread[Thread-173,5,main]
HTMLTranslator caught UnsupportedEncoding exception when translating
java.io.UnsupportedEncodingException: ibm-1047
     at java.lang.Throwable.fillInStackTrace(Native Method)
     at java.lang.Throwable.fillInStackTrace(Compiled Code)
     at java.lang.Throwable.<init>(Compiled Code)
     at java.lang.Exception.<init>(Compiled Code)
     at java.io.IOException.<init>(Compiled Code)
     at java.io.UnsupportedEncodingException.<init>(UnsupportedEncodingException.java:37)
     at sun.io.Converters.getConverterClass(Compiled Code)
     at sun.io.Converters.newConverter(Compiled Code)
     at sun.io.ByteToCharConverter.getConverter(Compiled Code)
     at java.lang.String.getBTCConverter(Compiled Code)
     at java.lang.String.<init>(Compiled Code)
     at java.lang.String.<init>(Compiled Code)
     at com.iplanet.portalserver.gateway.connectionhandler.HTMLTranslator.translateSB(Compiled Code)
     at com.iplanet.portalserver.gateway.connectionhandler.HTMLTranslator.translate(Compiled Code)
     at com.iplanet.portalserver.gateway.connectionhandler.Session.processNextRequest(Compiled Code)
     at com.iplanet.portalserver.gateway.server.HTTPConnectionManager$1.run(Compiled Code)
     at com.iplanet.portalserver.gwutils.ThreadPoolThread.run(Compiled Code)
Is there any workaround for this situation?
Thanks in advance.

Modify iwtPlatform.xml file by including charset value IBM=IBM-1047. May be its accepting only ISO charset. Thats why its generating this error. To modify this XML file access this file as
ipsadmin get component iwtPlatform > somefile.xml
Add the charset value to
<iwt:Att name="iwtPlatform-HTMLcharset>.
After that apply these changes to original file using ipsadmin change comoponent ... command.
Please make sure that backup the original file before modifying.

Management Services Response Time; per Role or per Deployment? Broken?

Management Services, Alert on Response Time, appears to be incorrect.
Can you help me understand why the Response Time charts for each Role in a deployment are identical? is this a bug?
My Testing
I have 3 Roles in my Cloud Service Deployment. A Web Role, a worker Role, and an extra Worker Role for this test.
Web Role: Light load 0 to 10 requests per hour.
Worker Role: Generates massive net work traffic. Constantly making web service calls to 3rd party data services & running a Distributed Azure Data Cache.
Extra Worker Role: No load. I added it to test this problem. It does nothing other than respond to Management Services Alert pings.
I created 3 Alerts to Track response time. One to each of the 3 roles, testing from within the same data center; San Jose to San Jose. Then added 3 more to test from Singapore to San Jose.
The Result (over several Months)
A. The San Jose charts are identical for each role.
B. The Singapore charts are identical for each role.
C. The set of Singapore charts are very similar to the San Jose set of charts. With slight variations likely due to internet latency & extra load at Singapore.
D. (most Important). The latency is directly correlated to the network load created by the Worker Role. When I turn it off or throttle it, Response Time on all boxes drops. When working at "full speed" network response time rises
to 10 - 30 secs. (very bad)
Conclusion
The Network Monitoring Response Time Alert is not working as designed. It does not show the Response Time for that Role. But seems to be displaying the Response Time for the whole deployment.
My Request
Can you assist me to understand if this is "By Design". If my testing is flawed? or what is going on.
My Goal
Alert the Azure dev team to this so that maybe they can fix it to function as I expected it to. (or for them to explain why my expectations are unreasonable.)
Why is this important?
Poor Network Perf is a showstopper bug that is preventing us from moving into production. I may be able to solve this with multiple instances. But as I lack the granularity to track even at the Role level. I'm stuck.

Thank you for your reply.
I forgot to update this with the answer.
It turned out that Azure Portal's Create Alert/monitoring UI has a bug.
The monitoring service is only capable of pining your public endpoints. ie: your Web Roles. Yet it lets you think you can create an Alert to monitor any of the roles within your solution. To make things worse, it shows a chart that suggests it actually is
pinging your worker roles, when it is actually only pinging the public Web Role.
That is why all the Roles in your solution display identical metrics regardless of their load. Even stopping your worker roles will not affect their ping response times.
If your worker role fails, Azure Alerts don't tell you about it.
NB: Using Application Insights to create alerting does not have this issue. But only because it doesn't attempt to monitor your web roles, not because it can alert you to any issues.

How to improve server response time...

we have two solaris5.8 machines with 4G of Ram in each of the them. So they are pretty big. We have round-robine load balancing... We are using loadRunner to monitor the performance of our applicaion.So far our applicaion has not even pasted 50 users load test. If we are performing the load test the site is not usable because it gets extremely slow. So we thought by
increasing the request thread (in both kxs & kjs)and connection pool size, we might increase the performance but it did just the oppsite. AS the number of user increases the response time inscreases as well...Our application framework is based on MVC. What is the best setting for application server for the large number of users....Currently in production our old application supports 200 users at a time(using loadRunner)
we are using oracle 8i database....
any help will be appreciated!!
Thanks

Hi,
we are having problems with server response on Solaris although this is with
an Informix database and we do not have load balancing set up. Our server
has 3GB of RAM. However the same app ran fine on an NT iPlanet installation
where the server only had 512MB RAM.
This has occasioned a lot of hard work trying to pin down the problem. We
are currently investigating the JDKs. JDK 1.3 almost completely eliminates
the problem we had. Using the java profiler we noticed that in JDK 1.2.2
most of the CPU time seemed to be spent in the socket read and socket accept
methods. I've attached the profile output for JDK1.3 and 1.2.2, there is
certainly a marked difference.
I would be really interested to see if you get similar results. I simply
added -Xrunhprof:cpu=samples to the JAVA_ARGS in iasenv.ksh.
Andy
"Mansoor Quraishi" <[email protected]> wrote in message
news:[email protected]..
we have two solaris5.8 machines with 4G of Ram in each of the them. So
they are pretty big. We have round-robine load balancing... We are
using loadRunner to monitor the performance of our applicaion.So far
our applicaion has not even pasted 50 users load test. If we are
performing the load test the site is not usable because it gets
extremely slow. So we thought by
increasing the request thread (in both kxs & kjs)and connection pool
size, we might increase the performance but it did just the oppsite.
AS the number of user increases the response time inscreases as
well...Our application framework is based on MVC. What is the best
setting for application server for the large number of
users....Currently in production our old application supports 200
users at a time(using loadRunner)
we are using oracle 8i database....
any help will be appreciated!!
Thanks
Try our New Web Based Forum at http://softwareforum.sun.com
Includes Access to our Product Knowledge Base![Attachment profile.log.jdk12, see below]
[Attachment profile.log.jdk13, see below]

Response Time of MSON Application

Hello SDN Users,
we want to use the Mobile Sales Online for CRM (MSON) Application for our Sales Representatives. Currently I'm testing it on a BlackBerry 7230 with a T-Mobile Germany connection. On the location I'm testing the antenna indicates 2 active bars. The BlackBerry Browser is in the default setting. I've started the MSON Application and stopped the time between Get link of "My Activities" and the display of the Search Form. The time was 15 seconds. I think this will not be accepted by our users. I've also tested this with the BlackBerry Device simulator on my local PC connecting directly to the Portal server where the Mobile Sales application is
running. The time was 4 seconds. Some other pages I've tested:
Search to Result list:
BlackBerry 10 sec
Simulator 2 sec
Result list to Details
BlackBerry 9 sec
Simulator 2 sec
Questions:
1. Are they any suggestions to reduce the Response Time of this application?
2. Has anyone else implemented MSON in real world enviroments?
3. What where the response times?
Best regards
Gregor Wolf

Hello Gregor,
There are numerous factors that effect the performance of the MSOn application like data volume and the landscape of the WebAS and CRM backend. In this application the WebAS makes intensive call to the CRM backend, its very important to have the WebAS and the CRM backend close together in order to reduce the network latency. The connection type also has great effect on the application performance.
Also, in addition to the landscape, the data volume could also reduce the response time of the application. In the search forms if some condition is given , instead of blank form the response time is much faster. In the new SR3 version of the application, atleast one search criteria has to be specified.
Due to all such variables its difficult to give concrete figures regarding the testing results within SAP.
Kind Regards
Gitin

Slow response time Java 2 (Standard Edition) v1.4.2_13

Hello all, I'm a new user to Java and not a developer or a programmer, but I thought this forum might be the best place to start seeking answers to my questions.
The organization I work for uses a centralized web-based Oracle (v10g) accounting program to monitor it�s worldwide inventory. Our network has very strong firewalls and uses a lot of security software to insure network integrity. Users navigate to a portal website and login to their account where they post transactions and such. When users login and initiate a session we use Microsoft VM to create the virtual environment where we communicate with the accounting program. Microsoft VM works well in our environment and provides quick GUI response times (seconds) as transactions are posted.
Understanding the real possibility that Microsoft VM might go away in the not so distant future we tried using Java 2 (Standard Edition) v1.4.2_13 (build-B06) to create the virtual environment and connect to our accounts. The response time was incredibly slow, it took minutes for the GUI transition from screen to screen or provide any response to input.
As a test I bypassed our network and used a standard commercial ISP (cable) connection and Java 2 (Standard Edition) v1.4.2_13 (build-B06) to create the virtual environment then logged into my account. The GUI response time I experienced using this connection was very fast.
Our internet browser is IE 6 or better, changing to a different browser is not an otion.
My question is: Are the security settings of our network affecting the system response time when using Java to create the virtual environment? Or could it be something else?
Thanks for your help
Gambert

RC4 algorithm is patented by RSA Security and it does NOT included into J2SDK 1.4 (there are Blowfish, DES, DESede, PBEWithMD5AndDES and PBEWithMD5AndTripleDE ciphers in SunJCE provider). You can download a free and JDK 1.4 compliant JCE provider with RC4 (and many other ciphers) from http://www.bouncycastle.org/ But keep in mind you may have to pay some royalty if you are going to use it in the production.

Incorrect Alerts CPU Time and Average Response Time

BUG
My Alert on Web app is consistently behaving incorrectly.
If I set an alert to monitor CPU Time at threshold 1.5 seconds, it's actually set to 1.5 milliseconds even though the graph shows a dotted red line at 1.5 seconds. The alert will trigger at 1.5 milliseconds threshold. So, instead, I have to set threshold
to 1500 seconds - the graph is completely useless showing a red dotted line at ~24 mins but the alert does then go off when CPU time goes above 1500 ms (aka 1.5 seconds).
The original portal alert seems to be fine - this problem is on new portal only. Unaffected by browser type - all browsers do the same thing (Just has to be said before someone asks).
This seems to apply to all metrics using seconds (Average Response Time acts in the same way)

Hii...
Am sure you are much aware " Components Response Time spectrum"
########Analyzing Performance Problems#################
--High wait time: Insufficient number of free work processes
--High roll-wait time: Communication problem with GUI, external system,
or large amount of data requested
--High load and generation time: SAP buffers (Program, CUA or Screen)
too small
--High database request time: CPU/memory bottleneck on database server;
communication problem with database server, expensive SQL statements,
database locks, missing indexes, missing statistics or small database buffer(s)
--High CPU time: Expensive ABAP processing, for example, processing
large tables; inefficient programming
--Processing time more than twice CPU time: CPU bottlenecks
If time aloows look at
Note 0000008963 - Definition of SAP response time/processing time/CPU time
Rgds

POP3 support with iPlanet portal server

Greetings,
We are looking to use a web client (NetMail lite) or thin client (NetMail
lite applet) with iPlanet Portal Server 6.0 sp3 out-of-the-box that will
support POP3. It appears that NetMail lite used to support POP3, but it
appears to only support IMAP now? Is this true?
We are not using iPlanet Messenger Server, we are integrating with existing
POP3 servers. Has anyone found an iPlanet client for this or used a third
party tool for this?

Mike Mahon wrote:
Greetings,
We are looking to use a web client (NetMail lite) or thin client (NetMail
lite applet) with iPlanet Portal Server 6.0 sp3 out-of-the-box that will
support POP3. It appears that NetMail lite used to support POP3, but it
appears to only support IMAP now? Is this true?To my knowledge, NetMail, either the applet or lite) have never supported
pop3.
>
We are not using iPlanet Messenger Server, we are integrating with existing
POP3 servers. Has anyone found an iPlanet client for this or used a third
party tool for this?The manual talkes about writing netlet rules that lets you your pop3 client,
through the netlet, to access your pop3 server.
Kent
"For everything there is a season, and a time for every matter
under heaven: ... a time to love, and a time to hate; a time
for war, and a time for peace."
-- Ecclesiastes 3:1,8

Unable to capture the Citrix network response time using OATS Load testing.

Unable to capture the Citrix network response time using OATS Load testing. Here is the scenario " in our project users logs into Citrix network and select the Hyperion application and does the Transaction and the Clients wants us to simulate the same scenario for load testing. We have scripted starting from Citrix Login and then launching Hyperion application. But the time taken to launch the Hyperion Application from Citrix network has not been captured whereas Hyperion Transaction time have been recorded. Can any help to resolve this issue ASAP?

Hi keerthi,
1. I have pasted the code for the first issue
web
                         .button(
                                   122,
                                   "/web:window[@index='0' or @title='Manage Network Targets - Oracle Communications Order and Service Management - Order and Service Management']/web:document[@index='0' or @name='1824fhkchs_6']/web:form[@id='pt1:_UISform1' or @name='pt1:_UISform1' or @index='0']/web:button[@id='pt1:MA:0:n1:1:pt1:qryId1::search' or @value='Search' or @index='3']")
                         .click();
                    adf
                    .table(
                              "/web:window[@index='0' or @title='Manage Network Targets - Oracle Communications Order and Service Management - Order and Service Management']/web:document[@index='0' or @name='1c9nk1ryzv_6']/web:ADFTable[@absoluteLocator='pt1:MA:n1:pt1:pnlcltn:resId1']")
                    .columnSort("Ascending", "Name" );
     }

IPlanet Portal V3 - response time anomaly

Similar Messages

Maybe you are looking for