I am testing a "three-tier" WebFOCUS 7.6.5 environment (though a knowledgeable person told me what I have is a "four-tier" environment.

Tier One: "Web Application Server" - the Web Server (IBM HTTP Server) and the Java Application Server (IBM WebSphere) [considered a separate tier?]

Tier Two: "Application Server" - the WebFOCUS Server.

Tier Three: "Database Server" - IBM DB2.

It seems to me that this environment is a little slower than our current two-tier production WebFOCUS 5.3.4 server.

I plan to do a lot more testing but was wondering if there would be a degradation of speed when the WebFOCUS Server returns a large report to the Web Server - and they're on different machines?

Thanks.

Just to make sure that I understand:

You have the WF client on T1.

EDASERVE (the reporting server) is on T2 and is a hub and has synonyms.

T3 has the native adapters and the 'real' masters.

If yes, that is what we had when I got here 4 years ago and moved away from the hub/sub except for one instance. Data has to do 2 hops and by definition this would be slower.

I think I'm understanding your configuration a little differently than Ginny. Seems similar to what we have. WF Client/IBM HTTP Server/IBM WebSphere on a P5 Linux box. WF Server with all synonyms on a Linux partition of Mainframe. DB2 Client also on that tier. DB2 databases on Mainframe.

I guess I wouldn't know any differently because we were never on a two-tier environment, but our performance seems satisfactory even return data sets of thousands of rows or thousands of pages (as in PDF docs).

If your environment is similar to ours, there shouldn't be any noticieable performance hit. If it's like Ginny describe with a hub-sub where MFD's are on one platform and EDA synonyms are on on main WF server platform, you'll definitely experience some degradation.

And IMO, I would describe your environment as three-tier. Tiers are describing the separate platforms on which the pieces reside, not the individual pieces of software involved. Web server and app server are on the same platform so that isn't a separate tier. You probably already knew all that, but that's how IBI diagrams it.

Darin is correct. I misread.

Darin, Ginny, thanks for the input. I think Darin's interpretation is correct.

Tier One: "Web Application Server" - the Web Server (IBM HTTP Server) and the Java Application Server (IBM WebSphere) [considered a separate tier?] On Windows Server 2003.

Tier Two: "Application Server" - the WebFOCUS 7.6.5 Server. On IBM AIX.

Tier Three: "Database Server" - IBM DB2 v9.1. On IBM AIX.

I'll be doing a lot of comparison testing.

I've never really understood what the "WebFOCUS Client" is - is the "webfocus" java web application?

One more question, I successfully configured Dev Studio to access this environment by simply specifying the Web Server URL - Dev Studio figured out the details of the "Web Component", the "WebFOCUS Client" and the "dataservers". If the Web Server is on one machine and the WebFOCUS Server is on another machine, how do I determine the server name or IP address of the WebFOCUS Server?

Thanks a bunch.

WF Client is kind of the short name for a lot of different stuff. It does contain all th servlets necessary for the web application (using MRE) Also contains the configuration for and communications to the WF Server, and frequently the entire MRE repository (don't know if that applies for Realm driver.) Report Caster "client" is much the same thing but we just call it ReportCaster.

Since you've got the client correctly configured, when you specify the web server in DevStudio, it uses the client's configuration to determine all the rest of that information.

The server name/ip address of the WF Server is the name/ip address of you IBM AIX where it was installed. You simply specify the port number that the WF server is listening on, which you entered when it was installed (or used the default - 8120 on most platforms.) For standalone development from DS, you add that servername/port number to your local server (EDASERVE) as a remote server and you can connect directly to the WF Server Console by entering in a browser the servername or ip address: port number (8121 by default).

For example, say my server is name Taurus and my domain is wcfgroup.com. I would specify taurus.wcfgroup.com as the server name or I could use the actual (internal) IP address of that box. I tend to stick with the server name as defined by DNS. you can ping the server name and it will return the IP address equivalent.

Don't know how much of all that was helpful, but let us know if you have any other questions.

This message has been edited. Last edited by: Darin Lee,

Darin, that was very helpful, just one thing - if I only know the name of the Web Server and can get to the WF Admin console, can I determine the name of the IBM AIX server where the WF server is installed, without asking someone? Would it be available in a config file or via the Admin Console (I haven't found it yet)? The reason I ask is for the standalone development in Dev Studio.

Thanks,

If you have direct access to the web tier, you can look at the odin.cfg file (In 7.1 it's in WebFOCUS71\client71\wfc\etc) Look for the block that contains your WF server name and it will specify the server name (HOST=) and Port.

You can also log in to the WF Admin console (webservername/ibi_apps/console/webfocusconsole.jsp) and look at the WF Servers under "Remote Services." I apologize if some of this is only valid for 7.1. We are not on 7.6 yet, but will be following your example and taking the leap in the near future.

Darin,

Thanks a lot - I'll poke around.

Here's one thing I've found that is a lot slower: After running an HTML report I save it by clicking File > Save As... > Web Page - this takes a very long time - sometimes over a minute, while on the two-tier environment, it's instantaneous.

Interesting - I don't seem to be experiencing that. It seems like that would be completely a local operation (on the PC) since all that info is already in local memory (or swapped to disk). Could be a version difference as well for 7.6. but not a problem on my version (716)

quote:

Originally posted by Francis Mariani:

Here's one thing I've found that is a lot slower: After running an HTML report I save it by clicking File > Save As... > Web Page - this takes a very long time - sometimes over a minute, while on the two-tier environment, it's instantaneous.

Francis,

Sounds like your browser is reissuing the report request. What happens if you unhook from the network before you try a "save as"? Does it take as long or do you get an error?

Cheers

Stu

Hi Francis,

I've worked with WebSphere, IIS and Apache. My experience is that WebSphere needs a very good configuration to work fast. I worked on a project where we had a lot of performance issues with WebSphere and after a reconfiguration, all worked much better.

Performance issues in multi tier environment are network issues most of the times. Are the servers in the same location, same 1 Gb network?

Stu,

I thought it was reissuing the request too, but it takes around the same time for a 18 second or a 4 minute report to be saved. It also doesn't really matter how big the report is.

I'm putting it down to the UAT environment not being configured perfectly (as Frans suggests) and will report it to the support team.

Thank you,

Stu, you're right! It is rerunning the request. I imagine this is a consequence of a three-tier environment.

I don't think that's the reason. We have a three-tier enviornment as explained above and it doesn't re-run the report. Just saves what's there in the browser so it is (more or less) instantaneous.

Darin,

Then it looks like we have a problem.

Thanks,

Sorry I can't be of more help on that one. Try as I might, I cannot get it to re-run a procedure just by saving it.

The only other thing I thought of is trying a different browser like Firefox or Chrome to see if it exhibits the same behavior.

Darin,

Thanks for that good suggestion.

I've tried it and it does not rerun when saving from Firefox 3, but does when saving from Internet Explorer 6. When running a report from the two-tier environment, it does not rerun the report. This could be caused by some cache setting on the IBM HTTP Server in the three-tier environment.

Thanks for your help.