[Jasont]

We are currently experiencing slow performance using OTM and have been told that the OTM screens are very heavy and the size of screens is a chief cause as they are too large. Further to this, we have also been advised the way the screens handle javascript and the frame on pop up detail screens is also very heavy.

As our implementation is worldwide, we are seeing huge performance issues in both Asia, as well as external partners.

I would like to solicit feedback from the OTMFAQ community on any experiences and potential changes and solutions people have deployed to address screen performance and whether or not the claim of screen sizes and the way OTM handles javascript is in fact an issue and has any worked around this issue.

regards

j

[nick.polites]

This issue does occur a lot escpecially in Asia-Pacific. OTM is not dial-up friendly as some of the screens do get between 50-500k. You can modify some of your screensets to reduce how much data is transferred but in most cases won't get it down to really make an impact.

[Jasont]

Thanks Nick,

From your experiences has compression and caching helped at all? If so, what benefits have you seen or anyone else seen by using compression and caching.

Being a non-techie, I am trying to pin point where the underlying issues are and how these may be addressed. We are going through an exhaustive testing process, but it is a long winded process and really need to identify quick wins and leverage people's experience to date.

Therefore if yourself or anyone else out there could provide any more insights as to what the major cause of size has been, e.g is it the way the data is handled, or is the page construction and javascript, would be great Providing specifics would be beneficial so we can go back to Oracle with a very clear and concise view.

Thanks for your feeback

[acuartero]

You might want to consider the use of Citrix to host the sessions at a central site and present the sessions to remote users. The bandwidth requirements of a typical Citrix session is around 18-20Kbps and is well suited for dial-up. You should be able to host a significant number of users on each server. This would also give you the ability to shadow users for troubleshooting, maintain browser standards (and settings) and otherwise provide a secure hassle free user experience.

-Alan

[chrisplough]

I've also seen good success with products like the Riverbed line of accelerators. If the users are within a single location, then pairing these up can work. If the users are distributed, but not too numerous then their SoftWOC solution can work. Otherwise, I'd check to see the network provider you're using - and which your users are using.

For our Hosting, we've done a lot of testing with various providers. They were chosen based on their ability to provide lower-latency connection to as much as the globe as possible. At the same time, rural China and India still remain a challenge, regardless of where your datacenter is located and what provider your using.

I do agree with Nick that anything that can be done to reduce the elements on the screen, or the number of returned results will help - and comes at no cost, other than config and testing time.

Hope this helps!

--Chris

[acuartero]

Chris,

Riverbed looks like an interesting solution. Have you done any benchmark tests using it with OTM that you would care to share?

Thanks,
Alan

[chrisplough]

Alan,

Unfortunately, I don't have any empirical results to share. While the solution seemed to work well, the company we were working with decided to hold off on this phase of their OTM roll-out. I do know that Riverbed provides eval units, so always possible to test before making a capital investment.

--Chris