5500 specific question on USSD handling upgrade

Question

There is an update for my 5500 available but... I use a prepaid sim card in it so I'm plagued with the dreaded ussd pop up.
Up until the other day the phone would start to ignore the balance messages after a little while which was great. Well the phone has some issues with the music player going and if a text message comes in and you go to read it the phone locks up after a few messsages...
So I have to restart it and the messages usually stop bugging me. Well this last restart... It still shows the message after a few days.
So what does this improved ussd message handling really mean? Does it mean that every one will show no matter what or does it mean that they won't show up at all.
Unfortunatly I haven't found any information on how to permanantly block these messages or get the phone to ignore them. There has to be an add on app to do that cause allot of people have to deal with the annoyance of balance messages.
Any ideas or thoughts on the matter. I have put off upgrading the firmware because I was worried that the messages would show all the time after the uprade.
FYI I am in the US and the 5500 was never marketed here.

Answer 1

Well, it looks to me like you've fallen into the "make the time dimension dense" mistake (which Oracle often helps to propagate - a lot of their documentation says to make the time dimension dense).
Looking at your numbers, I see the following. First off, for every combination of I, R, and S you'll be creating 131 buckets of data (because you have 131 weeks, and you've made the time dimension dense). You say that there are about 11 million IR combos, so I'll take a guess that the number of IRS combos (which defines how many tuples your composite has) is somewhere around 30 million (this is a complete guess - can you tell us exactly how many IRS combos are in your fact table)?
Based on reserving 30 million IRS tuples * 131 dense weeks, your cube is going to size out to be 4 BILLION CELLS LARGE.
Another way to look at this - assume my guess of 30 million IRS tuples is correct. Based on the fact you have about 100 million rows in your fact table, that means that each IRS tuple, on average, only has data for 3 weeks. However, you're reserving space for 131 weeks by making time dense. This not only kills performance from a size standpoint, but ties up the aggregation engine doing (literally) billions of calculations on empty cells.
I'd try again with compressed composites, but put ALL 4 dimensions in the composite. This should make your final cube size roughly 44 times smaller, and make the solve finish roughly 44 times faster (131 / 3).
Hope this helps!
Scott