TM1 Forum

Hi everyone,

I haven't gathered a lot of detail on this yet, but for a specific activity like exporting data from a cube (either as a datasource in TI or using Export as ASCII data) TM1 9.4 (MR1) seems to be much more sensitive to cube order. Specifically, if you had a suboptimal dimension order (like a large, sparse dimension closer to the top of the list) we're seeing dramatic increases (orders of magnitude) in the time it takes to do this export of data compared to a more optimal dimension order.

The really odd thing is that we don't see anything at all as dramatic in TM1 9.1.

Anyway, I thought I'd see if people had any similar experiences - quite possibly not that many people are exploring 9.4 that much yet, but it's worth asking.

Regards,
Mike

I have not run into this yet, but I have blocked off time next week to focus on alot of 94 testing. I will try out some of our processes with different dimension orders and post some results.

I've heard through the grapevine that it takes along time to do run a Process Worksheet type function in 9.4 vs 9.1 which is (I think) very similar type fucntionality, don't know any details or if the source cube is sub-optimal.
Cheers

(in response to Steve's comment) In a recent upgrade from 9.0.2 to 9.4 MR1, I found that a flat-file Process Worksheet DID take significantly longer to load data (the cube dim order was optimal). Have not worked with models that might exhibit the symptoms Mike posted.
See my other thread for other recent experience with Process Worksheets.

Steve Rowe wrote:I've heard through the grapevine that it takes along time to do run a Process Worksheet type function in 9.4 vs 9.1 which is (I think) very similar type fucntionality, don't know any details or if the source cube is sub-optimal.
Cheers

One of our clients upgraded to 9.4 from 8.4 because of their move to Office 2007. They carried out after fairly extensive testing and their results were:

1) Most users complained about general "sogginess" of the user interface in 9.4 compared to 8.4.

2) They had 2 or 3 main cubes accessd by end-users. Under 8.4 the cube views and Excel slices of these cubes opened virtually instantaneously after the first viewing. In 9.4 the same views were taking 15-16 seconds to open after repeated viewing of the same data.

3) As long-time TM1 users they had fairly extensive process workbooks for some very specific end-user functions. These were taking 20 (twenty) times longer to load in 9.4 than 8.4 which made them effectively unusable.

The response from Cognos support was effectively:
- Issues 1\2: 9.4 "is doing much more so you expect it to be slower"
- Issue 3:"they are taking 20 times longer but process workbooks are old technology so dont expect them to work well" The 9.1 release notes did mention slower perfomance of process workbooks but 9.4 is even slower than 9.1 and I do not think the 9.4 release notes mention this problem at all.

In this case the client HAD to move to ensure compliance with Office 2007; they ended up reluctantly accepting slower performance for issues 1\2 and then spending several days of our time writing TI to replace the process workbooks. From their point of view this outcome is not ideal and it reflects poorly on TM1.

In respect of process workbooks I do not think it acceptable to degrade perfomance of existing features whch are still supported. Can I encourage other TM1 user who are experiencing similar problems to ensure they are reported to Cognos support and escalated if necessary?

Not to be the bearer of bad news on Process Worksheets, but I found the following statement buried in the 9.4 on-screen help documentation:

“Processing worksheets, while a valid means of importing data, are deprecated functionality in the current TM1 release. We strongly recommend using TurboIntegrator to import data into TM1 cubes.”

I interpreted this to mean that continued use of Process Worksheets in 9.4 is a take your chances" and "Cognos won't help" situation. I rolled the dice already and lost (see separate PW posting). A couple posts here have (IMO) incorrectly stated that PW's are still supported.

Hmmm...
My take on 'Deprecated' is 'still supported, but this will stop soon'. I actually agree with IBM Cognos that TI is far better that PWs for data loading - but unless/until they stop supporting the feature it should work properly. At least we now know it is awful - but we and our clients shouldn't need to find out the hard way.

Maybe they made them crap on purpose, to get more people to use TI? I started using TM1 about 4/5 years ago and i could never get my head around the way those process sheets worked anyway. We used 8.1 and one of my first tasks was to look at the loads and convert them to TI - our admin users were over the moon because it was so much quicker to load their data, and less work for them too. I have yet to find anyone able to convince me that loading data by a process sheet is better in any way than a TI

Onwards and Upwards!

I never realized process worksheets were fast, they've always been slow. hence TI. So slow that we used to have 2 pc's one to run loads with PW's and one to work on. Agree with Steve, in no way would I use a PW over TI.