Your Experience with TM1 10.2.2 and IBM's Support

[nanobaka]

Hi All,

This is not a technical question. For those who have upgraded or tested TM1 10.2.2, would you care to share your experience with this particular version?

My organization has been using TM1 for about 6 years, starting with v9.5. We have created a budget & forecast application with it and we also created a number of reports for our users who access these through TM1Web. We have been testing v10.2.2 since late last year and we finally upgraded to it from v10.1 couple of months ago. The upgrade to 10.2.2 has been painful. While our rules/feeders/TI processes worked fine, TM1Web causes a lot of problems for us. The switch from using IIS to Apache Tomcat "broke" a large portion of our reports and worksheets. For the majority of them I had to manually fix a number the formatting issues while a small number of them required me to recreate the templates from scratch. (I guess the better way to put it is, I spent a lot of time trying to fix a template and failed. When I recreated it from scratch, the problems I had were gone. Hey, as long as I can make it work, right?)

During the testing process we identified a number of issues and we reported it IBM support. At the end we had to wait until FP3 is released to go ahead with the upgrade as it has most of our issues fixed. By FP3 we only have 1 outstanding issue (or so we thought) remained: if you have a report that is wide and long, when you export it to PDF from TM1Web only page 1 will have both column and row headers. Subsequent pages will have either one or both headers missing. We did report it to IBM and was told the fix was too late to make it to FP3 so we had to wait until FP4. We agreed since this is not exactly a show-stopper because we thought snapshot to Excel is working just fine.

Not long after we upgraded to 10.2.2 FP3, a user told us snapshot to Excel on TM1Web is not really working. For columns where values are generated by TM1RPTROW formula, if there is a bunch of rows that have the same values (eg. column 1 may have the same employee names from row 5 to 10 because row 5 to 10 are showing different types of data about the same employee), usually only the 1st one will be shown and the rest will be hidden by setting the font color and background color the same for the cell. In older versions when you snapshot to Excel these 'hidden' values are still in the cells, they are just hidden cosmetically. In 10.2.2 FP3 the values are just not in the cells at all. They all became empty cells. This basically makes the snapshot feature useless. Combine that with the PDF issue, the entire export feature is broken for us. (Users cannot use slice because they can only access TM1 through TM1Web.) This becomes a serious problem as our users use the export feature a lot.

Here is where "IBM's Support" in the subject of this post comes in. I reported this issue to IBM and I spent more than a month trying to convince them this is actually a defect. The support rep I talked to does not seem to think this is an issue. I can reproduce the issue and showed him and he still does not seem convinced. Last week FP4 is released and I saw that the PDF issue is fixed. I gave it a try on our test server and it indeed fixed the PDF issue. The problem is it introduced a new formatting issue that messed up the formatting for most of our templates. I cannot install FP4 on our production servers as is. I reported the issue to IBM and they confirmed it as a defect. I asked for a hot fix asap and they agreed but the wait time is 6-8 weeks. Last time I asked for a hot fix they also promised 6-8 weeks but no one contacted me after 8 weeks. It seems like at this point there is nothing much I can do. (FP4 did not fix the snapshot issue.)

On the other hand, there is no progress on the snapshot issue. There is no update from the IBM support rep at all. I am left hanging with a broken export feature on TM1Web. I tried to contact our organization's IBM account manager and explained the situation and asked if there is any way he can help. It seems that he has no interest in helping at all. I did not even get a reply.

Looking back I definitely regret not doing a more thorough testing. Otherwise I might be able to find the snapshot issue and delayed the upgrade. (It is one of the last things I would expect to break.) This is definitely a tough lesson for me.

For those who have tried 10.2.2, do you see this as a very, very buggy release? What is your experience? They are on FP4 now and it seems like it is still pretty buggy. (To be fair for us almost all the problems are in TM1Web.) Also what is your experience with IBM's support? I have dealt with them since we started using TM1 but this is getting really frustrating. Spending more than a month trying (and still trying) to convince them the issue I am having is a defect feels like death by a thousand cuts.

Sorry for the long rant. I really would love to hear other people's experiences. Especially you experts who have been using TM1 much longer than I do.

Thanks!

[mvaspal]

Also what is your experience with IBM's support?

mine is same as yours
maybe try to buy another 1000 pvus and they will listen to you

[Steve Vincent]

I cannot comment on specifics, but i might be able to bring this up later today with some IBM people who happen to be here. I do know a bit about the bug management process and 6-8 weeks is normal, also reasonable when you understand how that process works. I know that doesn't help us as end users when our models are broken and each subsequent fix pack seems to break something else, we've been hit by that hard recently, but understanding it does help us to be a bit more pragmatic and manage our patches / testing better.

[nanobaka]

I cannot comment on specifics, but i might be able to bring this up later today with some IBM people who happen to be here. I do know a bit about the bug management process and 6-8 weeks is normal, also reasonable when you understand how that process works. I know that doesn't help us as end users when our models are broken and each subsequent fix pack seems to break something else, we've been hit by that hard recently, but understanding it does help us to be a bit more pragmatic and manage our patches / testing better.

I can tolerate 6-8 weeks if they do keep their promise. I only asked for a hot fix once couple of years back. By the end of 8 weeks I did not hear from anyone about it. Obviously I would like to see bugs fixed asap, especially considering export on TM1Web is pretty much broken right now. I do understand fixing bugs takes time and as long as they keep their promise and send me a fix on time, I am ok with that. I am simply dreading that after 8 weeks I would not hear from anyone at all.

What makes me getting so impatient is the fact that snapshot to Excel is broken. I can live with PDF export not working and wait patiently for a fix as long as snapshot works. When I found out snapshot also broke it changes everything.

My biggest issues with this whole mess are:

1. Why is 10.2.2 remains this buggy? I understand that the move from IIS to Apache Tomcat is huge, but we are on FP4 now. To make it worse, FP4 introduces new bugs bad enough that I cannot install it. (To be fair, almost all the problems I have encountered are on the TM1Web side.)

2. IBM's support can be bad at times. I have dealt with a couple of reps who are quite competent. But the one I am dealing with right now with the snapshot issues makes me want to jump off a bridge. As for the account managers and those sales folks? Everytime I tried to talk to one, a new guy will show up and tell me "I'm your new account manager for your organization. Nice to meet you!" It is pretty crazy. Mvaspal couple of posts above is right. Those guys want nothing to do with you unless you are ready to give them more money.

[George Regateiro]

1. Why is 10.2.2 remains this buggy? I understand that the move from IIS to Apache Tomcat is huge, but we are on FP4 now. To make it worse, FP4 introduces new bugs bad enough that I cannot install it. (To be fair, almost all the problems I have encountered are on the TM1Web side.)

I think your biggest issue is the amount of people on the latest version of web. Going to things like Vision and talking to other customers there are many users that are still no where near 10.2.2. Even in my company we went there for the server but not the web. We are testing it out but unfortunately for us and IBM alot of the issues with the web will not be discovered until it is in prod across alot of companies and users have had time to break it. Even in our short time with it alot of the issues have been specific things we did within excel and not really something IBM could test for. Though time will play a factor, adoption is probably one of the biggest hurdles to finding the issue in Java web.

[tomok]

I don't deal with IBM Support that much since my clients expect me to know the answer without calling IBM. However, I have had occasion to call them for a client and have been less than impressed with their knowledge. You have to keep in mind this is often where companies put their new hires (sounds kinda counter-intuitive doesn't it?) so their knowledge of the product is probably limited to common stuff and errors that customers make when using the product. Actual bugs are probably a small percentage of their calls.

TM1Web has ALWAYS been a strange animal and very buggy. The most buggy part of TM1Web, IMO, has always been merging cells, conditional formatting, and exporting. If you have extensive cell merging and/or conditional formatting then expect to have problems in TM1Web at some point. I've yet to see a release that was perfect. 10.1 was pretty good but now that IIS is out and Tomcat is in, I suspect it will take a while to get it all worked out. You may want to dumb down your web reports by limiting the conditional formatting and/or other fancy formatting.

[jim wood]

You have to keep in mind this is often where companies put their new hires

This isn't quite true. When IBM took over they picked up all the Applix support staff but they have now all left as they didn't like working for Big Blue. They do have a few guys that have been there for a couple of years, but not many. They do however put guys in there but that's normally part of the IBM culture of training people on multiple platforms. I would say however that this differs by region as well.

[nanobaka]

I think your biggest issue is the amount of people on the latest version of web. Going to things like Vision and talking to other customers there are many users that are still no where near 10.2.2. Even in my company we went there for the server but not the web. We are testing it out but unfortunately for us and IBM alot of the issues with the web will not be discovered until it is in prod across alot of companies and users have had time to break it. Even in our short time with it alot of the issues have been specific things we did within excel and not really something IBM could test for. Though time will play a factor, adoption is probably one of the biggest hurdles to finding the issue in Java web.

We do not get to go to conferences unless it is local so we know very little about what other folks are doing out there. You do have a point about the rate of adoption. By the time 10.2.2 FP3 was released, 10.2.2 branch has been out for quite a while. I thought it should not be *that* bad. For 9.5 and 10.1, most of the issues we had were resolved by around FP2. That is why it feels like 10.2.2 is particularly bad.

On the other hand, I did not know you can run TM1 server in one version but TM1Web in another version. That works??

[jim wood]

We do not get to go to conferences unless it is local so we know very little about what other folks are doing out there.

That's a shame. A good portion of the IBM development crew including the likes of Mr Corbet are there. It's great way to bypass support and get an answer straight form the horses mouth.

[BrianL]

On the other hand, I did not know you can run TM1 server in one version but TM1Web in another version. That works??

I'm not aware of an official statement, but generally speaking new versions of the TM1 server are backward compatibly with older versions of the clients (TM1Web, Perspectives, etc...). IBM will accept bugs if that doesn't work. The reverse however, is NOT true. You will likely hit errors if using say 10.2 TM1Web with a 10.1 TM1 Server. That's documented at least http://www-01.ibm.com/support/knowledge ... ml?lang=en

[nanobaka]

TM1Web has ALWAYS been a strange animal and very buggy. The most buggy part of TM1Web, IMO, has always been merging cells, conditional formatting, and exporting. If you have extensive cell merging and/or conditional formatting then expect to have problems in TM1Web at some point. I've yet to see a release that was perfect. 10.1 was pretty good but now that IIS is out and Tomcat is in, I suspect it will take a while to get it all worked out. You may want to dumb down your web reports by limiting the conditional formatting and/or other fancy formatting.

TM1Web does have lots of quirks especially with formatting. I worked with it long enough I design reports in a certain way to work around those quirks and avoid getting into messes that are hard to fix. 10.2.2 is so different than 10.1 that I spent weeks fixing our existing reports. Some of them cannot be fixed for whatever reason and I had to start over with a clean slate. They have this template conversion utility but it does not work all that well that I just sucked it up and tested/fixed each template manually.

I can work around problems with merging cells and conditional formatting but I have few options with exporting. It is missing headers (PDF issue that was fixed in FP4) or actual data (snapshot). There is no workaround.

The issue that prevents me from installing FP4 does have to do with formatting. In previous versions (FP3 and prior) if you have a cell you did not set to wrap text and the text in it is too long to fit, TM1Web will happily truncate the text. I am fine with this behavior. In FP4, TM1Web decides to act "smarter" and will auto-wrap the text that is too long. When the text is wrapped, the height of the cell will increase. But it does not work well with Freeze Pane in Excel where it allows you to lock columns and rows (eg. headers) that you do not want to scroll outside the screen. If you have both columns and rows freezed and you have a cell in the freezed zone that has auto-wrapped text, that row's height in the freezed zone will increase, but for the rest of the cells on the same row outside of the freezed zone, their height will not increase. This will cause misalignment for the rest of the report. (eg. header and data are not lining up correctly)

Many of our reports are quite big with many rows and columns, so we very often will lock the header rows and columns to make them more readable especially for people with smaller monitors. This particular bug introduced in FP4 "broke" those reports. I do have the option to unfreeze panes but it will make the reports a PITA to read. I simply do not have a good workaround for this so I had to ask IBM to fix it.

[nanobaka]

I'm not aware of an official statement, but generally speaking new versions of the TM1 server are backward compatibly with older versions of the clients (TM1Web, Perspectives, etc...). IBM will accept bugs if that doesn't work. The reverse however, is NOT true. You will likely hit errors if using say 10.2 TM1Web with a 10.1 TM1 Server. That's documented at least http://www-01.ibm.com/support/knowledge ... ml?lang=en

I know about the backward compatibility with older clients such as Perspectives and Architect, but not TM1Web. I guess I look at TM1Web more as a server than a client. That is good to know.

[George Regateiro]

On the other hand, I did not know you can run TM1 server in one version but TM1Web in another version. That works??

I'm not aware of an official statement, but generally speaking new versions of the TM1 server are backward compatibly with older versions of the clients (TM1Web, Perspectives, etc...). IBM will accept bugs if that doesn't work. The reverse however, is NOT true. You will likely hit errors if using say 10.2 TM1Web with a 10.1 TM1 Server. That's documented at least http://www-01.ibm.com/support/knowledge ... ml?lang=en

There was a statement at one point. I got confirmation that we would not be painting ourselves into a corner. My google-fu is failing me at the moment finding the technote. This was a while ago so their official stance may have changed.

[Alan Kirk]

On the other hand, I did not know you can run TM1 server in one version but TM1Web in another version. That works??

I'm not aware of an official statement, but generally speaking new versions of the TM1 server are backward compatibly with older versions of the clients (TM1Web, Perspectives, etc...). IBM will accept bugs if that doesn't work. The reverse however, is NOT true. You will likely hit errors if using say 10.2 TM1Web with a 10.1 TM1 Server. That's documented at least http://www-01.ibm.com/support/knowledge ... ml?lang=en

There was a statement at one point. I got confirmation that we would not be painting ourselves into a corner. My google-fu is failing me at the moment finding the technote. This was a while ago so their official stance may have changed.

It's interesting; I'd never thought of the Web service (as opposed to the Web interface) as being a client before so much as an adjunct to the server but upon reflection... yeah, why not? After all, people in a non-Windows server environment always had it tacked on in the IIS days, since they needed a Windows server box to run it on. It was never part of their server environment as such. It sucks data out of the server and renders it, so by any reasonable standard it qualifies as a client. That being the case there's no theoretical reason why it shouldn't work.

I'm disappointed as anything that 10.2.2 Web is still a bugfest though. We can't stay on unsupported 9.5.2 (I don't care, others do) for next year's budget and if something as basic as slicing to Excel is borked then we can't go to 10.2.2 Web either. Which is a pity because 10.2.2 web actually renders websheets properly instead of as randomised cr@p (action buttons, links, column widths, etc), and doesn't leave dozens of orphaned Excel sessions scattered all over the place as 9.5.2 Web often does. So while my Apache skills aren't advanced yet, I'd rather have had that than another round of "guess what this will look like when it's uploaded, now go kill off those 20 Excel sessions before anyone starts work".

But if 9.5.2 web is not an option, and 10.2.2 web isn't an option, that only leaves 10.1. A version that I probably have less experience with than any since the 8.3/8.4 series, sigh. Of course, I figured that 10.1 wasn't important as 10.2.2 set the new direction. So much for that.

[mattgoff]

With a few exceptions, I've found everyone at support to be competent and pleasant, but the result has been far from meeting my expectations. This screenshot should illustrate why. I guess at least they have recent "modified" dates, right? Right?

Granted, the first one I graded Sev4, but the issue ended up being so bad/unfixable (10.2.2 doesn't support NTLM, SOX requires integrated login, immovable object meet irresistable force) that I just shut down TM1 Web entirely and forced everyone to use Remote Desktop. TM1 Web has always been a pain for me to administer (what with getting websheets to look right) so maybe IBM did me a favor. I also found a few users who really just needed a couple of static reports, so I wrote a script to dump the data to HTML every hour and deleted their TM1 accounts entirely-- probably not the direction IBM would like.

On the second one, I finally was able to get someone on the dev team on the phone last week, so if FP4 doesn't fix the issue (there's a related PMR, fingers crossed) I'm optimistic he's on it.

Matt

[nanobaka]

I'm disappointed as anything that 10.2.2 Web is still a bugfest though. We can't stay on unsupported 9.5.2 (I don't care, others do) for next year's budget and if something as basic as slicing to Excel is borked then we can't go to 10.2.2 Web either. Which is a pity because 10.2.2 web actually renders websheets properly instead of as randomised cr@p (action buttons, links, column widths, etc), and doesn't leave dozens of orphaned Excel sessions scattered all over the place as 9.5.2 Web often does. So while my Apache skills aren't advanced yet, I'd rather have had that than another round of "guess what this will look like when it's uploaded, now go kill off those 20 Excel sessions before anyone starts work".

But if 9.5.2 web is not an option, and 10.2.2 web isn't an option, that only leaves 10.1. A version that I probably have less experience with than any since the 8.3/8.4 series, sigh. Of course, I figured that 10.1 wasn't important as 10.2.2 set the new direction. So much for that.

I was going to say slice *seems* to work fine in 10.2.2 but I decided to do a test before I open my mouth. It turns out slice has the same issue as snapshot, although I can only reproduce it with my model, but not on the planning sample that comes with TM1. Our users cannot use slice so it is not a big deal for us. Here is a sample template I created from planning sample:

screenshot_01.gif (21.67 KiB) Viewed 16364 times

In the cells I highlighted, when you do a snapshot in 9.5/10.1, these cells will have 'Jan-2003' as values. But you cannot see them because TM1Web hides them by setting the font color and background color the same. Change the colors and they will show up. In 10.2.2, those cells are just empty. I thought this only happens for snapshot and I can see reproduce it with a slice on our production instance. (But not with planning sample. I do not know why.) This drives me insane because our power users download snapshots and put the data through pivot table.

The guy I talked to from IBM still thinks this may be "by design".

There are a lot of things I like about 10.2.2's TM1Web. Like you said, I do not need to go in and kill those orphaned Excel processes anymore. It also supports Google Chrome now which our Mac users are happy about. (They wish it will support Safari.) But these bugs really should be fixed by now.

[mvaspal]

The guy I talked to from IBM still thinks this may be "by design".

This is exactly why I gave up going to IBM Support with bugs like yours. The issue is not with the 6-8 weeks fixing time, I would agree with others. But you get tired of explaining why it is a bug and then you give it up.

[tomok]

In the cells I highlighted, when you do a snapshot in 9.5/10.1, these cells will have 'Jan-2003' as values. But you cannot see them because TM1Web hides them by setting the font color and background color the same. Change the colors and they will show up. In 10.2.2, those cells are just empty. I thought this only happens for snapshot and I can see reproduce it with a slice on our production instance. (But not with planning sample. I do not know why.) This drives me insane because our power users download snapshots and put the data through pivot table.

The guy I talked to from IBM still thinks this may be "by design".

He may be right. The reason you don't see it in Planning Sample is because none of the example reports having an active form with multiple row dimensions (which is where they hide the duplicate elements by changing the cell background and font color. Once you create one like that in Planning Sample you'll get the same result, empty cells. If the issue you are facing is because users want to dump the report into Excel and do a pivot table perhaps you could create a report that isn't based on an active form so they can dump that out. Zero suppression should help make that report not so big. Maybe not the best answer but I doubt you're going to get a bug fix on this because it may well not be a bug in their definition. Maybe an enhancement request instead of a bug report.

[nanobaka]

He may be right. The reason you don't see it in Planning Sample is because none of the example reports having an active form with multiple row dimensions (which is where they hide the duplicate elements by changing the cell background and font color. Once you create one like that in Planning Sample you'll get the same result, empty cells. If the issue you are facing is because users want to dump the report into Excel and do a pivot table perhaps you could create a report that isn't based on an active form so they can dump that out. Zero suppression should help make that report not so big. Maybe not the best answer but I doubt you're going to get a bug fix on this because it may well not be a bug in their definition. Maybe an enhancement request instead of a bug report.

The snapshot issue can be reproduced reliably in both planning sample and our own instance. It is only Slice to Excel that I cannot reproduce in planning sample. In any case, the problem is this is *not* the old behavior. From 9.5 all the way up to 10.1 those duplicate elements are only hidden but values are still in the cells. They basically changed the old behavior that I have expected to something else. I hope they are not telling me "this is a feature, not a bug" or "the old behavior is really a defect and we decided to fix it after all these years."

Creating a separate report as data dump is not a practical option. A lot of our reports are created fully utilizing the flexibility offered by active form. It would be difficult to not possible to recreate them in say, a cube view.

[declanr]

From 9.5 all the way up to 10.1 those duplicate elements are only hidden but values are still in the cells. They basically changed the old behavior that I have expected to something else. I hope they are not telling me "this is a feature, not a bug" or "the old behavior is really a defect and we decided to fix it after all these years."

Creating a separate report as data dump is not a practical option. A lot of our reports are created fully utilizing the flexibility offered by active form. It would be difficult to not possible to recreate them in say, a cube view.

I first looked at this and have to admit I was on the side of IBM with the specific issue in question as I feel that TM1 Web 10.2 is an entirely different beast to 10.1 (and prior) and in my opinion a better one. With the case of snapshots now showing exactly what was rendered in the web as opposed to what was in an excel sheet in the background (which most users should never have had to interact with) I would consider a good thing - they get exactly what they see...

I was about to recommend that you could use conditional formatting on the tm1rptrow which as of 10.2 does in fact allow you to show the repeated elements in an active form; this shows up in excel and the web - so logically it would snapshot out the values... it does NOT (I just tested) - as such I would consider that something that you perhaps could get accepted as a BUG.

But that being said I don't interact with IBM support much and can't speak for what they will or will not consider a bug.