TM1 10.2.2 FP6

User avatar
gtonkin
MVP
Posts: 591
Joined: Thu May 06, 2010 3:03 pm
OLAP Product: TM1
Version: PAL 2.0.1
Excel Version: 2016 64-bit
Location: JHB, South Africa
Contact:

TM1 10.2.2 FP6

Post by gtonkin » Wed Jul 13, 2016 5:44 pm

Looks like it is almost available - received a notification this evening.
Release Notes
Critical installation, configuration, and usage issues

Link to download not working, fix central does not contain FP6 for download yet.
TLS1.2 now supported other details scant but guessing updates will happen in due course...

Edit: 2016-07-26
New TM1 Server and Admin Server config parameters to support older clients (incl. BI on Unix) and Transport Layer Security 1.2
List of fixes

Download via this page or
Server-Windows 64-bit
Client-Windows 64-bit
Last edited by gtonkin on Tue Aug 02, 2016 5:59 pm, edited 4 times in total.

User avatar
qml
MVP
Posts: 1054
Joined: Mon Feb 01, 2010 1:01 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2007 - 2016
Location: London, UK, Europe

Re: TM1 10.2.2 FP6

Post by qml » Thu Jul 14, 2016 3:17 pm

That's a nice little change introduced in FP6...

TM1 10.2.2 Fix Pack 6 - Using multi-threaded query to execute a view as a TurboIntegrator data source
IBM Support wrote:
Abstract

As of TM1 10.2.2 Fix Pack 6 (FP6), multi-threaded query (MTQ) is used by default to execute a view extract that is the data source for a TurboIntegrator process.

Content

In previous versions of TM1, you had to execute a ViewConstruct in TurboIntegrator (sometimes in an entirely separate process) to take advantage of MTQ when a cube view was the TurboIntegrator data source.

As of TM1 10.2.2 FP6, this is no longer the case. By default, MTQ is used directly on the execution of a cube view data source, provided the following conditions are met:

1. The MTQ parameter must be enabled in Tm1s.cfg and set to a value of 2 or greater. By default, this parameter is not enabled in the Tm1s.cfg file for TM1 10.2.2 FP6 For details on the MTQ parameter, see http://www.ibm.com/support/knowledgecen ... t_mtq.html.

2. When defining the view extract for the TurboIntegrator data source, the Skip Consolidated Values option must be disabled and the Skip Zero/Blank Values must be enabled.

If you do not want to use MTQ when executing a view data source, you can disable MTQ in the TurboIntegrator process by adding the following function to the Prolog of the process:

DisableMTQViewConstruct( );

This function must appear in the Prolog procedure, it has no effect in any other procedure within a process.
I also like the new TI function CubeDimensionCountGet and others might like the fact that TM1RunTI can now execute chores.
Kamil Arendt

BariAbdul
Regular Participant
Posts: 414
Joined: Sat Mar 10, 2012 1:03 pm
OLAP Product: TM1,Cognos
Version: 9.5.2
Excel Version: 2007

Re: TM1 10.2.2 FP6

Post by BariAbdul » Fri Jul 15, 2016 4:03 am

I am interested to know more about CubeDimensionCountGet ,It seems the link is broken.Could find anything on google either.Thanks.
"You Never Fail Until You Stop Trying......"

User avatar
Michel Zijlema
Site Admin
Posts: 700
Joined: Wed May 14, 2008 5:22 am
OLAP Product: TM1, PALO
Version: both 2.5 and higher
Excel Version: 2003-2007-2010
Location: Netherlands
Contact:

Re: TM1 10.2.2 FP6

Post by Michel Zijlema » Fri Jul 15, 2016 4:47 am

This FP6 seems like a premature e-publication to me :-)
Regarding the MTQ usage I'm wondering how much benefit this bring if you have to include consolidated values in the view extract...

Michel

Andy Key
MVP
Posts: 337
Joined: Wed May 14, 2008 1:37 pm
OLAP Product: TM1 CX
Version: 2-5 to 10-2-x
Excel Version: 2003 2007 2010 2013
Location: Sydney
Contact:

Re: TM1 10.2.2 FP6

Post by Andy Key » Fri Jul 15, 2016 7:34 am

Michel Zijlema wrote:Regarding the MTQ usage I'm wondering how much benefit this bring if you have to include consolidated values in the view extract...
If you set the view up so that you don't have any consolidations in the view definition to start with (why have them there if you are only going to skip them anyway?) then there need not be any extra work being done. Also, once you have done the ViewConstruct and the view is cached, then you can apply SkipConsols further down the Prolog if you still need to.

As for the speed improvements, and with a list of caveats that IBM would be proud of, such as whether the system is idle enough to be able to fully utilise the full number of cores that the MTQ setting allows, the slight overhead of splitting the threads up and combining them again, and possible re-work done on the MTQ threads if there are complex rules on the cube, etc. etc. then you should see a fairly linear improvement in the amount of time the TI takes to evaluate the data source view.

As an example, with MTQ at 7, I'm getting a TI that used to take 120 seconds now taking 20.

If you're not seeing the improvement that you expect, add the stargate loggers to the tm1s-log.properties and check that you are getting all the values that you expect in the stargate definition and adjust the layout of the source view if necessary. Also check that VMM is set to an appropriate value.
Andy Key

User avatar
Michel Zijlema
Site Admin
Posts: 700
Joined: Wed May 14, 2008 5:22 am
OLAP Product: TM1, PALO
Version: both 2.5 and higher
Excel Version: 2003-2007-2010
Location: Netherlands
Contact:

Re: TM1 10.2.2 FP6

Post by Michel Zijlema » Fri Jul 15, 2016 8:29 am

Andy Key wrote:
Michel Zijlema wrote:Regarding the MTQ usage I'm wondering how much benefit this bring if you have to include consolidated values in the view extract...
If you set the view up so that you don't have any consolidations in the view definition to start with (why have them there if you are only going to skip them anyway?) then there need not be any extra work being done. Also, once you have done the ViewConstruct and the view is cached, then you can apply SkipConsols further down the Prolog if you still need to.
Hi Andy,

Creating a view extract using ViewConstruct was already possible before FP6. If you still need to go this way after FP6, than this new feature does not bring much new in my opinion. Indeed you can leverage this feature by only selecting leaf elements on the various dimensions, but the nice thing of the Skip Consolidations feature is that it saves you from quite some coding (which is one of the things that makes TM1 to be considered a technical/IT tool).

Michel

User avatar
qml
MVP
Posts: 1054
Joined: Mon Feb 01, 2010 1:01 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2007 - 2016
Location: London, UK, Europe

Re: TM1 10.2.2 FP6

Post by qml » Fri Jul 15, 2016 9:52 am

Andy Key wrote:If you set the view up so that you don't have any consolidations in the view definition to start with (why have them there if you are only going to skip them anyway?) then there need not be any extra work being done.
This. There would be no benefit of MTQ if there are no consolidations to be performed.
Andy Key wrote:Also, once you have done the ViewConstruct and the view is cached, then you can apply SkipConsols further down the Prolog if you still need to.
The change introduced by FP6 is that there is no longer the need to do a ViewConstruct in Prolog. MTQ will kick in automatically if the source view contains consolidations and is zero-suppressed.
Kamil Arendt

User avatar
qml
MVP
Posts: 1054
Joined: Mon Feb 01, 2010 1:01 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2007 - 2016
Location: London, UK, Europe

Re: TM1 10.2.2 FP6

Post by qml » Fri Jul 15, 2016 10:19 am

BariAbdul wrote:I am interested to know more about CubeDimensionCountGet ,It seems the link is broken.Could find anything on google either.Thanks.
Apologies, the links worked yesterday, but this seems to be a case of someone in IBM pressing the 'publish documentation' button too quickly and then retracting it. I'm hoping the links will work again in a few days.

CubeDimensionCountGet( CubeName ) returns the number of dimensions in a cube, something that up until now was only possible via workarounds like WHILE loop + TABDIM combination.
Kamil Arendt

David Usherwood
Site Admin
Posts: 1318
Joined: Wed May 28, 2008 9:09 am

Re: TM1 10.2.2 FP6

Post by David Usherwood » Fri Jul 15, 2016 11:04 am

This. There would be no benefit of MTQ if there are no consolidations to be performed.
My colleagues went to the Vision session where this was stated by IBM - but we did do some testing when MTQ first came out with a very large, rule based model and achieved a speedup factor of x3 on ruled content at base level. This may no longer be the case for newer releases but I'd be surprised if that were to be the case.

User avatar
qml
MVP
Posts: 1054
Joined: Mon Feb 01, 2010 1:01 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2007 - 2016
Location: London, UK, Europe

Re: TM1 10.2.2 FP6

Post by qml » Fri Jul 15, 2016 11:14 am

David Usherwood wrote:My colleagues went to the Vision session where this was stated by IBM - but we did do some testing when MTQ first came out with a very large, rule based model and achieved a speedup factor of x3 on ruled content at base level. This may no longer be the case for newer releases but I'd be surprised if that were to be the case.
You're absolutely right, MTQ also works for rules (albeit not providing the same performance increase as for natural consolidations). What I meant to say (and probably should have phrased it better) is that MTQ will not increase performance if there is nothing to calculate because the view only contains leaf-level values.
Kamil Arendt

Andy Key
MVP
Posts: 337
Joined: Wed May 14, 2008 1:37 pm
OLAP Product: TM1 CX
Version: 2-5 to 10-2-x
Excel Version: 2003 2007 2010 2013
Location: Sydney
Contact:

Re: TM1 10.2.2 FP6

Post by Andy Key » Sat Jul 16, 2016 9:12 am

Michel Zijlema wrote:Creating a view extract using ViewConstruct was already possible before FP6.
Indeed it was, and that was what I was describing.
Michel Zijlema wrote:If you still need to go this way after FP6, than this new feature does not bring much new in my opinion. [...] but the nice thing of the Skip Consolidations feature is that it saves you from quite some coding
Yes it's not going to help in every situation, but if you are relying on SkipConsols to, well, Skip Consols, but your view contains rules such that it would still benefit from MTQ, then it could be time to do that extra coding... and that is, of course, assuming that the manual ViewConstruct and SkipConsols will still work with the new behaviour of FP6. And given that FP6 seems to have disappeared into the black hole of Fix-Packs-that-were-not-quite-as-ready-to-release-as-IBM-thought, we'll just have to wait and see.
qml wrote:
David Usherwood wrote:My colleagues went to the Vision session where this was stated by IBM - but we did do some testing when MTQ first came out with a very large, rule based model and achieved a speedup factor of x3 on ruled content at base level. This may no longer be the case for newer releases but I'd be surprised if that were to be the case.
You're absolutely right, MTQ also works for rules (albeit not providing the same performance increase as for natural consolidations). What I meant to say (and probably should have phrased it better) is that MTQ will not increase performance if there is nothing to calculate because the view only contains leaf-level values.
The performance improvement that I mentioned was on a view that contained a mix of rules and a few required consolidations.
Andy Key

Edward Stuart
Community Contributor
Posts: 210
Joined: Tue Nov 01, 2011 10:31 am
OLAP Product: TM1
Version: All
Excel Version: All
Location: Manchester
Contact:

Re: TM1 10.2.2 FP6

Post by Edward Stuart » Tue Jul 26, 2016 11:59 am

Appears that this is live again... for now

32 bit servers are deprecated which should at least make the Software Access Catalog easier to navigate

pandinus
Posts: 78
Joined: Tue Mar 18, 2014 8:02 am
OLAP Product: TM1, Cognos Express
Version: 10.2.2
Excel Version: 2013

Re: TM1 10.2.2 FP6

Post by pandinus » Tue Jul 26, 2016 1:38 pm

Being able to use temporary views and subsets as source is a good fix.
Graphs in TM1Web are still buggy though..

nanobaka
Posts: 26
Joined: Tue Jul 26, 2011 4:24 pm
OLAP Product: TM1
Version: 10_2_2 FP5
Excel Version: 2010

Re: TM1 10.2.2 FP6

Post by nanobaka » Thu Jul 28, 2016 5:52 pm

pandinus wrote:Being able to use temporary views and subsets as source is a good fix.
Graphs in TM1Web are still buggy though..
Looking at the fix list I think they only fixed the temporary subsets issue. We still can't use temporary views as data source. If they fixed that too that would be really awesome. I actually asked IBM support about it back then and was told these 2 issues will not be fixed until 10.3...

BrianL
MVP
Posts: 262
Joined: Mon Nov 03, 2014 8:23 pm
OLAP Product: TM1
Version: 9.5.2 10.1 10.2 PA2
Excel Version: 2016

Re: TM1 10.2.2 FP6

Post by BrianL » Thu Jul 28, 2016 8:03 pm

nanobaka wrote:Looking at the fix list I think they only fixed the temporary subsets issue. We still can't use temporary views as data source. If they fixed that too that would be really awesome. I actually asked IBM support about it back then and was told these 2 issues will not be fixed until 10.3...
I don't see it in the FP5 fix list, but I'm pretty sure you can use temporary views as a data source in FP5.

Andy Key
MVP
Posts: 337
Joined: Wed May 14, 2008 1:37 pm
OLAP Product: TM1 CX
Version: 2-5 to 10-2-x
Excel Version: 2003 2007 2010 2013
Location: Sydney
Contact:

Re: TM1 10.2.2 FP6

Post by Andy Key » Fri Jul 29, 2016 3:19 am

nanobaka wrote:We still can't use temporary views as data source.
Are you saying that you have tried them in FP6 and they don't work?

Using temporary views as a datasource was fixed in one of the early FP4 IFs, so even if you didn't get that particular IF, it should have been an early enough fix to have been included in FP5, and you would hope FP6 as well.

I've got systems running on FP4 IF15 and IF19 currently, and temporary views as a datasource work, temporary subsets don't.
Andy Key

nanobaka
Posts: 26
Joined: Tue Jul 26, 2011 4:24 pm
OLAP Product: TM1
Version: 10_2_2 FP5
Excel Version: 2010

Re: TM1 10.2.2 FP6

Post by nanobaka » Sun Jul 31, 2016 2:09 am

I haven't installed FP6 yet. I thought the release notes were up again but not the actual downloads.

I tried temporary views/subsets when they were introduced in FP4 last year. At the time they didn't work. I am running FP5 right now and I didn't know the temporary view issue was already fixed. Last year I was working on a TM1 application that could really use both features. Since they didn't work at the time I went with the usual create/delete method instead. I didn't look at this again until I read the FP6 fixes.

I will give them a try again with FP6 and report back. I'll be working on v2.0 of that application later this year. This will be a great opportunity to make good use of them.

suman23479
Posts: 15
Joined: Fri Jul 29, 2011 6:27 am
OLAP Product: Cognos
Version: 8.4
Excel Version: 12

Re: TM1 10.2.2 FP6

Post by suman23479 » Fri Aug 05, 2016 4:27 pm

All,

After installing the fix pack 6 you will face difficulty to access tm1 architect.
Please read the fix pack release notes and make correction in cognos configuration and .cfg files respectively.

Olaf Franke
Posts: 4
Joined: Mon Jul 07, 2008 8:40 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2010
Location: Germany, Frankfurt/M.
Contact:

Re: TM1 10.2.2 FP6

Post by Olaf Franke » Thu Aug 11, 2016 6:18 pm

All,
on my opinion the FP6 includes now a mojor bug letting crash the server. This is related to the Architect and TI.
Preconditions:
- use function "Export as text data" for a cube (in cube context menu)
- the data set as a result of selected N- and/or C- elements will be empty
- the option "Skip Consolidated Values" is not checked
==> Click "Export" and "Save" and the TM1 Service is gone.

The same thing will happen if you run this in a TI, using the an "empty" View as a datasource and set: ViewExtractSkipCalcsSet (sCubeName, sViewName, 1);
You can check this out by yourself, simply using the SData sample database in local server mode. E.g. try to export the an empty PriceCube view.
The chrash does not depend on used MTQ settings in tm1s.cfg. You can use it or not.

I found a workaround for Turbo Integrator (TI) processes:
- use the new function DisableMTQViewConstruct() in the Prolog if you use a view as a datasource:

Because of this bug it seams to me sensles to use MTQ at this time (after installed FP6).

Best regards,
Olaf

User avatar
qml
MVP
Posts: 1054
Joined: Mon Feb 01, 2010 1:01 pm
OLAP Product: TM1
Version: 10.2.2
Excel Version: 2007 - 2016
Location: London, UK, Europe

Re: TM1 10.2.2 FP6

Post by qml » Fri Aug 12, 2016 10:16 am

Olaf Franke wrote:Because of this bug it seams to me sensles to use MTQ at this time (after installed FP6).
That's probably going too far. Thanks for bringing this defect to our attention - it's a good thing to know. But considering that it is possible to check in the Prolog if your source view will return at least one value, I would rather add a bit of code to do just that and prevent the crash from happening than throw the baby out with the bathwater and not use MTQ entirely. In many applications these days MTQ has become more than just a nice-to-have.
Kamil Arendt

Post Reply