IBM Cognos TM1 V9.4 Scalability

Post Reply
ryan1971
Posts: 4
Joined: Thu Jul 02, 2009 7:59 pm
OLAP Product: tm1
Version: 9.4
Excel Version: 2003

IBM Cognos TM1 V9.4 Scalability

Post by ryan1971 »

All

I've seen some articles that question the scalability of TM1 Version 9.1

http://forums.olapforums.com/viewtopic. ... 94&start=0

As a potential new customer I would like views from customers on this new version 9.4 to see whether they seen any scalabilty issues or experienced slow responses across networks globally.

Thanks

RS
User avatar
jim wood
Site Admin
Posts: 3951
Joined: Wed May 14, 2008 1:51 pm
OLAP Product: TM1
Version: PA 2.0.7
Excel Version: Office 365
Location: 37 East 18th Street New York
Contact:

Re: IBM Cognos TM1 V9.4 Scalability

Post by jim wood »

We found the memory usage a lot higher in 9.1 compared to 8.4.2. (Which is bizzare as they are supposed to be very similar) We haven't moved to 9.4 yet so I can't comment on whether the memory usage is still increasing per version.
Struggling through the quagmire of life to reach the other side of who knows where.
Shop at Amazon
Jimbo PC Builds on YouTube
OS: Mac OS 11 PA Version: 2.0.7
ryan1971
Posts: 4
Joined: Thu Jul 02, 2009 7:59 pm
OLAP Product: tm1
Version: 9.4
Excel Version: 2003

Re: IBM Cognos TM1 V9.4 Scalability

Post by ryan1971 »

Jim

Thanks, does this pose any issues with 64 bit technology e.g slower responses?, I was under the impression that the increase in memory is offset by not having 3GB limit as per 32 bit.

With regards to your memory I take it your cubes have the same design and order?

Thanks

RS
User avatar
Alan Kirk
Site Admin
Posts: 6606
Joined: Sun May 11, 2008 2:30 am
OLAP Product: TM1
Version: PA2.0.9.18 Classic NO PAW!
Excel Version: 2013 and Office 365
Location: Sydney, Australia
Contact:

Re: IBM Cognos TM1 V9.4 Scalability

Post by Alan Kirk »

jim wood wrote:We found the memory usage a lot higher in 9.1 compared to 8.4.2. (Which is bizarre as they are supposed to be very similar)
No, I'm afraid that 9.1 is where the "big bump" (or potentially big bump, depending on model) occurred because of a significant change from 9.0 (and consequently 8.x). There's a warning about it on page 9 of the 9.1 Release Notes:
Memory Usage
In TM1 9.1, the TM1 Server delivers greatly improved concurrency and overall stability, as well as better diagnostics and monitoring capabilities. A side-effect of these necessary changes is that the TM1 Server may consume more memory than in previous versions. The amount of additional memory is primarily model-dependent. Customers who in previous versions have observed that they operate close to the memory limits of their hardware, should anticipate adding additional memory and/or modifying their models using documented TM1 memory optimization techniques. Customers who are currently using 32-bit TM1 Server near the 3 GB limit, may need to upgrade to a 64-bit TM1 Server.
That's why I'm looking at migrating to 9.0 rather than 9.1.
jim wood wrote:We haven't moved to 9.4 yet so I can't comment on whether the memory usage is still increasing per version.
Reputedly (and it's seriously hard to get good comparisons because there are so many variables) you probably won't see as much of a hike in 9.1 to 9.4 as in 9.0 (or earlier) to 9.1. The biggest difference in 9.4 is the use of Unicode, and that should only impact you if you store lots and lots of strings.
"To them, equipment failure is terrifying. To me, it’s 'Tuesday.' "
-----------
Before posting, please check the documentation, the FAQ, the Search function and FOR THE LOVE OF GLUB the Request Guidelines.
User avatar
Alan Kirk
Site Admin
Posts: 6606
Joined: Sun May 11, 2008 2:30 am
OLAP Product: TM1
Version: PA2.0.9.18 Classic NO PAW!
Excel Version: 2013 and Office 365
Location: Sydney, Australia
Contact:

Re: IBM Cognos TM1 V9.4 Scalability

Post by Alan Kirk »

ryan1971 wrote: does this pose any issues with 64 bit technology e.g slower responses?, I was under the impression that the increase in memory is offset by not having 3GB limit as per 32 bit.
Not sure what you mean by "offset by"; using a 64 bit server means that you're no longer limited by the memory limit of 32 bit servers. That means that you can run larger cubes. (Of course, you need more memory to start off with compared to a 32 bit platform but that's just the nature of the 64 bit beast.)

This doesn't DIRECTLY relate to speed, though obviously if you have much more data in your cubes then you can conceivably have more calculations which could in turn slow response times, depending on how much of the data is queried and when. It doesn't mean that one is a direct effect or a direct cause of the other, it's just that as with locking and calculations (the question that David posed in another thread tonight), things can interact.
ryan1971 wrote: With regards to your memory I take it your cubes have the same design and order?
A cube which was originally built even in version 6 can still be loaded up on a version 9.4 server without any changes to its design. However... because 9.4 uses Unicode, it's the one version of TM1 that there's no going back from. Once your cubes have been converted to run with Unicode under 9.4, you can never again load them in an earlier version.

Again, this only affects the cube files, not the dimension order or anything like that.

The only other change that had a similar effect was when the maximum number of dimensions was increased from 16 to 256 in (If memory serves) 8.3. Obviously you can't load a 17 dimension cube in an 8.2 session, not that you'd want to.
"To them, equipment failure is terrifying. To me, it’s 'Tuesday.' "
-----------
Before posting, please check the documentation, the FAQ, the Search function and FOR THE LOVE OF GLUB the Request Guidelines.
lotsaram
MVP
Posts: 3648
Joined: Fri Mar 13, 2009 11:14 am
OLAP Product: TableManager1
Version: PA 2.0.x
Excel Version: Office 365
Location: Switzerland

Re: IBM Cognos TM1 V9.4 Scalability

Post by lotsaram »

This thread is also pretty interesting as is this one in terms of some discussions about TM1 scalability.

Memory pointers are larger in x64 than x86 so a given TM1 model will be substantially larger in x64 than x86. (Between 30% and 100% larger according to the documentation). But the benefit of x64 is that model size is essentially limited only by hardware.

By scalability do you mean size of model or number of users?
ryan1971
Posts: 4
Joined: Thu Jul 02, 2009 7:59 pm
OLAP Product: tm1
Version: 9.4
Excel Version: 2003

Re: IBM Cognos TM1 V9.4 Scalability

Post by ryan1971 »

Guys thanks for your responses should been a bit clearer in the reasons for my questions

We currently have an old TM1 version 8.1.6, we were limited in rolling out a group solution due to 3gb cap in this version on a non 64 bit server

We are now revisiting options and infrastructure and looking at creating 6 dimensional cubes that could contain in the region of 60,000 Org Units, 35,000 Lines (Accounts) , 240 time period, 30 balance types, 120 products, 150 projects on a 64 bit server

This would obviously fall down on previous versions but i wanted to know would 9.4 cope with this and maintain speed when retrieving data and drill down

I would also like to know if you work in larger organisations and have seen any deterioration with larger userbases e.g 500+ & if your users have senn any issues accessing globally

Thanks

RS
Jeroen Eynikel
Community Contributor
Posts: 139
Joined: Mon Sep 15, 2008 1:45 pm

Re: IBM Cognos TM1 V9.4 Scalability

Post by Jeroen Eynikel »

@ RS

I have no experience with TM1 9.4 but I have experience wiuth cubes that are several factors larger (in number of cells anyways) than the cube you mention. So that should not be a problem.

As to deteriorating performance... Depends a lot on what the system is used for, your exact model and ofcourse network bandwith.

- Pre 9.1 at least performance sometimes became problematical if you had a large userbase (100-200+) inputting data at the same time. If it is mostly read operations impact is a lot less.

The largest TM1 installation I have personally worked with had about 250 users, 245 of them just browsing data and the largest cube consumed about 18GB of memory on start up (I believe it had about 16 dimensions with at least two of them containing 200000+ elements). There were some rules on that cube as well but no ultra complex ones. Performance was acceptable with educated users.

Performance was horrible with uneducated users (auto-recalc on for instance is a killer...) Then again I am not aware of a competing product that is going to perform better. If you do have a large userbase doing data input you might want to take a look at combining Cognos Planning with Cognos TM1. Never tried so myself but I guess you could use planning for data input and TM1 for the calculations.

- I have never seen TM1 perform well in limited bandwith environments. Use Citrix/remote desktop/ a second server if there are bandwith issues.
belair22
Posts: 68
Joined: Wed Feb 25, 2009 2:26 am
OLAP Product: TM1, Cognos Express
Version: 9.5 9.4 9.1 9.0 8.4
Excel Version: 2007 2003

Re: IBM Cognos TM1 V9.4 Scalability

Post by belair22 »

Jeroen - you touch on a few interesting points.

The biggest being a single 'uneducated user' is capable of making an entire TM1 system perform like a Excel 97 spreadsheet. I've had users call seeking an explanation as to why there are performance issues at critical times - a quick snapshot/email of TM1 Top might show a single rogue user running something like a full system Version Copy processes mid-day (with a nice queue 'waiting' behind). Naughty Naughty..
User avatar
Martin Ryan
Site Admin
Posts: 1988
Joined: Sat May 10, 2008 9:08 am
OLAP Product: TM1
Version: 10.1
Excel Version: 2010
Location: Wellington, New Zealand
Contact:

Re: IBM Cognos TM1 V9.4 Scalability

Post by Martin Ryan »

My other favourite after auto calc is the F9 button. I'm still surprised at the number of people with years of Excel experience who don't understand/know that Shift+F9 calcs one sheet while F9 calculates every single worksheet in every single workbook you have open.

Had a server lockup for 23 minutes the other day with that one. Tearing my hair out thinking a TI process was looping or something.

Martin
Please do not send technical questions via private message or email. Post them in the forum where you'll probably get a faster reply, and everyone can benefit from the answers.
Jodi Ryan Family Lawyer
lotsaram
MVP
Posts: 3648
Joined: Fri Mar 13, 2009 11:14 am
OLAP Product: TableManager1
Version: PA 2.0.x
Excel Version: Office 365
Location: Switzerland

Re: IBM Cognos TM1 V9.4 Scalability

Post by lotsaram »

ryan1971 wrote:We currently have an old TM1 version 8.1.6, we were limited in rolling out a group solution due to 3gb cap in this version on a non 64 bit server. We are now revisiting options and infrastructure and looking at creating 6 dimensional cubes that could contain in the region of 60,000 Org Units, 35,000 Lines (Accounts) , 240 time period, 30 balance types, 120 products, 150 projects on a 64 bit server

This would obviously fall down on previous versions but i wanted to know would 9.4 cope with this and maintain speed when retrieving data and drill down

I would also like to know if you work in larger organisations and have seen any deterioration with larger user bases e.g 500+ & if your users have seen any issues accessing globally
To try and get this thread back on track ...

Will 9.4 cope with proposed data volume and cube sizes?
Yes. And moreover, probably quite easily.

Will 9.4 maintain speed when retrieving data and drill down?
Yes. As with any TM1, OLAP or BI solution, scope user reporting requirements carefully. You are not talking about large dimensionality for TM1 but it is probably going to be beneficial to have a layer of "summary reporting" cubes for high level reporting. Usually 95% of users only want to look at summary level 95% of the time so it makes sense to have summary cubes tailored for this, no sense making the TM1 server consolidate millions or billions of leaf cells needlessly just so people can look at a summary level report. For the other 5% of the time for the majority or for the other 5% of people who always want the detail then use the full "detailed cube". Yes it's duplication of data (but tiny) but implementing these standard BI design principles can have significant performance benefits for TM1 models as well.

Deterioration with larger user bases? 500+
Possibly, but with 9.4 then any potential issues can be overcome with design.
  • If you are talking about a read only reporting environment or "reporting server" then no problem at all.
  • If you are talking about a "planning server" with lots of concurrent read/write activity then this could potentially cause locking and performance issues. But this can be avoided in 9.4 by using a multi-cube / micro-cube approach for data entry. There has already been lots written on this topic elsewhere on the forum.
  • If you are talking about a "mixed use server" with lots of readers and a small to moderate amount of writers then if the "planning" and "reporting" applications (by "TM1 application" I just mean a related group of cubes and reports) are separate then there shouldn't be any issues with write locks interfering with read activity. If the readers also need to report on the the results from planning then the same "multi-cube" principles apply to take advantage of granular cube locking.
  • The new user based licensing model does offer the ability to split over as many servers as you want. Trickier for planning but very handy for reporting in terms of achieving scalability with no reduction in performance.
Issues accessing globally?
This has always been TM1's Achilles heel. Due to a "chatty protocol" (lots of small packets back and forth between client and server), latency as opposed to bandwidth is the real killer. For 500+ users a single server is fine, but for users accessing from far flung locations the only real option is Citrix or Terminal Server.
User avatar
jim wood
Site Admin
Posts: 3951
Joined: Wed May 14, 2008 1:51 pm
OLAP Product: TM1
Version: PA 2.0.7
Excel Version: Office 365
Location: 37 East 18th Street New York
Contact:

Re: IBM Cognos TM1 V9.4 Scalability

Post by jim wood »

ryan1971 wrote:Jim

Thanks, does this pose any issues with 64 bit technology e.g slower responses?, I was under the impression that the increase in memory is offset by not having 3GB limit as per 32 bit.

With regards to your memory I take it your cubes have the same design and order?

Thanks

RS
We moved from windows 32-bit to Solaris 64-bit within the same version. We noticed that the memory usage under 64-bit is a lot higher and we also got slower response times. We were however able to put the response times down to slower Unix processors.
Struggling through the quagmire of life to reach the other side of who knows where.
Shop at Amazon
Jimbo PC Builds on YouTube
OS: Mac OS 11 PA Version: 2.0.7
Post Reply