TM1 Comparison to Essbase

[Thondar]

I am a longtime Essbase user and have now been exposed to a company using TM1. I have read the TM1 manual and used the Server Explorer but none of the other tools (TI, Perspective etc.). I would like to get a better understand the differences in how data is stored and memory used by TM1 as compared to Essbase. I don't currently have access to the TM1 server, so I can not get an understanding of TM1 as I understand Essbase. The separate creation of dimensions from cubes seems to be one of the major differences. Also can you specify the Dense and Sparse dimensions in TM1 as you do in Essbase? The optimization process sounds similar between dense and sparse dimension orders in cubes but I can not tell how a dimension is determined as dense or sparse in TM1. I know this is vague but any input would be appreciated.

Larry

[jim wood]

I am currently working with both so here goes:

Essbase has BSO and ASO. BSO is pre-calculated hard drive based. ASO is memory based but it only loads in to memory what it thinks it needs based on what you tell it. TM1 is completly memory based and calculates on the fly. In all of the cases there are strengths and weaknesses. I would tend to suggest that TM1 is a lot stronger than Essbase for what if. I would however tend to Essbase for weekly KPI reporting on a large scales.

Essbase has many interfaces where as TM1 only really has 2. This means that while TM1 is easier to get to it is not capable of delivering the same quality of report.

Tm1 is more capable of handling multi hierarchy dimensions. You can have multiple hierarchies within dimensions within Essbase but you can't share parents.

Essbase has built in functions for time and is easier to use on this front.

Feeders in TM1 can be hard to understand but on the whole I feel while you can do most things in both calculation engines, TM1 is easier to optimise. (Compared to ASO anyway) For example to make a calculations apply to a specific element set in Essbase (ASO) you have to use IF statements. In TM1 you can just specify what calculations apply to.

TI Vs Rules files: TI has the chief benefit of having the scheduler. It also has a more powerful script language. Rules files have more built in functions for easy use. (For example truncating cubes.)

I hope this helps. I should point out at this point that I have used TM1 more than Essbase and naturally I may lean more towards it. I do however see lots of posative points in Essbase. As always it depends on what you want to use a OLAP tool for,

Jim.

[RobertMilli]

Hi Jim,
I also know quite well Essbase as I've been working with it since 1994. Almost at the same time, I had experiences also with TM1 and today, I'm managing TM1 Center Of Expertise in a french IT company.
Having said that, I can tell you that, IMHO, some of the main differences between Essbase and TM1 are :

• - TM1 works in memory only where Essbase run on disk and use memory as a cache for data blocks and indexes. TM1 is often presented as a "real time calculation engine". It's true because if you type in a data, you don't have to run any process to get a result but at the same time if your result involve calculations of billions of cells it can take a while...
  - Dimensions are shared within cubes for TM1 where that can't be done in Essbase unless you use EIS (Essbase Integration Services) and even so it's not the same level of integration. In TM1, as soon as you change a dimension, every cube which was using this dimension is automatically changed
  - Data entry on consolidated element in TM1 is forbidden where it can be done in Essbase. On the other hand, TM1 provide very powerful and sophisticated spreading method in order to get on a consolidated results by populating all the leaf cells
  - TM1 manage data and calculations separately and it provides very powerfull way to understand a result (Trace calculation) where within Essbase as there is no split between a data and a result you can't know where the figure come from
  - There is no "block storage" concept within TM1, TM1 is much more close to the ASO (Agregate Storage) feature. As you said, changing the dimension order within a cube can affect performance, but as there is no block storage and the optimization process is quite different between these two technologies
  - There is no use to run calculation in TM1. The calculation rules are applied as soon as you request any cells which needs to be calculated (like if every calculation was tagged as Dynamic Calc). In Essbase, you can have many calc scripts (one for budget, one for forecast...). With TM1 to manage multiple calculation methods you have to use parameters and tests it within the calculation rules.
  - TM1 is truely multicube where even if this feature exist in Essbase (@XREF for example) it's not as usable as in TM1
  - TM1 provide a much stronger ETL : Turbo Integrator. You can do much more things with TI (Turbo Integrator) than with the Essbase DLR (Data Load Rules) as it provides a scripting language

It's my vision of the differences, it's probably incomplete but I'm conviced it's a good start...
Regards,
Robert

[abetschen]

Hi all,

Most of my experience is with TM1(Application development including cube design, TI processes, rules and websheets). In regards to Essbase the only experience I have is using the Excel addin and retrieve data. I am interested in learning more about Essbase as we are using both Tm1 and Essbase; Essbase is more a corporate application used mostly for reporting; TM1 is uses by Finance groups mostly for planning and reporting. And off course periodically, we are running into discussions why we are using TM1 vs Essbase.....

So the information that was submitted so far on this thread sheds some light on the underlying technology. What about the following topics:

1. TM1 data can be accessed through the following interfaces: a) Excel Addin b)TM1Web or third party web applications (Olap Object, Arcplan,....) c) TM1 EV d) may be there are more interfaces that I don't know about. As far as I know most of these interfaces are read/write except may be TM1 EV but I am not sure. What about Essbase? Is there any other way to access the data other than through the Excel addin "Retrieve"? Is there a web interface available? Is it read/write?

2. TM1 as a work flow module "Planning Manager" (additional purchase required). For example this allows you to setup a work flow for forecasting or budgeting (Dept Mgr enters budget and then submits to the division manager for review, once approved it moves on to the next level until it reaches the end of the work flow and the approved and reviewed data from all the departments are aggregated). Is this a functionality that exists in Essbase?

3. In Tm1 it is easy to build an Excel report or Websheet that pulls data from multiple cubes "side by side" (or however you want to format your report). You can even share common report header drop down if the dimension is the same in both cube or if the values in both dimension are similar. Is this achievable in Essbase Retrieve (Excel)?

Larry, if you have a TM1 Perspectives license (stand alone local server capability) then it could be of great interest to you to first read through the TM1 Rules section of the online help file and then follow some of the exercises about the "Fishcake factory". I believe that the examples highlight quite well the power of the rules and also explain quite well the whole sparsity, skipcheck and feeders concept.

I would like to keep this thread going and collect as much information on both tools.

Thank you in advance.

Andre

[Michel Zijlema]

1. TM1 data can be accessed through the following interfaces: a) Excel Addin b)TM1Web or third party web applications (Olap Object, Arcplan,....) c) TM1 EV d) may be there are more interfaces that I don't know about. As far as I know most of these interfaces are read/write except may be TM1 EV but I am not sure. What about Essbase? Is there any other way to access the data other than through the Excel addin "Retrieve"? Is there a web interface available? Is it read/write?

Hi Andre,

I think Essbase was the first OLAP database that EV was supported on and EV still works on Essbase (as on TM1 and SSAS). Arcplan also connects to Essbase.

Michel

[RobertMilli]

Hi,
It's true, Temtec, the company who created EV, began with a Essbase only version. Then it came to Microsoft Analysis Services (2000 and later on 2005) and finally, due to the fact that Applix bought Temtec, it came to TM1. The TM1 support which was very basic before has become much more mature. For any questions around EV I recommand http://evusers.org which is a forum managed by many Temtec ex employees.

Regarding Andre questions, what I can say is :

• - As for TM1, Essbase can also be accessed by many interfaces : Excel, Hyperion Analyzer (web), EV, ex Brio, Tableau Software... With EV you can access both within the same interface, having a view connected to a TM1 cube and another view connected to an Essbase cube. You can even open a view connected to TM1 and connect it to Essbase (reverse operation can also be done), if the cube structure is not that far, it will show the data along with a warning message that will tell what are the metadata differences
  - I've never headr about a workflow manager for Essbase. The one which came with TM1 wasn't, IHMO, really mature for version prior to 94. For 94, I've not gone enough in deep to have a true opinion on it.
  - An Excel report involving multi cube can also be done with Essbase even if it's not as easy as TM1. This is due to the fact that TM1 is formula driven where Essbase isn't managing the data retrieval in the same way. To do so in Essbase you can have several worksheet, each connected to a dedicated cube, and a synthesis worksheet which grab the figures using formulas, or have a single worksheet with area connected to different cubes but you'll need to selected individually each area and retrieve data.

Hope this help,
Regards,
Robert

[adma76]

hi guys,
how do u describe the necessary programming background to write rules and etl at Essbase/hyperion compare to TM1?
can non programmers can maintain Essbase (new rules, process, reports (design & publish)) ?

[DeeS]

I am trying to compare performance of Essbase against TM1, I am using Windows server with 128 GB or ram and 24 processors for this test. Essbase ASO cube is performing better than TM1 as long as the entire cube is in memory. I am now trying to figure out how to make sure that the Essbase 11.1.2 ASO cube built using AIS with 50 GB of data and aggregations in memory at all times. Anyone has any suggestions?

[ykud]

I am trying to compare performance of Essbase against TM1, I am using Windows server with 128 GB or ram and 24 processors for this test. Essbase ASO cube is performing better than TM1 as long as the entire cube is in memory. I am now trying to figure out how to make sure that the Essbase 11.1.2 ASO cube built using AIS with 50 GB of data and aggregations in memory at all times. Anyone has any suggestions?

With ASO you shouldn't be focused on pushing it in memory, but rather on aggregate design. Try to run aggregate design wizard after you've tracked some query statistics (turn this option on in admin console). It should give you rather 'close-to-fit' precomputed values and whole model will fly up.

In my experience TM1 is a better input\calculation tool than BSO Essbase option, but ASO run circles around when it comes to reporting on billion-rows scale.