TM1 Forum

Hi,

We are looking to upgrade our TM1 10.2.2 to Planning Analytics, both PAX and PAW. We use Perspectives across Citrix and typically use Excel interfaces to admin data and run TI processes. Some cube work is done, but it is limited due to the fact there is no charting in it; looking forward to PAW on that front.

As part of this upgrade, due to the fact we need to ideally add a new individual server for PAW, we are looking to move both the TM1 server itself as well as the PAW server into Azure. TM1 is currently onsite on a dedicated server.

Are there any recommendations on which particular server types in Azure would be best optimised for TM1?

https://docs.microsoft.com/en-us/azure/ ... dows/sizes

A memory optimised one is most likely, but we are pretty compute heavy too. Also, i know TM1 historically didn't play particularly well with low speed multi-core processors, although newer versions since the introduction of Parallel Interaction have somewhat improved things on that front. Maybe we should be more balanced (General Purpose)? Or is it all down to the particular use case/model?

Any experiences or recommendations would be appreciated.

Before moving to a VM environment, especially a cloud based one, I would strongly suggest you look very carefully at your licensing.

If you are licensed on a PVU basis as soon as you move to a VM, even a local one, you will be entering the murky waters of 'sub capacity licensing' and the IBM License Metric Tool (ILMT)...

https://www-01.ibm.com/software/passpor ... aqgen.html

We are on a Cognos Express license and so I don't think need to worry about the PVUs. Although i've just seen there might be a limit of 1600, but that sounds a lot and so i don't think we need to worry about it. Does that sound right?

Your business partner should be able to help with the licensing questions and may also be able to provide infrastructure for testing your models in a cloud environment.

The memory requirements should be reasonably clear from the onsite server and you should also be able to monitor this server for processor use. The leap between 10.2.2 and 2.0.5 is not as great as to revolutionise your models but there are certain enhancements that can improve overall model/ stability/ running. For instance Multi Threaded Feeding introduced in 2.0.0 may have a significant impact?

Planning Analytics 2.0.5 - June 25, 2018 - Fix Lists, the 'more topics' link on the right should 'hopefully' reference versions 2.0.0, 2.0.1, 2.0.2, 2.0.3 and 2.0.4

The considerations I would have would be around external data sources and latency between onsite systems and cloud infrastructure and you won't really see the impact of these until you test

Hi Ed,

Thanks for this. Multi-Threaded Feeding is not something i had looked at yet! Will definitely look into it now.

Yes, we have a fairly clear idea of the current memory and processor usage, at least in broad strokes, but i was unsure how this would be reflected in the different types of VM setup.

I'll go through your fix list link and see what i can learn from that.

Most of our other systems are already in Azure and so our data transfer processes should speed up considerably once TM1 is in there as well.

Definitely speak with your IBM partner.

As for Express being limited to 1600 PVUs (I think you are right about that by the way) the actual number is irrelevant. In a VM environment you are *required* to run ILMT to monitor your VM isn't using more PVUs than you are licensed for. In the event of an IBM audit, if you are not running ILMT and have records to prove you always have been, you will be required to license the full capacity of the hardware.

That means the hardware your VM is running on, ALL of it. I would imagine that's going to be quite a lot of hardware in the case of Azure!

Try googling "IBM Audit and ILMT".

I would be astounded if IBM support TM1 on any cloud platform other than IBM Cloud/Softlayer anyway.

MGrain wrote: ↑Thu Jul 19, 2018 11:20 am I would be astounded if IBM support TM1 on any cloud platform other than IBM Cloud/Softlayer anyway.

I don't know if the on-premise server license agreement specifically mentions restricting running it on your own hardware versus something in a cloud (other than IBM's) but I would imagine IBM could make it difficult by being less than cooperative around interpreting PVU restrictions on a cloud server. If you are thinking of going cloud I would definitely use IBM's. For the reason already stated but also because of the gateway they have for accessing on-premise ODBC data sources from the IBM cloud. I doubt that gateway would be available on Azure which could be a deal breaker.

Mike, Prepare to be amazed...

https://www-01.ibm.com/software/passpor ... olicy.html

Don't believe ILMT is required in Azure, though there are some mysterious caveats around the cloud being supplied by a third party.

I'm not an IBM license lawyer though...

Hi,

I was told by my IBM representative that I was still required to install ILMT to monitor my installation on a cloud service. I was also sent that very same link as Mike posted.

Basically everything under "Licensing requirements and conditions" means that I still have to install that horrible piece of crap called ILMT. It is actually very sad that IBM hasn't come up with a better solution for this.

Thanks for these insights. Reassuring that IBM at least acknowledges that people might want to stick it in Azure.

I'll look more into the licensing, but on a recent call i had with IBM about the overall upgrade from 10.2 to Planning Analytics I did mention that we would move to Azure at the same time and they didn't raise any issue with that. Licensing was discussed as well as i wanted to confirm what was and was not included coming from CX (PAX is, PAW is only for admin, regular users need additional PAW licenses). They did not mention the ILMT though, but i will follow up with them again on that.

I really wish IBM would allow the logs from Azure, AWS etc to be used instead of ILMT. This would really help out in installing PA.