TM1 Forum

Hi experts,

Is it better to have TM1 on the cloud or having the stand alone independent server-client components? There are respective pros-cons but the only thing which I feel a customer would go for TM1 on cloud, is proabaly because of the cost. But I still doubt how much the cost can be reduced by having TM1 on cloud? And more than the hardware configurations one would need to consider when installing TM1 on our servers (RAM, Harddisk space etc), one should be more worried about the security breaches and internet connectivity when connected to the cloud.
Any inputs/suggestions would be helpful. Thanks.

TM1 on the cloud is a new feature which only came in with 10.2 which was released 2 days ago. I doubt anybody has had chance to really asses this yet. You may want to post this question in the 10.2 sticky to see if anybody is running any testing with 10.2 in the cloud,

Jim.

jim wood wrote:TM1 on the cloud is a new feature which only came in with 10.2 which was released 2 days ago. I doubt anybody has had chance to really asses this yet. You may want to post this question in the 10.2 sticky to see if anybody is running any testing with 10.2 in the cloud,

Jim.

To be fair, the "Cloud TM1" you are talking about in 10.2 is IBM's software as a service offering, where they are proposing to host TM1 for companies. Running TM1 in the cloud has been possible for quite some time, I just don't think many companies have attempted it due to security concerns and the fact that Perspectives would likely perform poorly in that environment. To use Perspectives on a cloud-based TM1 server you would have to use it via Citrix or Terminal Services and then make sure the two cloud servers had mega bandwidth with each other. I also am not sure how the IBM licensing would work in that environment. Bottom line is it is possible because I did it with version 9.5.2 and 10.1.1 on Amazon, although it was not for a production instance, just test. FWIW, if you are looking for an OLAP tool that performs extremely well in the cloud then take a look at Jedox.

Good point Tomok, but as you mentioned any WAN performance (Cloud or otherwise) has historically been bad, so I didn't consider it as an option. With improvements to PM and the addition of Cafe IBM are able to start using the "Cloud" word. To put it another way, I wouldn't have considered it before 10.2. I would have gone to your last suggestion... else where.

Spoke to one of the Senior IBM reps ( not a coin operated one ) this morning while on the way to a client and I mentioned the cloud and specifically around TM1 in the cloud. The strategy according to him is TM1 in the cloud is to be seen that TM1 can be deployed in the cloud. This is to protect it against the likes of SAP and host analytics which are increasingly getting a large share of the pie.

Companies run ERP's in the cloud successfully and CFO's are gaining in confidence around security so the issue is performance and performance alone.

nick_leeson wrote:Spoke to one of the Senior IBM reps ( not a coin operated one ) this morning while on the way to a client and I mentioned the cloud and specifically around TM1 in the cloud. The strategy according to him is TM1 in the cloud is to be seen that TM1 can be deployed in the cloud. This is to protect it against the likes of SAP and host analytics which are increasingly getting a large share of the pie.

Companies run ERP's in the cloud successfully and CFO's are gaining in confidence around security so the issue is performance and performance alone.

Yes. Thats exactly the reason one of our customers was mentioning. Most of the ERP's and other SAP related tools are very much successful running in the cloud. So, the main reason to go with TM1 in the cloud was to compete against them and be in the race (successfully)
Thanks for the replies all.

Bear in mind TM1 on cloud is as much a new subscription licensing model as it is anything else. As many have pointed out its been feasible to run on cloud for some time, but IBM hasnt allowed a subscription based pay for what you use SaaS based licensing model. This is really what TM1 on cloud is. Its not a perpetual license, its a pay for what you use monthly subscription model with the ability to scale it up and down on demand. Im sure more details will emerge in good time. But this is great if you want to shift away from a CapEx model to an OpEx model which is often more friendly to the finance department.

nick_leeson wrote:The strategy according to him is TM1 in the cloud is to be seen that TM1 can be deployed in the cloud. This is to protect it against the likes of SAP and host analytics which are increasingly getting a large share of the pie.

I agree with this but only partly. As others have pointed out it is already possible to run TM1\CX in the cloud (We do so ourselves on AWS for demos etc) but TM1 10.2 brings both official support and a SaaS offering. This will help against the likes of Adaptive Planning, Anaplan and Host Analytics who make a big play about being pure cloud.

Bue there are sometimes valid reasons for going cloud, specifically if the organisation has an unresponsive IT department (sometimes internal but often much worse when outsourced). The cost and effort of specifying, obtaining and commissioning a server can be massive and the elapsed time many months. It shouldn't but often is! In this case starting a cloud instance is quicker, easier and cheaper.

One of our clients went to Anaplan recently for a specific app (despite it being functionally weaker and taking longer to implement) purely because their off -shored IT infrastructure made it so difficult to get a server up and running.

stephen waters wrote:One of our clients went to Anaplan recently for a specific app (despite it being functionally weaker and taking longer to implement)

Hush your mouth! You're directly contradicting their main marketing point. Let's hope no one from Anaplan is monitoring this forum. The problem I encountered with running an OLAP tool on the cloud is access to data. When your tool is in the cloud and your data isn't, you have to "dumb down" your ETL process and turn all the data sources into flat files and FTP them up to the cloud server. It works but is not nearly as elegant as what you can do with an ODBC source.

I'm not sure about "have to" - though IBM's Mr TM1 Cloud who took us through the details the other week said that was the preferred method. They don't block a TM1 Cloud instance from going outwards via other ports, so ODBC is not ruled out. You'd need to address latency, and security. Think I'll have a go sometime with AWS - I was quite surprised how more-or-less acceptable Perspectives was against an AWS instance when we tried it the other day. I'm thinking an SSH tunnel would be the way to go.

David Usherwood wrote:I'm not sure about "have to" - though IBM's Mr TM1 Cloud who took us through the details the other week said that was the preferred method.

My comment had nothing to do with the technical requirements of using ODBC versus text file. The reason I said you need to go text is because there is an extremely high likelihood that your ODBC source is behind a firewall, and is not going to be addressable by your cloud TM1 server unless your IT department exposes it. It's technically feasible, just not likely to happen thus the need to use flat files.

tomok wrote:Hush your mouth! You're directly contradicting their main marketing point. Let's hope no one from Anaplan is monitoring this forum.

Those were comments from the TM1 team at our clients. Did I mention the poor reporting of Anaplan as well? Ohh sorry.

If Anaplan are monitoring the forum I would be interested to know how they can justify a lot of the misleading stuff in their TM1 vs Anaplan doc. eg they illustrate the simplicity of Anaplan calcs vs the alleged complexity of TM1 with the following ( See screenshot). I am not a technical TM1 person but I think I could probably calculate "Total Employees Costs = Salaries + Other Employee costs" without writing a TI including a home crafted process logging routine!

@tomok

The reason I said you need to go text is because there is an extremely high likelihood that your ODBC source is behind a firewall, and is not going to be addressable by your cloud TM1 server unless your IT department exposes it. It's technically feasible, just not likely to happen thus the need to use flat files.

Absolutely - but I would suggest that it would be feasible and defensible to run a secured pipe from a TM1 server in the cloud to a database server on premise. I'd want to look hard at latency of course but you could probably tune the ODBC driver to be a bit more relaxed. But (before you point this out to me ) I am presupposing that the political relationship between the TM1 owners and IT is sufficiently constructive to allow that to happen - which will often not be the case. In fact, now I think about it, the number of clients who could pull their data directly from an internal database but IT won't let them and give them a flat file is.... statistically significant. The silliest one I can think of is where the data resides on an iSeries and they pull transactional data into Express cubes and use Reporter - why, oh why??

Tomok/David,

While Anaplan has been mentioned, how does that handle the old connecting ODBC pipe issue? Surely it would have to deal with same problem of connecting to none cloud databases etc?

Jim.

check out this video for how anaplan deals with this ...

http://vimeo.com/37695262

its essentially a process that you run on the client machine which connects to ODBC/JDBC sources and then posts the data into anaplan.

skinners666 wrote:check out this video for how anaplan deals with this ...

http://vimeo.com/37695262

its essentially a process that you run on the client machine which connects to ODBC/JDBC sources and then posts the data into anaplan.

Sounds like a processing worksheet in TM1, they way we used to have to get data into TM1 before they invented TurboIntegrator (yes, I AM that old ). Sounds like it would be pretty slow.

skinners666 wrote:check out this video for how anaplan deals with this ...

http://vimeo.com/37695262

its essentially a process that you run on the client machine which connects to ODBC/JDBC sources and then posts the data into anaplan.

So if the client machine is on a WAN or the DB machine is on the cloud it will run like a dog. In other words it doesn't solve the problem at all, it suffers from same problem. Nuf said for me.