Yes. They are different. Thanks for the info.
It would also be good to hear from IBM if PAaaS AWS is more performant, less performant, or about the same as PA Cloud or even as some generic, representative on-premises environment.
An analogy I'm working on is that an old-school Mr. Coffee coffee maker can make ten cups of coffee faster than one more expensive, modern, containerized Keurig. But if you share twenty Keurigs with three other businesses, will there be enough coffee for all during the month-end accounting close? It is either an analogy in progress or I'm just thinking about getting more coffee and the one-cup per Keurig option is calling to me since my Mr. Coffee pot is empty. LOL.
Original Message:
Sent: Thu March 21, 2024 06:29 AM
From: Bernd Siebert
Subject: Webinar: Ask Me Anything: IBM Planning Analytics with Watson - March Edition
With IBM's classic SaaS offering PAoC you get dedicated exclusive resources (CPU, RAM, disk), at least for the data tier running TM1 Server and the web tier running TM1 Web. It does not matter if these resources are used or not, they are always there, dormant if not or only partially used.
I think it is very different with the IBM's new containerized SaaS offering PAaaS on AWS:
Regarding how the compute power is allocated or assigned across instances, I assume it boils down to dynamically adjusted to what is currently needed by an instance up to an upper limit which cannot be passed.
Thus you may retrieve the amount of resources (CPU, RAM, disk) used by a container at a specific time stamp, or the minimum and maximum amount within a time period, but there is not really a value you can compare to the dedicated exclusive resources (CPU, RAM, disk) being assigned to IBM's classic SaaS offering PAoC.
PAaaS on AWS is running on top of Red Hat OpenShift on AWS (ROSA).
Here, a container description includes information on the required resources, usually the maximum authorized resources per replica = node aka the resource limits a container may not pass, occasionally the minimum required resources per node = replica aka the requests. It also includes the information how many copies = replicas of a container shall run concurrently, every copy = replica on it's own worker node. Two copies = replicas of the same container shall not share the same worker node.
An Openshift cluster is aware of the free resources ( RAM, millicores = CPU, disk, ... ) per worker node. Depending on this information it deploys the containers it is tasked to run.
I assume the containerized TM1 Server version 12 aka the Planning Analytics Engine (PAE) is working like PAW Distributed on OpenShift, which has a default configuration file ..\kubernetes\defaults.env with entries like
# container cpu limits in millicores
export PA_KUBE_ATLAS_CPU_LIMIT=1000
# container memory limits in Mb
export PA_KUBE_ATLAS_MEMORY_LIMIT=2048
export PA_KUBE_ATLAS_REPLICAS=2
So there is a container named Atlas.
The Atlas container is authorized for a maximum = an upper limit of 2 GB RAM and 1000 millicores = 1 CPU core. This is a maximum load value. It is not what this container is typically using.
2 copies = replicas of the Atlas container shall run on 2 different worker nodes.
Regards
------------------------------
Bernd Siebert
Original Message:
Sent: Wed March 20, 2024 01:06 PM
From: Walter Coffen
Subject: Webinar: Ask Me Anything: IBM Planning Analytics with Watson - March Edition
Mike,
Thanks for doing the AMA every month. I do have a few questions that I'll be asking:
- We get asked the following a lot: What type of CPU / Compute power is used for PA SaaS on AWS (specifically for TM1 Server instances) and how is that compute power allocated or assigned across instances? Customers want to know how it performs versus on-premise or PA Cloud.
- Is there a timeline for PA 2.1? PAE on premises?
- Any updates on the Secure Gateway to Satellite process? How customers are being informed?
Thanks.
------------------------------
Walter Coffen
Technology Manager
QueBIT Consulting, LLC
Original Message:
Sent: Mon January 08, 2024 11:28 AM
From: Michael McGeein
Subject: Webinar: Ask Me Anything: IBM Planning Analytics with Watson - March Edition
Summary
Modeling, budgeting, forecasting, reporting, and multidimensional analysis-there's a lot you can do with IBM Planning Analytics. Chances are you're just scratching the surface.
Join us live for our monthly "Ask me Anything" session designed to answer all of your burning Planning Analytics questions from product roadmap to upgrades, hosted by our product management team.
- Engage with the product management team
- Learn about new features
- Ask questions
Please join us on 21 March 2024, 11:00 AM ET. Please share any questions by clicking on the Reply button. If you have not done so already, register to join here and get your calendar invite.
------------------------------
Michael McGeein
------------------------------