SAP MII PCO Data Store

[emre_bicer]

Hi everyone,

I configured PCO and getting data succesfully. I am storing energy and machine data to database every 10 seconds. But this data is to big and actullay I am thinking there is someting wrong in my scenario.Does it make sense to save this data ?I need a development idea for PCO scenario.

Thank you in advance for your suggestions.

[agentry_src]

Hi Yunus,

What are you using the data for? What kind of machine are you pulling it from? Is the database a data historian or some sort of SQL database? What is your purpose in saving the data (raw?)?

Hard to make suggestions when you have told us so little about your situation.

Regards, Mike

[emre_bicer]

Hi Mike,

For example there are more than 50 different variable for a machine(There are 8 of the same machine) .Customer want to store this data for analyze . But as I say this data is to big . I am storing to hana db .

[Kevin_Hunter]

H Yunus,

As Mike says you have not provided enough information for us to advise properly but in general your scenario is a far too common one.

It sounds like from your description that your client does not know what data they want, how often they want it and why, so have simple said give me everything and as often as possible. This leads to a huge volume of data being stored and potentially never being used. From a technical point of view huge volumes of data is not a problem, but does give performance and cost issues.

Since you are a consultant this is something you need to discuss with your client if they REALLY need the data. "Analyzing" the data is not an answer, you need to drill much deeper to understand what does the data show, how are they going to analyze it, how long do they need to store it, whats the appropriate sample rate ..........

If the requirement turn out to be genuine they your solution architect should have recommended using a separate data lake or a separate database specifically for this data.

In my experience the data is often not required or is being over sampled, for example measure the temperature every 10 seconds of a machine with a large thermal mass is pointless as rate of change for the temperature is pretty slow, therefore measuring it once every 5 minutes may be more appropriate.

The fact that they want to measure 50 parameters from one machine every 10 seconds suggests now thought or design went into coming up with the requirement.

Kevin

[agentry_src]

Hi Yunus,

So this is tag data coming directly from equipment in a factory. It would probably be best to put in a data historian which is really designed for handling that sort of data. HANA is a great database, but it is not tailored for effective use of tag data. Reach out to one or more vendors (OSIsoft, GE-Proficy, Aspentech, Honeywell, etc.) and they will help you understand how to use their tools to both more efficiently store the data as well as the kinds of analysis which would be appropriate for your data.

Cheers, Mike

[emre_bicer]

Hi Mike,Kevin,

Thanks for your comment. Your comments are so precious for me.

I understand that when the data is large, I have to write to the historian. I will define all machines in one PCO as agent and notification. Would a single PCO be enough for all machines?

[Kevin_Hunter]

Hi Yunus,

Its impossible to answer your question without a lot more data, eg server size etc. Please refer to the PCo sizing guide for more information.

https://help.sap.com/doc/21b91997de5d415c88f9b49f2a1190df/15.4.2/en-US/PERF_SIZING_PCO154_final.pdf

In general 8 machines with 50 tags each should be OK on a single instance. We have clients running hundreds of machines on a single instance.

Thanks

Kevin