Size of a Client in HANA

[Former Member]

So I'm familiar with how to get the size of a client via RSSPACECHECK or by running a test client copy; But when the system is using a database that utilizes columnar tables, like HANA, I have yet to find a good way to see how much space a client actually takes up.

For example, you have 3 clients.  In each client's tables, there are unique values per column that exist in all clients. And since a columnar table only stores unique values once in each column, the size of all three columns is only the addition of difference between the clients plus the unique values (counted only once).  So if you were to run the RSSPACECHECK report, it counts the expanded size of each client and the total of the three would add up to more than what HANA is actually storing.

I'm sure I could sit down and write up a query to figure this out but I wanted to see if anyone else has done this already?  If not, I'll share once I have it figured out.

[lbreddemann · Accepted solution]

While you could "de-fiddle" subsets of the table contents to assign to separate clients this way, I'd say there's little actual value to this approach. This is because separating the data sets would implicitly change the compression that SAP HANA applies to the single columns, thus leading to different actual space requirements.

In the end this would lead to over-counting, since it each client has it's own range of distinct values and those likely will be overlapping.

So, what I would propose for this is to apply a form of fixed costs accounting.

One simple approach would be:

1) find the share of data volume measured in records each client takes up in the specific table.

2) take the total space requirement for the table and distribute it to the clients based on their share from step 1).

This would provide a reasonable accounting for an as-is analysis.

Obviously, this would only be so useful when planning for a system copy (maybe to another platform) since then you cannot count on the same compression factors.

my 2 cts

[Former Member · Accepted solution]

You may use SQL: "HANA_Data_Clients_CommandGenerator" (SAP Note 1969700). This tool basically implements the algorithm proposed by Lars. Be aware that the command generates another command in the first step. This generated command has to be executed in the second step.

[Former Member · Accepted solution]

Perfect!  That is exactly what I was looking for... thank you!

[Former Member · Accepted solution]

Lars, I didn't want to leave your reply unnoticed and I wanted to explain the value to me.  We are in the middle of several big projects and we are setup in a N+1 landscape, using one project landscape for all projects.  We already have some projects finished so I have clients in the project landscape that will give me some idea of what "a set" of clients for a project size would be.  Being able to calculate the size of that set allows me to tell the business whether or not we can support their project or if we need to add resources or do we need to build another landscape to support the project.

I do understand that this is not exact, but sizing and performance analysis never are, they more of an art than a science.

Thank you for your input, it is much appreciated.