Slow insertion and high cpu usage when inserting huge points

Dear influx developers,

I’ve discovered that if you try to store a bunch of “big points” it is just too slow, and consumes lots of CPU.

My test is something like 4 keys with 4-10 possible values, and then something like thousands of field keys with it’s value.

The reason of doing this is the fact that I cannot put some value as a key (due to high card), and I finally opt to use it as the prefix for field keys.

I cannot put this value as another field, because then I will rewrite the point.

Is there any way to achieve this and avoid high cardinality issues with this approach?

Would someone be so kind to explain me why is it so slow and CPU demanding?
My high cardinality version was high mem demanding and went slow when it just hit the TSI limits.

Best regards,

Hi @Carlos_Tripiana_Mont, have you looked at the hardware sizing guidelines to see if you have the right hardware for such high cardinality?

Yes. My card is so high to eat my hardware if I use this value as key.

But the option of making points with thousands of fields do not consume my hardware. It just use lots of CPU and is slow without any explanation available in the documentation and/or internet.