How to design schema for large number of metrics?

Store
,
1

We are having 10000 different metrics which we need to record and they are currently supplied like

<Key_1>.<Key_1_1>.<Key_1_2> = “long value”
<Key_1>.<Key_1_1>.<Key_1_3> = “long value”
<Key_1>.<Key_1_4>.<Key_1_5> = “long value”
<Key_1>.<Key_1_4>.<Key_1_6> = “long value”

<Key_2>.<Key_2_1>.<Key_2_2> = “long value”
<Key_2>.<Key_2_1>.<Key_2_3> = “long value”
<Key_2>.<Key_2_4>.<Key_2_5> = “long value”
<Key_2>.<Key_2_4>.<Key_2_6> = “long value”

Now i can design schema for this in 3 ways.

  1. Use the complete key (<Key_1>.<Key_1_1>.<Key_1_2>) as measurement name. This will explode the number of measurement we have in a database. This is already said to be a bad approach here InfluxDB schema design and data layout | InfluxDB OSS 1.5 Documentation
  2. Split the measurement key by “.” and create record like
    <Key_1>,tag_1=“key_1_1” key_1_2=“Long value”
    This keeps the measurement count to 1/8 (around 1200 measurements) the number we are count of metrics we are capturing.
  3. Split the measurement key by “.” and create record like
    FixedName,tag_1=“Key_1”,tag_2=“key_1_1” key_1_2=“Long value”
    This will keep the measurement count to 1. But will increase the tag count.

What is recommended approach for above use case?

Let me know if you need any other details.

Thanks
Tosheer