The latency of paliHandleGet is currently tracked in granularity of millis. This is not very valuable for us because when we report to T2, values are often reported as 0, 1 or 2ms (see image below from a test cluster). I believe this is because we cast to an integer when we output it FTDC.

<img width="1657" height="36" alt="image" src="https://github.com/user-attachments/assets/5af83060-8596-4452-8c7f-93dfba81046c" />

Changing our granularity to micros gives us better visibility to what is occurring with our reads