How HBase responds in milliseconds

[#1]

Recently encountered a tricky issue with HBase 2.4.11
3 VMs, each VM has a total memory of 32 GB allocated to hbase24 GB memory, the read and write of the regionserver account for 9.6 GB respectively, and the listTable table has two fields.
A rowkey (line number) empid, a field elc, a total of 1 million rows of data, use Phoenix to query a piece of data 3 ms, 10,000 pieces of data 109 ms, 100,000 pieces of data 1.62s,
10.5s for millions of pieces of data, whether it is a region or divided into multiple regions, the speed is the same.
If you query data and filter it by conditions, there are millions of data, how can HBase respond to it in milliseconds?

Any help is highly appreciated and thanks in advance.

[#2]

@listne, Welcome to our community! To help you get the best possible answer, I have tagged in our HBase experts @smdas, @PrathapKumar @rki_ @SVB  who may be able to assist you further.

Please feel free to provide any additional information or details about your query, and we hope that you will find a satisfactory solution to your question.