Fit latModel to actual BlueStore

[carlosmalt]

This pull request is to document fitting the radossim latModel to an actual BlueStore deployment in a Ceph cluster.

[carlosmalt]

Please don't merge this one.

[carlosmalt]

Now that we have K-S test scores the next step is to find an automatic way to fit the model across all sizes. Maybe using pypet?

[carlosmalt]

Excellent. The difference indeed shows it's all positive. I pushed a small change to skip the first 1000 measurements. One concern is the variance in transaction sizes. Can you compute the CDF and average of 10,000 transaction sizes for each request size (again skipping the first 1000)?

[yzhan298]

Excellent. The difference indeed shows it's all positive. I pushed a small change to skip the first 1000 measurements. One concern is the variance in transaction sizes. Can you compute the CDF and average of 10,000 transaction sizes for each request size (again skipping the first 1000)?

Sure, CDF should not be hard. actually for all graphs, I have skipped the first 1000 points, in the read_csv function(skiprows=1000).

[carlosmalt]

... actually for all graphs, I have skipped the first 1000 points, in the read_csv function (skiprows=1000).

Almost all -- check 0d5f9bf.