https://foreman.wc1.dfw1.stabletransit.com/reports is much faster.

It uses 'describe SELECT * FROM `reports` ORDER BY reports.created_at DESC LIMIT 0, 20;'

Which i would actually expect to be slower, as it has (in my case) 10x's as many rows to sort.

After investigating this further:

This is all improved by tuning the innodb_buffer_pool_size, but I do have notes for additional improvement.

Adding the index reports(status, created_at) was pointless as the optimizer never used it.

Adding the index reports(created_at) and altering it to:
SELECT id,status FROM `reports` HAVING (status != 0) ORDER BY reports.created_at DESC LIMIT 0, 20;
avoids the filesort and is magnitudes faster. However i am unsure of portability.

WHERE operates before the sort forcing a filesort and HAVING is just a filter afterwards allowing it to use the reports(create_at) index.

Plus the reports(created_at) index would also used by:
SELECT id,status FROM `reports` ORDER BY reports.created_at DESC LIMIT 0, 20;

SELECT count(*) AS count_all FROM `reports`;
Is just slow with innodb, however very dependent on your innodb buffer pool size.

Just ensuring my buffer_pool was large enough to hold at least the entire reports table helped tremendously.