-4. Inefficient collection of data. I'm calculating stuff for the
-things I'm not actually displaying, so I should split this so that
-the display part does the calculations when needed and only the raw
-values are collected.
+4. Sometimes I see this:
+
+pstop 0.0.12 - 8:46:41 xxxxxxxxxxxxx / 5.6.21-log, up 1d 22h 59m 16s [ABS]
+Latency by Table Name (table_io_waits_summary_by_table) 239 rows
+Table Name Latency %| Fetch Insert Update Delete
+xxx.xxxxxxxx 5124.10 h +++.+%|100.0%
+xxxxxx.xxxxxxxxxxxx 5124.10 h +++.+%|100.0%
+xxxxxx.xxxxxxx 5124.10 h +++.+%|100.0%
+xxxxxxxx.xxxxxxxxxx 5124.09 h +++.+%|100.0%
+xxxxxxxx.xxxxxxxx 5124.09 h +++.+%|100.0%
+xxxxxx.xxxx 00:07:49 49.2%| 98.6% 0.3% 1.1%
+xxxxxxxxx.xxxxxxxxxxx 00:06:23 40.2%| 97.4% 2.4% 0.1%
+xxxxxxxxx.xxxxxxxxxx 26.25 s 2.8%|114.2% 0.1% +++.+% 0.0%
+xxxxxx.xxxxxxxx 15.71 s 1.6%| 99.6% 0.4%
+
+with a very long latency. Related to doing backups. Need to check the cause
+of the 5124.10h ++ issue. Probably a subtraction problem of some sort.