Performance issues when processed_messages gets "large"

[bendavies]

Hi,

Loading the dashboard is taking 20-30 seconds for us when processed_messages becomes a few million.
We currently have 6,552,864 rows in that table, so not really that big.

I know this table can be purged, but 6 million rows doesn't seem like enough to warrant that.

query 1 that takes up to 20 seconds:

SELECT DISTINCT
	id_15,
	MIN(sclr_16) AS dctrn_minrownum
FROM
	(
		SELECT
			p0_.run_id AS run_id_0,
			p0_.attempt AS attempt_1,
			p0_.message_type AS message_type_2,
			p0_.description AS description_3,
			p0_.dispatched_at AS dispatched_at_4,
			p0_.received_at AS received_at_5,
			p0_.finished_at AS finished_at_6,
			p0_.wait_time AS wait_time_7,
			p0_.handle_time AS handle_time_8,
			p0_.memory_usage AS memory_usage_9,
			p0_.transport AS transport_10,
			p0_.tags AS tags_11,
			p0_.failure_type AS failure_type_12,
			p0_.failure_message AS failure_message_13,
			p0_.results AS results_14,
			p0_.id AS id_15,
			ROW_NUMBER() OVER (
				ORDER BY
					p0_.finished_at DESC
			) AS sclr_16
		FROM
			processed_messages p0_
	) dctrn_result
GROUP BY
	id_15
ORDER BY
	dctrn_minrownum ASC
LIMIT
	15;

Second query which takes 10 seconds to load the statistics page:

SELECT
	count(DISTINCT p0_.id) AS sclr_0
FROM
	processed_messages p0_
GROUP BY
	p0_.message_type;

I've not looked into improving these yet. just raising for visibility.

Thanks,
Ben