Table shown below is:
- searched only [simple select], no deleting/inserting/joining with others etc.
- columns are set / no changes here
- 99%+ of the time searches will go by same column, which is indexed (column: "key" in example)
- engine: MyISAM
- row format: dynamic
- there are 2 other indexes (columns: id/location)
Since this table will be used with every page load, I'm concerned with size vs. speed ratio. Up to how many rows (roughly) will it be very fast, then fast, then slow, then reduced to crawl?
[columns name] | [data type] | [collation] id | int(11) name | varchar(64) | utf8_general_ci key | varchar(64) | utf8_general_ci | [ 99% used for search: is indexed] value | text | utf8_general_ci identifier_id | int(11) sort_order | int(5) last_adjusted | datetime location | varchar(255) | utf8_general_ci group_no | int(3)