Hi all, I'm testing using a HEAP table to maintain transient state in a high volume TP environment. So far, it looks promising, and I want to ask anyone about similar experiences and solicit some input.
Some background: * I have one table; the row length is ~200 bytes and the table will typically never have more than 10,000 rows (more like 7,000 on average). * The table is accessed via one of two keys. I have an index on each. The EXPLAIN output indicates these are used. * The ratio of access is ~ 7 selects : 1 insert : 1 delete : 1 update. All selects are singleton row selects (using limit 1 even though it may be redundant) or select count(). * The rate of access is ~15,000 - 20,000 / second across ~80 processes. * The Mysql machine is not memory constrained. * The Mysql instance will be on a separate physical machine from the app machines performing the processing (4 app machines) - they all need to share this data. Questions: * Is this feasible? Has anyone done something like this? * What specific items should I tune for the Mysql instance that supports this? * In initial testing, I see the row selects that return all the data taking 3-5 ms where the count selects often take < 1 ms. Is this about what to expect? Is the delay in the row select due to data size? * Are there any gotchas I should be aware of? I already understand the transient nature of HEAP tables, so anything beyond that. Thanks for your help. Your feedback may help me justify this approach. Regards, Rick --------------------------------------------------------------------- Before posting, please check: http://www.mysql.com/manual.php (the manual) http://lists.mysql.com/ (the list archive) To request this thread, e-mail <[EMAIL PROTECTED]> To unsubscribe, e-mail <[EMAIL PROTECTED]> Trouble unsubscribing? Try: http://lists.mysql.com/php/unsubscribe.php
