Philippe Poelvoorde wrote:
Hi,
I changed a few columns, bidsize and asksize are integer now, and i changed ticknumber to smallint unsigned.
At first I used the ticknumbers by the feedserver, now I count up to 65,000 and then reset the counter back to 0. I need that additional column to handle multiple ticks within one second.
now I have a row_length of 41 instead of 61 on the "geldbrief"-table, but there is still just way to much memory-usage
quelle is the stock-exchange ("source");
So normally a contract is traded on a principal exchange, not two, I would eventually suggest doing that :
Hi Philippe,
That might be right if you only watch one country, but we currently have 35 international exchanges. So we don't have just one major exchange for IBM but we have NYSE, LSE, Xetra (german) etc.
It looks like we'll try something disc-based and have a memory-database only with ask- and asksize for the "important" exchanges. We want to get rid of our old solution and it seems like the massive amount of data just doesn't fit into memory with mysql because of the overhead mysql has.
The idea of introducing a numeric code instead of the char(12)-symbols and have a translation-table might be interesting. It makes everything a little less comfortable but saves a few bytes... I could split the data in individual tables for each exchange... It would save another 2 bytes for the exchange-ID... but the applications will have to choose the right table... I guess we'll have some discussion on that here in the company next week.
I don't think that the disc-based table engines can respond quick enough and handle all the inserts at the same time... but I might give it a try next week.
or one table per symbol maybe ?
ehmmm.. >300000 tables? not a good idea ;) I'd split it in tables by the time since that makes cleaning it up much easier (truncate table is much faster than delete from table where datefield<date_sub(now(),interval 5 day))
thanks for your help! Jan
-- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
