Yor are right. It's some kind of queue mechanism. Right now i am working i
telco company (We used to send sms)
Users will be inserting records into send_sms @ 30,000msg/min Then those
record will be updated and moved to alt_send_sms and deleted from send_sms.
After that 30,000msg/min will be taken out from alt_send for processing and
sending to client. All the above task are happening concurrently. We will be
dealing with million of records/hour
On Fri, Aug 21, 2009 at 6:11 PM, walter harms <wharms@stripped> wrote:
> Krishna Chandra Prajapati schrieb:
> > Hi list,
> > I have two tables send_sms and alt_send_sms. Users are inserting records
> > into send_sms @ 500/sec ie 30000/min. After applying some updates to
> > send_sms data are transferred to alt_send_sms and deleted from send sms.
> > same thing is happening with alt_send_sms table.
> > Is it possible to insert 1000records/sec in send_sms table and taken out
> > the rate 1000records/seconds from alt_send_sms.
> > Which engine is more better for the above senario.
> Hi Krishna,
> i see you are using some kind of queue mechanism but
> to get a useful answer you need to be more specific:
> e.g. what are your safety requirements ?
After moving the data to next stage Data is deleted from current table.
> Tables in RAM are very fast.
> e.g. do you need forgein keys ?
> When will data be copied (send->alt) ?
> after 1 day ? 1 hour ?
> how long to you need to store data at alt ?
Min 1 minute (we need to process immeaditely and send to the users as sms)
> how often is the access ?
> If speed is a concern do you need a database at all ? (KISS)
> where does the current system spend its time ? and why ?
> You see your request is far from simple and demands detail knowlegde about
> your requirements going beyound what can be done in such a ML
> (and this is only software, there is also hardware an economics).
> Here you can ask "how can i improve SQL statement XX ?"