Hi!
>>>>> "venu" == venu <venu@xxxxxxxxx> writes:
>> Has any thought been given to bulk operations? It's the next logical
>> step for increased performance after placeholders - and relatively
>> straightforward to do. The performance gains can be far more dramatic
>> than with just placeholders because you're eliminating many network
>> round-trips.
<cut>
>> @multitable @columnfractions .30 .70
>> @item Size @tab Comment
>> + @item 2 @tab iterations. Number of times to execute the
>> statement and
>> + corresponding number of repetitions of parameter
>> data in package.
<cut>
venu> This looks like a perfect idea for bulks; and even we can make a bit in
venu> [types_supplied] byte itself to represent the bulk_operation is more
venu> than one or not; so that we can avoid extra 2/4 bytes to represent the
venu> bulk count for normal single execution. As the [types_supplied] can be
venu> represented in a one bit(currently we are using a byte for this); so
venu> this can take another bit; and still we have 6 more free bits here :)
venu> Lets wait for Monty approval for this.
Yes, we can have a bit in the types_supplied to tell if the next 2
bytes is a count of parameters.
We should do this in 5.0; Venu, can you please make a worklog entry
for this.
Regards,
Monty
--
__ ___ ___ ____ __
/ |/ /_ __/ __/ __ \/ / Mr. Michael Widenius <monty@xxxxxxxxx>
/ /|_/ / // /\ \/ /_/ / /__ MySQL AB, CTO
/_/ /_/\_, /___/\___\_\___/ Helsinki, Finland
<___/ www.mysql.com
--
MySQL Internals Mailing List
For list archives: http://lists.mysql.com/internals
To unsubscribe:
http://lists.mysql.com/internals?unsub=gcdmd-internals@xxxxxxxxxxx
|
