Hi Alex, I've considered this - but I'm worried about performance in and out of serializing complex data structures as well as additional networking overhead that will be added in this case.
Looking for minimal barrier and minimum performance regression :) Cheers, D On 3 July 2014 09:27, Alexander Malysh <[email protected]> wrote: > Hi Donald, > > this should work but I wonder why you just don't use redis or memcached > because their are designed > for your case... > > Alex > > Am 01.07.2014 um 16:56 schrieb Donald Jackson <[email protected]>: > > > Hi everyone, > > > > I'm not experienced in using shared memory so I wanted to get opinions > from the list. > > > > I use gwlib in some other modules to store hashed data mostly using > dict.(h|c). In order to save memory on certain high use instances I'd like > to potentially look at creating a dict_create_shm() function which will > store the dictionary in a shared memory path. > > > > This would use semaphores to handle locking and contention between > processes. I am wondering if I am missing something here or if this could > be a feasible approach? I'd be interested to hear your experience. > > > > Thanks, > > Donald > > -- Donald Jackson http://www.elite-sms-software.com http://www.ddj.co.za http://www.thearchitech.com donald(a)thearchitech.com
