Yes, thanks, I think we all know what memory-mapped files are! Do you hope to gain:
* Bigger cache sizes by mapping a file bigger than physical memory? * Caches that persist across restarts? * Caches that you can share among concurrently running memcached processes by pointing them at the same file? * Caches that can be directly accessed by client applications by pointing them at the file maintained by memcached? * Caches that can be accessed remotely using a network filesystem? * Administrative benefits (e.g., setting the cache size by changing the file size)? * Performance benefits? Hopefully that list makes it clear that ³What do you hope to gain?² is not an idiotic question. You could be looking for any or all of those things. What feature(s), specifically, do you want to have that memcached as it currently exists doesn¹t provide? Are you looking specifically for one of the above benefits, some combination of them, or something else entirely? -Steve On 7/4/07 7:43 PM, "Joel Poloney" <[EMAIL PROTECTED]> wrote: > ... do I really need to explain this one? I thought memory mapped files were > pretty self explanatory. See wikipedia: > http://en.wikipedia.org/wiki/Memory-mapped_file > <http://en.wikipedia.org/wiki/Memory-mapped_file> . > > Yes, I know there are drawbacks, but if implemented correctly, the benefits > far outweigh them. > > -- Joel > > On 7/5/07, Steve Grimm < [EMAIL PROTECTED] <mailto:[EMAIL PROTECTED]> > > wrote: >> What do you hope to gain by doing that? >> >> -Steve >> >> >> >> On 7/4/07 5:26 AM, "Joel Poloney" <[EMAIL PROTECTED]> wrote: >> >>> Hello list, >>> >>> I was curious if it was possible to extend memcached to memory map files as >>> well. I know there are other memory management systems out there that do >>> this, but I would like to use only 1 memory management system if possible. >>> Memcached already deals with memory... so technically this should be >>> possible... but is it easy to do, I'm not sure. >>> >>> Thanks! >>> >>> -- Joel >>> >> > >
