nyc;3491334 said:
Our web based financial trading platform averages eight million plus requests a day. I can't imagine filtering quotes especially if using memcached (we don't) can't handle the load. (we use Java though rather than php) Still, it should be a feasible project.
Well, it could be an issue if everything (web server, database, and primary processing is all one server) It could easily be done with a single DB and webserver.
Sorry, off-topic.
memcached is a nice compliment caching system but it can only cache data, it doesn't actually help with processing performance itself. Many netbooks can watch videos, 3D animations, etc. without many issues but try rendering those same videos, 3D animations, etc. on that netbook.
I really did not want to stray off-topic as I have repeated several times now *sigh* however, the problem with filtering posts as you said is that kind of filtering would require on-the-fly filtering .. that has nothing to do with memcached or caching. If it was not user-by-user, it would be feasible because you could process posts, then cache them while only creating processing overhead on the first pass.
To support a user-by-user level filtering of posts that included manual quote checking, it would require on-the-fly processing which means allocating additional CPU/memory resources. Sure you could use memcached to save the processed results but since it would only be seen by one user and unless that one user is going to read not only the same thread but the same page over and over again, any benefits gained would be outweighed considerably by the additional resource usage by memcached itself in storing the data each time.
memcached has its advantages but it's not ground-breaking technology, it's simply a cog or tool among many that compliment other pieces of the puzzle to make things run more smoothly. In many ways, I find memcached extremely limited and in some cases, my own caching systems performed much better than memcached. Still, it can provide a nice performance boost when dealing with previously processed or queried data.
The software I am developing is extremely optimized and I will be avoiding adding features that cannot fit within tight resource requirements. The mistake that most programmers make is assuming just because something can be done, that it should be done. Providing a user-by-user quote filtering system simply because someone cannot take someone else's comments while at the same time increasing processing overhead is a waste of resources especially when there are many ways a user can get around it such as altering the quoted text slightly or using different user accounts.
So .. back to those
new ideas!!
-Reality