2014-05-15 14:26 GMT+02:00 Antonio Fernández Pérez < antoniofernan...@fabergroup.es>:
> > > > Hi, > > I have in my server database some tables that are too much big and produce > some slow query, even with correct indexes created. > > For my application, it's necessary to have all the data because we make an > authentication process with RADIUS users (AAA protocol) to determine if one > user can or not navigate in Internet (Depending on the time of all his > sessions). > > So, with 8GB of data in one table, what are your advices to follow? > Fragmentation and sharding discarted because we are working with disk > arrays, so not apply. Another option is to delete rows, but in this case, I > can't. For the other hand, maybe de only possible solution is increase the > resources (RAM). Adding more RAM will only save you for a few weeks/months until the data isn't able to fit in memory any longer. You will face the same problem soon (if your data is and will be still growing). There will be a point where you just can't buy more and better hardware (actually you kinda can, but you will spend load of money and might end up with nice servers just doing nothing because they support more memory in their motherboard so you need to upgrade it too). You should give your application a thought and start considering noSQL/table sharding/partitioning/archiving. Maybe it is too late, but before needing another hardware upgrade, yo should've thought about a solution that would allow you keep growing without needing to spend all in hardware (unless you have unlimited money). Good luck! Manuel.
