لطفا منتظر باشید ...
Final Thoughts
We’ve looked at many different approaches to scaling for a spam-filtering solution. There’s much room for further optimization, including performance tuning, intelligent purging, and even the possibility for neural declustering, as we’ll cover later in this book. All of these approaches require the right amount of human resources and skill. Large nationwide networks have these resources to dedicate to implementing high-performance, distributed solutions on the network. For a filter developer, the key to developing a scalable tool is foresight regarding the many different types of scaling these large providers may require. Including support for different types of scaling, such as user-id distribution and peak hour statistics, can make life easier for the systems administrators installing and maintaining the software on their network.
