In practice, this maximum number of subscribers is influenced by traffic patterns and utilization ratio between components, which varies between deployments based on customer business models. As the number of users or associated traffic grow, depending on which component utilization is in question, optimizations must be done punctually to address the scalability of that particular component. As this is structural problem it can only be properly addressed by an architecture change where load can be distributed among multiple machines.

 

As a rule of thumb, beyond 75% utilization of any single component resource, the platform will likely start to display signs of stretch and corrections wherever applied will not fix the issue for long time. The performance of the platform as a whole is likely to depreciate when utilization grows beyond 75%, it is not safe to operate the platform outside this envelope.

 

You should plan for migrating to the Large platform design using SIP Thor when you go beyond this safe envelope.