If you install the Acrolinx Platform on a test system, you can use the Dashboard to analyze the results of a performance test.
To complete a good performance test, you should perform the following tasks:
Check a range of documents that are typical of what your writers would check on a daily basis.
Estimate how often your writing team will probably check their documents and run a batch of checks that reflect this checking volume.
There are two useful performance measurements on the language server pages of the Dashboard. If either of these measurements are unacceptable, you should upgrade your system to support one or more additional language servers.
A "token" is a term that comes from computational linguistics, but most of the time, a token equates to a word. Words make up approximately 80-85% of the tokens so if Acrolinx checks 250 tokens in one second, that measurement would correspond to about 200-215 words per second.
Once you know the number of words processed per second, you can estimate the number of pages that Acrolinx can check per hour and compare this number to your estimated checking volume.
If a user starts a check while all language servers are still processing checks from other users, the core server places the check in a queue. When a language server has completed a check, the core server takes the first check in the queue and sends it to the available language server.
The total queue time is a running total of the amount of time each check spends in the queue. You can divide the total queue time by the total number of checks to get an idea of the average amount of time each user must wait for their check to be processed.
The total amount of time a user must wait for their check to complete consists of:
The queue time (if any).
The time it takes for the Acrolinx Platform to process the content.
The time it takes for the integration to highlight issues in the content.
As well as the performance measurements in the Dashboard, there are also more in-depth profiling tools available that give you more detailed information about your memory usage. You could set up a test server and use a profiling tool to confirm how much memory the server computer needs to check a document.