Back in 2010, we wrote an article about the pricing problem when selling floating licenses for short duty-cycle applications. We summarized several approaches to address the problem using RLM license keywords “hold=” and “min_checkout=” to tell the RLM License Server to “hold” the licenses some extra amount of time to increase contention for the shared pool of licenses. You can read that article here. Now we have a new capability: caching rlm licenses, which allows you to increase performance even more for short duty-cycle applications.
In this article, we are addressing a different yet related problem: how to minimize the license server’s workload and maximize the speed to checkout licenses of parallel and/or short-duration applications.
To address this issue, starting in v10.1, RLM lets vendors create licenses specifying a cache time in seconds so that when checked out, RLM will create a cached license on the local client machine. If the same user on this host checks out the license before the cache time expires, then the local cached license is used, saving a connection to the license server.
License caching can be especially valuable when your application is run many times over a short period of time, such as a compiler. Caching is also useful when license servers are located several time zones away from the clients.
For a detailed technical description of this feature, please refer to the RLM Reference Manual section entitled “Client-Side License Caching.”