Reading articles like this and this, I started wondering if there is a more clear and simplified description on which "high speed and low latency" kinds of workloads LRDIMMs would not be ideal for.
Is it safe to say that these would be on the list for avoiding to use LRDIMMs? :
- A VM/machine running some real-time processing system. (duh... I guess)
- A VM running REDIS (= in-memory storage, used as a distributed, in-memory key–value database, cache and message broker).
- A VM running a firewall.
- A VM running software that reads large volumes of RAM data very regularly to finish a time-consuming task, where reducing the total amount of processing time from, say, 30 days to 25 days is very important.
If someone can tell me if any of the above can be crossed out from this list, or something else added, it would be great. Could all of the above be candidates for elimination from the list because the difference might be negligible? The linked articles seem old anyway and things might have improved since then.
