@networknerd said in Using In-Memory Databases, Anyone?:
@jimmy9008 said in Using In-Memory Databases, Anyone?:
@networknerd said in Using In-Memory Databases, Anyone?:
- They actually point out in the podcast that they had the best success with in-memory databases on local storage or SAN (NAS not a good choice) and that ethernet was the networking avenue of choice.
Not sure what this means... not that I know about this, but, if its in memory, its in... memory, right? So, what does it have to do with local storage, SAN, or NAS?
Anyway, other databases load what is used frequently in to memory so its quick to access. Its why lots of places would build an SQL server with a ton of RAM - so that SQL can put what is used often in to RAM (fast) and keep it available... right?
Well, at some point you do have to write the data to disk (whether synchronously or asynchronously) to keep a less volatile copy of the data (at least that is what I remember from the podcast). It seems like synchronous data writes would be near impossible because you cannot keep it going at the same speed.
I definitely do not claim to understand it all really well.
Not necessarily. Persistence to disk is not a guaranteed feature of in memory databases. It is common, for sure, but not guaranteed. But you should think of it as taking a backup of the database, not as storing it normally. That's the big difference between the two approaches.