Is master-slave replication my best setup?
I have a big database which writes 1Gb each day. This runs on a low-cost node with just a couple of cores and few Gb memory.
Once or twice per week for a couple of hours, I read from the db and analyse the data.
I wish to pay hourly for the 16vCPU / 64Gb ram node on which to perform these large db reads and computationally intensive analysis.
I seem to require two instances of my database, so as not to have any downtime. (Pausing the write scripts and unmounting the DB volume / mounting it to a high-spec node for 2 hrs is to be avoided.)
My idea is the have master-slave setup, where the master has 100% up-time and it streams to the slave, which is on a volume I unmount weekly for a few hours of analysis.
If the slave is unreachable for a few hours, will this cause problems?
Have I overlooked a superior setup?