Reputation: 475
we have a single AWS Aurora RDS instance, and it hit 60% CPU, and our site might get a lot more traffic tomorrow, so I am concerned that it will hit 100%. I would like to scale up the single instance to a better instance class tonight, just in case (we're currently on db.r4.large).
A couple things:
1) If I go into AWS and just edit the instance class, how long will the downtime be as AWS scales it up
2) Do I have to do anything special with my data? Will it lose any data?
3) If I initiate the change, will it scale up immediately or will it wait? I keep seeing stuff about some sort of maintenance window, and if I scale it up, I would like it to scale immediatel y.
This is currently somewhat of an emergency situation. Thanks!
Upvotes: 4
Views: 3914
Reputation: 9422
2 years passed but I will try to answer to complete this question:
- If I go into AWS and just edit the instance class, how long will the downtime be as AWS scales it up
Usually it needs about 5~7 minutes. ScaleUp/Down downtime is pretty close to downtime when you restarting instance. But if you using MultiAZ instance downtime will be less (in my experience failover operation between two Aurora instances in cluster needs about 2 minutes).
- Do I have to do anything special with my data? Will it lose any data?
Not sure if someone knows what happens under the hood.
According to my unconfirmed observations ScaleUp/Down operation doing Stop → Change Instance Type → Start. So all opened connections may be terminated without commit. Already stored data should be fine. Anyway, it will be better to create backup before change instance type (and that is applicable to any emergency situation)
- If I initiate the change, will it scale up immediately or will it wait? I keep seeing stuff about some sort of maintenance window, and if I scale it up, I would like it to scale immediately.
Up to you. Default behaviour is changing when maintenance time. But you can check Apply Immediately
to do modification immediately.
Upvotes: 2