March 2020, extremely annoying! In my mind the whole point was to be able to be back in business ASAP. Trying to do this is a pain in the ass. Why hasn’t DO addressed this yet is beyond me. I will have to buy my own servers or take my business somewhere else.

Hi

I’m more than aware the read and write rates of SSDs. 35 minutes to copy as many gigabytes is not down to the read or write rate of the SSD it’s down to DO’s infrastructure.

It’s a shame that there’s no automated way of telling the droplet to automatically turn back on once the snapshot is complete. I’m sure it used to turn back on with no option not to have that happen, which is just as bad. Seems like a silly oversight not to let the user decide what happens.

It’s also a shame that the shapshot isn’t quickly generated locally on the same server so the droplet can be turned back on. Then in the background the new shapshot can be slowly moved to the HDD’s at DO’s convenience (rather than our inconvenience).

I hate the speed of snapshot. if I have to perform server upgrade, I have to take backup as snapshot so I can restore to old version if upgrade fails. But since snapshot is unreliable in time it take and it is slow. It make upgrades harder.

Here’s an interesting fact. I have found that snapshots of large DO volumes are very quick to snapshot – much much quicker than snapshots of droplets.
My small droplets (under 5 GB) take about 15 minutes to snapshot. But when I backup my large (over 15 GB) volumes, it takes only a second.
If you have a large amount of data, consider moving that data to a volume to increase snapshot speed.
Hope that helps.

Hi,

I am not an expert but what you have faced its normal with DO snapshot process. Time actually varies with the amount of data you have on the disk and size of the disk. However deploying from snapshot would be much faster compare to taking a snapshot.

Why they are slow in nature ?

First of all, technically speaking, SDD or Solid State Disks are very fast when it comes to read, but when it comes to writing data, they are not as fast as their read speed. Secondly, when it comes snapshot, I believe the disk on which the data is written, is a (mechanical) HDD and may be the data is also written in a different physical node..

Thus the time with network delay along with mechanical HDD operation..

But its only my assumption, best is to wait for DO staff member to clarify !!

Thanks.

Is anyone from DO going to let us know whats up? crazy that I have to shut my VM down and then it still takes 13 minutes for a little 5gig snapshot… what gives??

This is really becoming a problem. I was able to copy my entire droplet to another droplet over networking in about 5 minutes. It’s been 20 minutes and I’m still waiting for a snapshot of the same droplet to finish.

Did you ever hear from DO? Same experience here, in 2019. Snapshots are like molasses.

yeah, this needs to be addressed I think. 40 minutes in and still not done.