Ask HN: Vultr lost customer data. How is EC2? | Hacker News Reader

Ask HN: Vultr lost customer data. How is EC2?

1 pointcodegeek4 comments

We had a catastrophic incident where vultr had a hardware node failure of some sort and we lost some data (we do take offsite backups but lost few hours of data that was not backed up yet and this particular customer has lot of activity).

After the whole night on tickets with vultr, their reply is "sorry suck it. we will give you 60 days credit". We pay them thousands per month for tons of customers. Not trying to bash them but more interested in asking HN if EC2 etc is more reliable ? This is not the first time we had issues with them recently (used to be great before). So I have decided that we are completely moving off Vultr.

We do use EC2 for some and DigitalOcean as well to mix it up. I am thinking that we should completely move to EC2 as I have found AWS Support to be the best.

Is this just the way it is with VPS companies or is EC2 more reliable ?

If they did tell you to 'suck it', then take ALL your business elsewhere.

You can set up ongoing replication to another region so if there is a failure you don't actually lose anything. With AWS just be mindful of cost, they can get exorbitant pretty quickly.

Are you saying that you had an agreement with Vultr that they would back up your data more frequent than 2 hours and they failed to do that?

Hardware fails regardless of VPS or physical server provider. Hard drives fail, raid controllers fail, raid arrays become corrupted on all providers EC2 and all other VPS providers equally. Data integrity is based on what backup strategy is in place. For highly critical data the application should be writing to n+1 locations, or a middle process such as ceph or gluster should handle that. The AWS version of this would be S3. What was your backup strategy aside from taking the offsite backups?

EC2 is good but AWS also offers many managed DB services as well. Regardless of what you pick, you need to have redundancies. This could mean you have a replica in another region.

I have had good luck with Vultr, but also with Linode. My small Linode VM (HashBackup upgrade server) has been up for 308 days. They took it down back then to do some kind of microcode upgrade as I recall. They're really good about notifying you of any scheduled downtime and have helped me on the phone when an OS upgrade didn't go so great (they got it straightened out).

Reply on news.ycombinator.com