Uptime - what to expect?
I'm a newcomer (customer) to the world of IaaS and cloud hosting.
But an old hand at runnning my own iron in-house, both virutalized and non-virutalized (close to 20 years).
For the last ca. 8 months I've had a server in the cloud.
A full virtual guest server running Ubuntu installed and managed by me.
And in that time have experienced hours long unplanned downtime on 3 occasions.
Each time requiring at support ticket to get the server online again. I.e. problem could not be solved by any form of remote management available to me as a customer. Restarting instance, console access, and so on not working.
And each time caused by some outage at the cloud provider, where the provider where aware of and dealing with the root cause.
But where (apparently) little were done to discover dead guests/hosts in need of manual intervention from the provider to get running again.
Now, "luck" obviously plays a part in any kind of hosting, when it comes to uptime.
Nothing can be completely under the control of us mere humans.
But I'd like to hear how you guys find this experience?
Does it sound about right?
I'm considering going back to running this server in-house, since uptime certainly isn't any better than what I can get with a cheap PC server, a single Internet connection, and perhaps a small UPS.
Costs are somewhat similar too, so I'm not sure any value is really added by running this server in the cloud.
And some "value" is certainly lost by the fact that I can't look people in the eyes (sometimes using a mirror), while I yell at them to get things fixed!
Is this a reasonably track record for a cloud provider in the segment we are talking about here on this board?
The point is not to single out my current provider, so I won't go into that.
But to hear what your experience is in this market, so I can set my expectations right.
If the provider says that the VM is in the cloud it doesn't mean there is some sort of HA, mate.
The only downtime I have experienced with the cloud providers I have used in the past was maybe 1-2 minutes (if that) every few months. I would say the majority of the issues I had with cloud providers is the control panels would go wonky (like you said) and shit just wouldn't work which is kinda annoying.
Actually, this provider does advertise HA + automatic fail-over to different node through monitoring.
But these things hasn't covered my cases of downtime it would seem.
HA etc. is the kind of value-add I hoped to gain by going from my own iron and to "the cloud".
Only times if have noticed control panels going wonky is when my server has also been down.
@netman - wrong provider.
EDIT: I always wondered how that migrate everyone to a new node would work. How can they push a few of TB of data to a node in a reasonable enough length of time to call it HA?
@netman have you asked why the failover didn't kick in?
No. That would be the obvious thing to do - I realise, now that you ask.