Hi all there,
a question for all ded/colo. server Users.

How much downtime - in hours - for you server/s (including all domains hosted on it) will you accept from your datacenter befor changing to another center if this downtimes happens every month?

Thanks

In what time period?

It would really depend on many factors. quality of the host, type of downtimes, etc.

but in general i'd say I'm willing to handle a few hours a month max.

20min per month max

The questions begs more than just a simple answer, so let me ramble a bit.

The answer depends upon, as someone else mentioned, what is the type and cause of downtime.

I have a very low tolerance of downtime due to stupidity. One of my previous employers for experienced near weekly downtime on their network. Downtime was short (about 15 minutes or so), but it occurred nearly every week. Finally I asked what was the cause of the downtime. The answer: "I tripped on the router's power cord again." When I asked why he didn't reroute the power cord so that it didn't cross the main aisle at ankle level (really!); he said, "that's a good idea." Stupidity, plan and simple. When I see that in any vendor of mine, I run fast and far.

On the other end of the spectrum, if the recurring problems are due to something beyond the vendor's control, and there appear to be occurring in an unfortunate clustering, then I am far more tolerant. One of the two hosts that I currently use recently went through what I call "character-building time". Between support staff leaving, upstream router problems, newborn baby on the home front, etc., he had his hands full. But I stayed with him. Why? Because I could have done no better under the circumstances he experienced, and I knew that he wants to provide a high-quality service. And seeing his actions in the recent past, I am glad I stayed with him.

I tend to view my hosts as business partners, people who help me to reach my goals. There is no room for adversity towards your business partners. There is plenty of room for goal alignment, i.e., both having the desire to provide high-quality hosting services to their customers.

Whatever you measure will be optimized. If you look for hosting based upon price alone, you'll wind up with cheap hosting. If you search for hosting based upon downtime alone, you will wind up with a host that has low downtime. But if you search for hosting based upon an alignment with your business goals, you'll wind up with a host that wants to help your business.

I think there is two types type of downtime

There is downtime caused by ie the datacenter and there is downtime caused by the client.

From the datacenters end a few minutes here or there is no big deal, overall it is probably going to happen, long dirations is not good.

If the client caused the downtime or one of his clients caused it then regardless of how fast support is, i dont think its the datacenters responsibility to babysit every client, and also not its job. Overall when you have alot of servers you just cant. I think its your job once the problem is known to fix it asap, but you shouldnt be blamed

Joe

Originally posted by Mike the newbie
I tend to view my hosts as business partners, people who help me to reach my goals. There is no room for adversity towards your business partners. There is plenty of room for goal alignment, i.e., both having the desire to provide high-quality hosting services to their customers.

True, very true for me. The most important thing for me is if the host follows this rule. If not, even very low downtime can't help.

When I was a reseller, the server my accounts where on faced a bad harddisk, but because it was not a big hd crash which they would have known immediately they first thought about bad memory and change this instead of hd. The whole issue went over a few days with the server crashing every now and then. I stayed with them because not only do I know that sh*t happens, but mainly because I knew every hour what was going on (communication!) and I knew from times before that they will manage it.

Originally posted by Mike the newbie
When I asked why he didn't reroute the power cord so that it didn't cross the main aisle at ankle level (really!); he said, "that's a good idea."

You are kidding, are you? :D

Originally posted by Walter
You are kidding, are you? :D True story.

Hey, you don't have an option for over 500 hours of downtime!

When you're with ***** you can expect at least 100 hours of downtime/month...it's their policy!

As a rule...
20min per month max


IMHO - I think this is unreasonable. I can deal with a server like - freezing or abending or something (it's just a part of life), at which point by the time it's rebooted it could take around 10 minutes just to do disk scan and come back around (more if it's god awful nt or 2k.) So right there we're at around 15 minutes - down the box for a reboot again to add a feature upgrade and your up to 25 - now certainly we're approaching 30 minutes and that is a bit severe - but I think that 20 is a bit to harsh.

-neil