I have set up (paid for) monitoring on 3 different hosts. I have set up the "check" time to happen on all 3 at the same time.

At the end of the month I want to see the % of all three. Who has the "most" uptime obviously.

Do you think this a fair test? Should the "month" be a longer time period?

Carol Parent

Instead of doing the test for one month only, you should extend it to maybe a few months so then you can average their uptime over an amount of time. This will give you a better taste of what their uptime is like. :)

What are you using to check our uptime?

Keep in mind that not only is uptime important but so is support. A host can have the best uptime around and real bad service. So I would not base your decision entirely on uptime alone.

One other thing to think about is what if one of the hosts you are monitoring has a schedule maintenance period during one of your testing times, you not being a customer would not be aware of such an event and may get false negatives.

The above is just food for thought, wish you the best in your search.

This is more for my own knowledge than anything else. Most of my clients are oblivous to the internet and "trust" me to take care of everything.

I do have a couple of sites tho that are used on a daily basis by clients as part of their presentations for their products and these are the ones I am the most concerned about.

Thanks for your replies.

Carol Parent

We do hot tape backups daily but we also do a complete bounce backup weekly (In the early hours Sunday morning) to ensure all data is captured. This is because tapes will miss any data that is active at time of backup and is so easily corrupted.

Bounce backup takes 45 minutes on one of our servers. Therefore if you were to monitor us, we not do aswell as other hosts. But, in fact, our service is better as we are ensuring our client's data is completely stored.

So your test would dismiss us, maybe for being down 3 hours a month. But our clients tell us continually that our service is the best they've experienced.

So, your test, while good in some respects, should not be used solely as a basis for choosing a host.

if any enquiry has been made previously you might find http://www.netcraft.com of some use to you regarding uptime.

You can check each host via the 'what's this site running' option (just type in their domain name), as well as check uptime for things like their neighbours in the same NOC

http://uptime.netcraft.com/up/hosted?netname=NAC-NETBLK04,216.118.64.0,216.118.127.255

(Network Access Corp for example where we are).

Of course the above posts count for a great deal as well. Support levels and system maintainence will count for more in the long run. I wouldn't call Netcraft the total solution, but it might be another addition to your arsenal.

Greg Moore

Originally posted by NexDog
...Bounce backup takes 45 minutes on one of our servers. ...

I agree that preserving the servers content is important and that most customers would forgive you for the "bounce" time of 45 minutes.

You may wish to consider building your servers a little different. For example, use three drives. Mirror the first two drives, rotate the Master Mirror drive every Sunday morning in the time it takes to reboot the server. You could shave the "bounce" time down to 100 seconds. With a catastrophic server failure you could be up in just 100 seconds in the condition the server was on Sunday by using the "cold" third drive.

Of course, the servers cost more now and so you may have to charge a little more for hosting - your customers may wish to keep the 45 minutes and their money. :)

Scott Hirsch

if any enquiry has been made previously you might find http://www.netcraft.com of some use to you regarding uptime.

Thanks very much for an excellent resource.

:)

Carol Parent

Originally posted by SecureWebs


I agree that preserving the servers content is important and that most customers would forgive you for the "bounce" time of 45 minutes.

You may wish to consider building your servers a little different. For example, use three drives. Mirror the first two drives, rotate the Master Mirror drive every Sunday morning in the time it takes to reboot the server. You could shave the "bounce" time down to 100 seconds. With a catastrophic server failure you could be up in just 100 seconds in the condition the server was on Sunday by using the "cold" third drive.

Of course, the servers cost more now and so you may have to charge a little more for hosting - your customers may wish to keep the 45 minutes and their money. :)

Scott Hirsch

Thanks, Scott, this is an excellent idea and well worth looking into. :)