I'm not sure what the most typical reasons for down time is perfect for Unix hosts, but on NT (in my opinion) I'd are saying the only most standard reason for down time is poorly written scripts over-running the server. Whether individuals scripts are ASP (most typical) or Perl (possible), NT's built-in safeguards will identify the over-run, and also to safeguard the server (and stop the whole operating-system from crashes), NT will rapidly and silently turn off the net server software.

That's a highly effective safeguard - it knocks the problem script offline instantly and brings the CPU back in check. Regrettably, additionally, it has got the effect of shutting lower all of the internet sites around the server.

At these times, a SysAdmin must log in to the server and restart web services. When the host has someone on-duty 24/7, then your down time most likely will not become more than a few minutes. However, when the host has only technicians on throughout business hrs, well... you can observe the end result.

However, I'll state that despite this problem being not unusual, 99.9% uptime is *easily* sustainable for those who have good sense, implement the correct monitoring systems, and also have staff prepared to cope with such problems.


Jason Ellis, Boss
Hosting Solutions, Corporation.
Now offering Fully Handled Servers!
"There are 720 hrs inside a one month month (744 for any 31 day month). 99.9% uptime would still give you about 7 hrs of down time monthly with different one month month."

Just my encounters, however 7 hrs down time is much more only then do we experience. It is not very difficult to possess less then one hour down time per month.

The issue happens when hosts not have the equipment/staff to correct any large problems/small issues that show up. For me the only real kind of down time that needs to be experienced is down time associated with server upgrade/hubs etc.

However this is not an ideal world, but to remain under 7 hrs, even 3 hrs per month of down time isn't everything hard.

[ Jason Berresford Admin]
[ http://www.can-host.com ]
[ Admin@can-host.com ]
[ (905)765-8140 ]
would someone please explain why website hosts are afflicted by Down time ?

and just how much UPTIME will be practically expected from the good webhost ?

i frequently read 99.9 % UPTIME... something informs me case Baloney !!! shall we be held correct ??!!!!

however, you will find some companies who condition 90% or 97% and so forth etc.

if only to be aware what "downtime" is actually because of.
&gtwould someone please explain why hosting &gtcompanies are afflicted by Down time ?

Down time could be credited to many factors including server crashes, router problems, or hardware/software glitches.

Another facet of server down time is planned maintenance. Normally, this really is down time that's introduced, and it is ususally with regards to improving hardware, operating-system or any other vital software towards the server

(Home windows NT 4 is well known for reboots triggered by software installs).

&gtand just how much UPTIME will be practically &gtexpected from the good webhost ?
&gti frequently read 99.9 % UPTIME... something &gttells me case Baloney !!! shall we be held &gtright ??!!!!

I disagree with this particular somewhat. 99.9% is very inside the arena of achievable. Think about this:

You will find 720 hrs inside a one month month (744 for any 31 day month). 99.9% uptime would still give you about 7 hrs of down time monthly with different one month month.

Sometimes mainly like a Unix/NT/WAN Sysadmin for a pretty big government entity. If my network would suffer 7 hrs of down time inside a month, it might be uncommon.

On the other hand, I additionally re-sell NT hosting (I'm most proficient with NT), and the organization I re-sell for has experienced from what I will tell about 12 hrs of down time (for my website and every one of my clients) within the last 24 months.

I haven't seen a lot of hosts that may provide that kind of stability (although I've been together mainly the entire time I've had my website).

So yes, 99.9% is very accessible, however it requires the organization to purchase good hardware, connectivity, and more importantly very competent sysadmins.

&gton another hands, you will find some companies &gtwho condition 90% or 97% and so forth etc.

For any personal site, 90% uptime is most likely in the low finish of acceptable (if acceptable whatsoever). I possibly could not fathom getting 21 hrs of down time per month. Most E-commerce sites would lose a lot of earnings from that type of average down time IMO.

&gti need to know what "downtime" is actually &gtdue to.

More occasions these days, it appears that down time is related more to hardware crashes/lost connectivity than other things, although I am unable to attest to the hosts available. Within the situation I referred to over the lost down time for my website and also the sites of my clients was triggered by hidden Fiber Optic cables that got cut.

I am certain the hosts out here could most likely give their undertake what's the most standard reason for down time.

Hope this can help,


Thomas A. Kangas
Down time may also have allot related to the rout between your server, the easiest method to tell your reason for lower would be to operate a trace route when you are lower. There has been many occasions which i get a complaint from the client saying the server is lower but they're the only real ones that may not interact with it. Everything has related to routing, plus some hubs are stupid plus they may take some time to locate a another path to the server. These I've discovered would be the downtimes that only last ten or fifteen minuets. When the route between your server continues to be damaged it might take the hubs a while to determine the following best path to the server. We're around the Alabanza network and also have been for a while, and I must say that they're up 99.9% of times. So the next time you decide to go lower don't let yourself be in this large hurry to visualize the server is lower, perform a trace route which can have were the path breaks off. I know that allot of servers do go lower allot, but when you are well on a trusted server odds are it's a routing problem not the server.

JRC Systems
This is a connect to a totally free trace rout program that's excellent.

JRC Systems

So true, Dork. Almost all of individuals on the web havent have a clue regarding how it operates so that they assume when they can't connect with a website, it should be the responsibility of the website. For your reason, I added a webpage known as Internet Issues and FAQ to my website more than a year ago, and today get it on my small server site too. It describes, basically, a few of the reasons there might be problems in hooking up and just how it's possible to check.

It's cut lower on questions regarding connectivity during my discussion forums and works as a quick reference for that periodic couple of ones that appear. It's simpler to touch on individuals to the page than answer their questions with short solutions.

The page I am mentioning to reaches http://techcellence.net/internetfaq.htm It features a link at the end to an infinitely more detailed coverage from the Internet and it is roots.
WOAH!!!! Careful all. There's a significant flaw within the following math:
You will find 720 hrs inside a one month month (744 for any 31 day month). 99.9% uptime would still give you about 7 hrs of down time monthly with different one month month.
Case erroneously incorrect...

First of all, Personally, i believe it's better to put into practice the moment since fifteen minutes of down time is serious plus an hour of down time. I am frequently upset by hosts which will calculate down time on an hourly basis the way it safeguards them from 20 and half an hour lower in addition to by hosts who wont calculate it whatsoever unless of course they're 'forced' to with a site owner but that is another problem...


Going from a one month month you will find 720 hrs which is equivalent to 43,200 minutes. Let's focus on the mathematics:

43,200 * .999 (99.9%) = 43,156.8 minutes.
43,200 - 43,156.8 minutes = 43.2 minutes.
Which means this enables the host to possess as much as 43.2 minutes of down time while remaining within 99.9% uptime. An apparent long way away from 7 hrs!

The next shows Uptime(ut) = An allowance as high as x minutes(m) down time(dt).

99.9% UT = 43.2m DT (under one hour)
99.8% UT = 86.4m DT (1.44 hrs)
99.7% UT = 129.6m DT (2.16 hrs)
99.6% UT = 172.8m DT (2.88 hrs)
99.5% UT = 216m DT (3.6 hrs)
99.4% UT = 259.2m DT (4.32 hrs)
99.3% UT = 302.4m DT (5.04 hrs)
99.2% UT = 345.6m DT (5.76 hrs)
99.1% UT = 388.8m DT (6.48 hrs)
99.% UT = 432m DT (7.2 hrs)

There's just far too huge of the difference... As you can tell a 99.% uptime guarantee would permit the 7 hrs but a 99.9% uptime guarantee certainly wouldn't.


- Http://www.FutureQuest.net/

For any personal site, 90% uptime is most likely in the low finish of acceptable (if acceptable whatsoever). I possibly could not fathom getting 21 hrs of down time per month.
98% UT = 14.4 hrs permitted down time.
97% UT = 21.6 hrs permitted down time
96% UT = 28.8 hrs permitted down time.
95% UT = 36 hrs permitted down time
94% UT = 43.2 hrs permitted down time.
93% UT = 50.4 hrs permitted down time
92% UT = 57.6 hrs permitted down time.
91% UT = 64.8 hrs permitted down time
90% UT = 72 hrs permitted down time


A decimal is really a terrible factor down the sink.

[This message continues to be edited by N (edited 07-04-2000).]

Good Catch N. I miscalulated. My appologies towards the board.


Thomas A. Kangas
  • 1