Has anyone ever considered what 99+% uptime actually is???

99+% uptime of what??

1% downtime of 525,600 minutes in a year is 5,256 minutes of potential downtime.

5,256 minutes divided by 60 minutes in an hour = 87.6 potential hours of downtime in a year!

So how is it typically calculated?

99% of what??

Thats why it is so easy to obtain, yet sounds good :D

one 8 hour day of downtime per month would be enough for most hosts to lose clients

In most cases, the uptime guarantee is calulated on a monthly basis. 99.9% uptime guaranteed per month gives a maximum of 43 minutes possible downtime in that month, for example.

99% uptime is something you shouldn't worry about. If a provider can't guarantee better uptime than that, they shouldn't be in the hosting business.

Simon

Marketing poop in other words ;)

So here is a comparison:

99.6% uptime = 2102.4 potential minutes of downtime per year, or about 35 hours.

99.9% uptime = 525.6 potential minutes of downtime per year, or about 9 hours.

I for one do not believe ANY host can avoid having 9 hours of downtime over the course of a year on any given server/internet system except by heavy intervention by the Almighty.

The downtime will be spread out hopefully, so it isn't noticed really.

But, I read over and over about disk/raid/connection failures etc. on the various host's forums. The time that elapses from the problem to the fix is rarely less than 30 minutes in the posts I've read.

That's only 4.5 of those kinds of booboos a year before the old 99.9% mark is defeated. Many of the snafus are more like hours in duration. This is all normal and to be expected.

All hosts are going to have downtime. I believe the best a host can do is 99.5% simply because the dang Internet is soooooo complicated in the first place. Frankly, if they can realistically achieve 99.5% it is a blessed miracle of human effort! Somebody probably died in the process or lost a limb!

It easily takes well over 43 minutes to rebuild a raid or replace/restore a drive. I've done it. So, unless a host has failover protection, which most don't, 43 minutes is going to be a hard number to deal with.

Now the big guys have all of this of course. So I hear you saying that there should be no small guys in the business. That's a scary thought.

:) Hopefully I didn't come off as argumentative with that last post. All I'm really saying is be prepared to dance for your clients when your reseller host has downtime because he will. It can quite easily be more than 43 minutes a month.

It might be worth noting that sometimes companies exclude scheduled maintenance from their uptime statistics. I don't know how common this practice is, though.

Uptime guarantees aren't specifying levels that the host won't go under. They're specifying the levels where various compensations are due. Second, more often than not, you have to ask for the compensation, or none will be awarded.

Yes, hardware issues can lead to extended downtime. On the whole though, for a well sized company, one server's poor uptime can be easily seen as more than compensated by the good uptime of hundreds of other servers (when analysing quality of service levels internally).

99.9% yearly uptime isn't impossible to be achieved over a decent sized fleet of servers, by a good host. 99.9%+ uptime every month for each server, undefinitely, is not anywhere so likely to happen.

A link I recalled: http://www.easyuptimecalc.com/

Marketing poop in other words ;)
Nope. A proper uptime guarantee can be the backbone of any company.

All hosts are going to have downtime. I believe the best a host can do is 99.5% simply because the dang Internet is soooooo complicated in the first place. Frankly, if they can realistically achieve 99.5% it is a blessed miracle of human effort! Somebody probably died in the process or lost a limb!
Again, much like 99% uptime, if a provider can't avoid 3 hours of downtime every month, they need to look at their business and contemplate whether they should be in this industry. 99.5% uptime is not a miracle; it's a very bad month. Redundancy is key... one bandwidth provider going to outage status shouldn't crush a company, it should just mean that the other providers take the load.

It might be worth noting that sometimes companies exclude scheduled maintenance from their uptime statistics. I don't know how common this practice is, though.
That's a valid point. Most providers will have it built in to the SLA that scheduled maintenance (with a mandatory warning period) will not be classed as downtime.

Now the big guys have all of this of course. So I hear you saying that there should be no small guys in the business. That's a scary thought.
While I do see your point, this industry is not about just splitting the big from the small. The big exist because of the small, and vice versa. If a small company can't live up to relatively easily to attain standards, then they need to look upstream and make better choices when it comes to their tier 1 (and 2) providers.

All I'm really saying is be prepared to dance for your clients when your reseller host has downtime because he will.
If someone has to "dance" to get their SLA credit, they need to move to a different provider. Sure, there are companies out there that tend to act that way, but they shouldn't be used to form an opinion of the industry as a whole. For every company that makes their clients "dance" for SLA credits, there's many, many others who honour their agreements.

Simon

Folks never do seem to realize the big difference an additional 9 can make to an uptime statement. Another thing they don't realize is the exponential cost increase for each 9 that is added. For most folks 99.9% uptime is sufficient, but that still leaves a good amount of downtime - downtime that a site such as Amazon.com could not deal with. There are definite needs for virtual 100% uptime, but this is expensive, and most users just don't need it.

I would personally much rather see a 99.9% or 99.99% offering than 100% though from providers. All of these hosts/server companies on WHT that advertise 100% are doing the marketing poop like dwest said. To maintain a 100% uptime requires multiple power feeds, multiple power supplies to said multiple power feeds, multiple high availabilty servers, redundant connected/san storage, multiple networks and much more. Even the huge boys such as Amazon, Yahoo, Google, Godaddy can't fully achieve 100% every month.

^^^ You forgot one: multiple data centers! :P And as you point out, even then it's still impossible. Even if there's only 1 minute of downtime in a year, that's still downtime.

Yea I left that one out as the more I started thinking about past experiences the more I came up with.

IHSL:
You misunderstand who I mean by clients. As a site developer who earns income from hosting the sites I develop, I have to "dance" for MY clients, explaining why their sites are down. All when the people addressing the problem are totally out of my control.

As for sla compensation from the host I rent from, it might as well not exist. Most hosts honor it of course, but I ask, what good is it? It certainly doesn't bring back angry clients I've lost as a developer. I mean really, we are talking mere dollars in refunds or credits. The time cost of getting it is more than it amounts to.

99.5% and 99.9% are nothing more than marketing claims. Hell 50% is nothing more than a marketing claim, albeit a bad one :).

The backbone of any host is whether they communicate to their resellers and provide timely responses to downtime. The marketing poop of putting a percentage figure on that is meaningless.

Anyone who places their faith in the uptime brags of a host is looking at the wrong information to find comfort. That is the reason I started this thread.:)

i thin most providers try not to have down time they only put that up as a just in case liek if there upgrading or somethign

Has anyone ever considered what 99+% uptime actually is???

99+% uptime of what??

1% downtime of 525,600 minutes in a year is 5,256 minutes of potential downtime.

5,256 minutes divided by 60 minutes in an hour = 87.6 potential hours of downtime in a year!

So how is it typically calculated?

99% of what??


99.9% per month uptime of the service (hardware - server and network), nothing is perfect in this world. :D

99% uptime is definitely not hard to achieve, but you have to remember that it's guaranteed every month. 87 hours of allowable downtime spread out over 12 months. Still that's not acceptable. Works out to be about 14 minutes of downtime every day.