What is reasonable downtime?

A survey such as you suggest would eiter have zero value because of the numerous variables involved, or be so complex as to make it unusable.

When talking about acceptable downtime, you first need to look at the big picture. What is the criticality of the processes being run. "Acceptable" downtime for a computer (or network) running an air traffic control system or a nuclear reactor is 0, while the "acceptable" downtime for a business that uses a minimal amount of automation might be days. "Acceptable" downtime might vary between the different servers run in a given organization.

Because what is "acceptable" can very much depend on a particular organizations "culture," there is no such thing as "normal downtime." What is acceptable to one person may not be acceptable to another. The bottom line is you need to determine, realistically (with the help of management), at what point a particular system being down will start to impact the business (or how long before you as thesystems administrator start feeling the heat from upper management). I don't know about you, but any time the email server is down for more than 10 minutes, I start getting calls from everybody, from the boss, on down. Therefore, 10 minutes is the "acceptable" downtime for my email server.

The bottom line reason for a system being down has no real bearing on this unless it is something totally beyond your, and your companies control (i.e. natural disaster, city-wide power outage, etc. This is why you need a good business continuity/disaster recovery plan).

Hope this gives you some things to think about.

Hi, I agree with everything that tim.walsh said, plus --- It also depends on how your organization defines and measures downtime. For example, if you support (as I do) some retail stores and an office, the company doesn't really care (and wouldn't notice) if the system were unavailable from 10PM to 6AM daily. I can use some or all of that time for maintenance and upgrades. As far as the company is concerned, there is "no" downtime! If, however, we had a website that did e-commerce globally,well, you see that the requirements change when you do business 24x7...

Is any downtime acceptible? In your office, it's at least good that you have multiple services running, yet have them running on different machines than lumping them all in on one device. This will almost eliminate other services being affected by the server that is down at the time. With business such as your, I'm assuming a lot here by guessing that you should be running RAID5 in all your servers to ensure uptime, and backing them up nightly. If you're not, then a very long time of acceptibledowntime must be allowed. If you're not at least running this, then you're looking for a long road, but if you are, you have a godo start. go forward with your contingency planning in case something does happen. This will be a good, creative, proactive way for you and management to be on the same page, and know that you've faced it in advance and are prepared rather then waiting and being reactive instead.

Perhaps the best way to determine what is acceptable downtime, is to look at the cost to the business of having a particular system down.

For an office manager, 2 days may be acceptable.

A clerk may be of no value to the firm as soon as their PC goes down.

For an inventory control system, 5 minutes of downtime may be too long.

It really depends on how critical the PC is to the workers task and the companies business.

Where I work, we have complex test systems that may be down fora week at a time.
This is 'acceptable' as the cost of back-up systems and hardware far exceeds the cost of a week of downtime.

Chas

IMO, scheduled downtime is not a problem. Need to patch the servers and give advanced notice what specific services will be down that weekend, or evening and you can even get up to 10% or so and be no problem, but where I work if you have 3 hours down in the middle of a day unscheduled, it is nearly crippling to some departments and incredibly expensive for the company.

I ask when a lot more than how much is acceptable. Patch, or limp hardware along enough to get through the busy hours and schedule time down ASAP when it's less critical.

Of course, some say this is impossible, but through redundacy, staying on top of things and time management I think most of it can be controlled with less work overall. Planning is everything, always have back-ups and back-ups and back-ups..... when one system fails, have another to switch to, the few times everything fails or gets close, squeeze that last drop out of the dilithium crystal until you get out of the warp zone. Then schedule a little downtime :)

There, that's pretty much my comment.

In some organizations you may need to consider personnel and labor limits when dealing with acceptable downtime. Said limits may prohibit you from responding quickly or outside 'normal' hours.

Of course, this usually happens when a union is involved, so you can often use the union as an excuse.