IT Employment

Ask something
Discussion Locked

Acceptable downtime percentage

0 helpful
20 replies

by campbellwj 1 view

Closed to new repliesThis discussion is closed. It stays readable, but it takes no new replies.

Hi all,

I was wondering if someone could point me in the right direction to find some information about uptime and downtime percentages. I would like to know what the acceptable industry standard is for uptime / downtime annual percentage. I did some numbers and found out that my MIS Department had a 98.9% uptime rating for 2001. I would like to have some input on this please.

20 replies

by james r linn

Is your organization a 24hr organization? - if so then your stats should be based on 24 hours(and scheduled downtime counts as an outage).

If your org is not a 24x7 org, you should think about scheduling outages outside your working hours, and counting either only your working hours(and expecting amazing availability) or somehow weighting your working hours more heavily than your non-working hours.

You should also consider server classes. We have servers in our datacentre which are mission critical, and we care a lot more about their availability than the 5 year old workstation that someone uses as an occasional file server. We stratify our servers into levels and weigh the mission critical(email, financials, HR) far higher than the others.

We strive for 3 9s or 99.9 for mission critical servers in working hours. We usually meet or come close.

So you see its highly variable based on how you decide to measure it and why.

James

by bklein

If your shop is 24/7 and your uptime is 99.0%, this means you're dead for 14 1/2 minutes every day.

by admin

With workstations our business considers repairs not to be downtime if the maintanance/repair occurs during a time the WS is not needed or a loaner is used to replace it. Makes 99.9% uptime much easier. Of course, it's really scheduled downtime to us.

Good points on different ways to measure it. Internally we watch downtime differently so we can say "This machine needs replaced" but the company directors want bragging rights so that they can say "99.9%" uptime to the staff as a benefit. The trick is not to let it impact our budget..... "What do you mean we need to buy a new server? It has 99.9% uptime- why would we replace it....." That's when we need a whole different measurement :)

by generalist

The part about 'not needing' a new server because of 99.9% uptime is SO true it is a problem. Maybe the thing to do is to prove that the 99.9% uptime claim is due to the server replacement policies. If they want to maintain that, they had better replace it.

by e-fellow

Six Sigma ratings are 99.9997%. So shoot for zero downtime and I am sure you do, however, 98.9% is 3.8 sigma IF YOU HAVE THE OPPURTUNITY TO BE UP 100% OF A 24 HOUR DAY AND YOU ARE UP 89.8%, then go to work. We have experienced about 3 hours of non-scheduled downtime in about 9 months and have only scheduled about 3 hours during operational hours. You have a problem somewhere, find it, fix it or live with it.

by james r linn

If we could keep the same suite of applications running for 1 year, with no downtime for upgrades, service packs, memory upgrades, disk upgrades, processor upgrades etc., that would be a realistic goal.

We have a scheduled downtime once a week, so we strive to make sure we do as little as possible outside the scheduled downtime.

James

by e-fellow

Think outside your box and listen to what you said to me, "We have a scheduled downtime once a week". Why do that, if you have a good reason to do that, find a way to stop, good grief. CSM, common sense management. Run the system do not let it run you. There is no such thing as a realistic goal, why set an unrealistic one,,,,,,,, there is a 99.999999% you are not going to invent the technology to make your situation better, so find out if you are the best, if not, mock the best. Good luck guy.

by james r linn

is something I follow every day.

First of all, we don't have a bad sitaution. We often have 99.95% availability, and if you looked at our normal working hours, we'd probably be close to the 99.999% goal.

We have a 3 hour maintenance window, outside of normal working hours. We only take down servers which require maintenance - and sometimes routine maintenance waits for an opportunistic time.

For example we like to keep our servers on the same Service Pack, but its not absolutely necessary for all servers from a technical pespective. So instead of taking it down at the first opportunity, we wait until the server needs a reboot anyway.

All I am saying is be realistic. Not all businesses need electronic services 24/7. Some scheduled maintenance may sacrifice short term availability, but improve long term performance and reliability.

You don't have to wish me luck thanks. I'm doing just fine thanks very much.

James

by e-fellow

You said that your uptime was 89.9%. If your users are not able to access critical systems 10.1% of the year then you are right, you need a lot more than luck.

by james r linn

I never said 89.9. I said 99.9 is our goal (measured 24x7) and we usually meet it. And as you will see our customers are not expecting our services to be available more than 8-6 Mon-Fri, because that is when they work. If you took that period of time, we'd be closer to 99.999.

Please get your facts straight before you insult others.

James

by james r linn

But if you really think that 98.9 is no better than 89.9, then I can suggest a few basic math courses.

Do you think a Mortgage at 5% is no better than a mortgage at 5.5%?

I've spent a good deal of time in my life crunching numbers. Part of my degree was sucessful completion of a statistics course. But my 8 year old could do the math on whether 89.8 and 98.9 are close to each other.

I'm not normally prone to insult, but you seem to be somewhat hostile and abusive of other posters here, and to boot, part of the problem is you have frequently mistated the numbers they have used.

Perhaps if you can't engage in a civil and reasoned dialogue you should consider not posting.

James

by techrepublic

Efellow:\Your apology doesn't make sense. Even the original says 98.9%. not 89.8%. You better start the day over or learn to read :).

by e-fellow

My message said 89.9% not 89.8% Like your statement says "Even the original says 98.9%. not 89.8%." Looks like we both do not read very well. However, in my mind, 98.9% is no better than 89.8%, you are either "first-class" or you are not.

by campbellwj

I just want to say thank you to everyone who participate in this discussion. Although I appreciate everyone's comments, I know that you should shoot the best possible and achievable goal. I was just wandering if there was an actual industry standard. I have to present this numbers to the CFO and CEO and just want to have some backup and set a standard in their minds and to remove any negative perceptions they may have.

Thanks again.

Bill Campbell

by generalist

If you're SELLING hardware, you want to convince the customer that six sigma is possible, especially in mission critical situations like keeping an airplane in the air or a life support machine functioning.

If you're BUYING hardware, the bank account, the accounting staff, the public relations staff and the high level executives determine the standard. They may decide that the 500% to 5000% increase in cost to get from 99.8 to an 'ideal' level may not be worth it. And, from an accounting standpoint, they may be right.

by e-fellow

100% is the goal, if you miss it be prepared to explain why and what it would take to keep it at 100%; costs and time.

Six sigma, 7 sigma????? Just remember the difference between winning and loosing sometimes is the STRETCH! Cost of QOS is a consideration, the bad news is you have to do the research and present the most feasible options, the good news?, you can do it! Good Luck guy?

by e-fellow

Not sure what happened to my response to you.
#1. I am not hostile or abusive.
#2. I have screwed up no more numbers than you. (You said "But my 8 year old could do the math on whether 89.8 and 98.9 are close to each other.")
#3. Next time you fly you tell me if 98.9% flight success rate is any better than 89.9%!

by e-fellow

Concerning flight success.

Delta flies over 1 million filghts per year. They experience a crash about 1 per every 4 million flights, which is far beyond 6 six sigma, six sigma is 3.4 crashes per 1 million flights. At this rate Delta would have over 13 crashes in 4 millions flights.

Also note that 89.9% success rates would yield around 100,000 crashes per 1 million flights.

You say 98.9% is better and you are correct, however, 98.9% would yield over 10,000 crashes per 1 million flights.

100,000,,,,,,10,000 or .25 crashes per 1 million flights, CHOOSE ONE!!!! Speaking of statistics.

I know that every single process cannot be 99.99999999% or six sigma, however, we must have goals that make us STRETCH!!!

And I guess the over$14,000.00 saved on a $200,000.00 mortage for 20 years is nothing?

I apologize for sounding like an "A-HOLE", please keep in mind that as we collaborate and debate these issues, we learn, I know I have learned a lot, even from people I do not like.

by http://stobie.home.sprynet.com/work/

From a business perspective you need to consider it by function.

Many businesses provide (or try to provide) continuous (no-down) service to their customers. In these cases the functional downtime preferred is 0. However, the business is may only be willing to spend so much on insuring uptime (cost per uptime gained evaluation), so certain outages might not be protected against (such as multiple castrophies in multiple places.)

Some business functions are only performed certain times ofthe day. In these cases, downtime should be concentrated in offhours and cost/performance evaluations of insuring uptime should be performed with different values for different times of day (week, etc).

You need to consider the true costs of downtime and evaluate if how much protection is warrented.

This discussion is closed to new replies.