Technical term for 'Guaranteed Uptime'?

Lobsang · November 28, 2005, 5:43pm

As refering to a website or ISP.

Mr.Slant · November 28, 2005, 6:10pm

Service Level Agreement? [1]

[1] http://en.wikipedia.org/wiki/Service_Level_Agreement

Voyager · November 28, 2005, 6:32pm

People usually talk in terms of availability, which is defined as the percentage of time a system is available for production use. This is distinguished from reliability, which measures failure rates, and servicibility, which measures how long it takes to restore a down system to use.

People also talk about mean time to failure, and mean time between failures. There is also a measurement called FIT (failures in time) which is the number of failures per billion hours of operation.

Availability is usually the best measure. If two system fails once a month, they both have the same reliability. If one takes a day to bring back up, and the second takes a minute, the second system has far better availability.

Ethilrist · November 28, 2005, 6:39pm

Another acronym is LOS (Level of Service).

Southessex · November 28, 2005, 6:55pm

Nah… the SLA is the overall level of service agreed to in a contract. It includes many other things related to service, such as help desk response times, help desk hours, changes to server configurations, etc. Uptime is an important part of the SLA, but isn’t the only one.

As to the OP, I always just called it “uptime” and no one’s ever looked at me funny.

UncleRojelio · November 28, 2005, 7:07pm

Zero downtime?

Squee · November 28, 2005, 7:43pm

Now what company can profess “zero downtime”? Normally availability is measured using the “nines”… as in 99.9999% uptime for example. Sorry if this seem pendantic.

ElvisL1ves · November 28, 2005, 8:03pm

For airliners, it’s “availability”.

Patty_O_Furniture · November 28, 2005, 8:21pm

I don’t think there is a term for this because no company would be foolish enough to make such a guarantee. SLA’s don’t guarantee up time; they outline what restitution will be offered in the event certain service levels aren’t met. Some of the service levels measured are (1) mean time to repair (MTTR), (2) network availabillity and (3) meeting installation dates.

Voyager · November 28, 2005, 9:57pm

Perhaps not guaranteed, but definitely promised. Customers whose systems are down more than the guarantee are often very unhappy, and restitution doesn’t change that.

Patty_O_Furniture · November 29, 2005, 12:40am

I’m not sure I see an important difference between a promise and a guarantee when you are talking to a customer. To promise 100% uptime is just as foolish as to guarantee it. I interface with 8 blue chip customers and none of them have expressed the notion that 100% uptime is a service requirement. And with only one exception, all of them have redundant lines in anticipation of service interruptions.

Contractors cut through our cables, manholes flood and cars knock down telephone poles. That’s just the way it is.

Patty_O_Furniture · November 29, 2005, 12:52am

I was speaking as the telco, and the ISP is our customer.

I see the OP was talking about ISP’s. An ISP might be able to get away with a 100% uptime assurance if they had a lot of redundancy. If a telco line is cut, they could re-route traffic through an alternate path and continue to provide service to their customers. But if I was the ISP, I’d still worry that I had a possible single point of failure somewhere.

ccwaterback · November 29, 2005, 1:32am

Five nines, six nines, etc.

Availability in networks is measured as an end-to-end calculation generally represented as a percentage of uptime, showing how much of the time the network is available for use. A network with “five nines” availability, for example, is up and available 99.999 percent of the year. The converse is to represent availability by the amount of downtime experienced in minutes per year. Using this measure, a “five nines” network represents about 5.3 minutes of downtime per year.

Manufacturers of older telephone equipment claim that their products have “five nines” availability. Achieving “five nines” uptime is impossible without redundancy (having backup devices always in place). Most private branch exchanges (PBXs) on the market are not redundant. With only one central processing unit (CPU), one memory area, and one hard drive, these systems typically have “three nines” or “four nines” availability. Three nines translates to 8 hours, 46 minutes, and four nines translates to 52 minutes, 33 seconds of downtime per year.

Achieving high availability is not a given. It’s based on design principles and best practices that apply to all systems, whether they are PBXs, IP PBXs, distributed IP telephony systems, server farms, or other network technologies. The main principle is the use of highly reliable components that are redundant.

Availability is based on a mathematical calculation that results in higher uptime as more elements of the phone system become redundant. Cisco CallManager clusters can achieve “six nines” or 99.9999 percent uptime with redundant elements, so that the loss of a line card, CPU, or network connection anywhere in the network will not bring the phone system down. That’s an uptime ratio with a calculated systemic downtime of only 31.5 seconds per year.