5

As far as my understanding goes the availability of a service or system is defined as:

(Time resource was available - Time resource was unavailable) / Total Time

My questions are:

Is describing a system/service as "Highly Available" a standard industry recognised term?

Which leads to my next question:

If a software vendor claims to sell a High Availability solution, is there a minimum level of availability one could reasonably expect for the solution to fit the description?

2
  • I don't think this question is too broad. In fact it's precisely the opposite. I'm looking for exact definitions, not opinions.
    – Aditya K
    May 20, 2015 at 8:14
  • 3
    It's too broad because no, there is no standard definition for "highly available." It's just a marketing term. To know what the vendor means by that, you need to get them to tell you what percent uptime they're prepared to guarantee, and what they'll do if they don't meet that. May 22, 2015 at 6:30

3 Answers 3

6

My question was to determine if there is an precise definition of how much availability constitutes High Availability

There is no official definition of how much availability constitues "high availability". Thus, there are several more or less established definitions around the world. I don't want to be more precise as I don't really know if there is any widely accepted standard. So I am listing those I know.

The Harvard Research Group defined several so-called Availability Environments Classifications and that is maybe as close as you can get in terms of precise definition.

  • AE4 Business functions that demand continuous computing and where any failure is transparent to the user.
    This means no interruption of work; no transactions lost; no degradation in performance; and continuous 24x7 operation.

  • AE3 Business functions that require uninterrupted computing services, either during essential time periods, or during most hours of the day and most days of the week throughout the year. This means that the user stays on-line. However, the current transaction may need restarting and users may experience some performance degradation.

  • AE2 Business functions that allow minimally interrupted computing services, either during essential time periods, or during most hours of the day and most days of the week throughout the year. This means the user will be interrupted but can quickly relog on. However, they may have to rerun some transactions from journal files and they may experience some performance degradation.

  • AE1 Business functions that can be interrupted as long as the availability of the data is insured. To the user work stops and an uncontrolled shutdown occurs. However, data availability is ensured. A backupcopy of data is available on a redundant disk and a log-based or journal file system is being used for identification and recovery of incomplete transactions.

  • AE0 Business functions that can be interrupted and where the availability of the data is not essential. To the user work stops and uncontrolled shutdown occurs. Data may be lost or corrupted.

  • * Disaster Recovery capability is a horizontal availability feature that is applicable to any of the Availability Environments (AEs). It provides for remote backup of the information system and makes it safe from disasters such as an earthquake fire, flood, hurricane, power failure, vandalism, or an act of terrorism.

Source: http://www.hrgresearch.com/pdf/AEC%20Defintions.pdf

However, there are unofficial terms bound to the AEs in certain parts of the world as well. Take this with a grain of salt.

  • AE0 Conventional
  • AE1 Highly Reliable
  • AE2 High Availability
  • AE3 Fault Resilient
  • AE4 Fault Tolerant

There are also so-called availability classes in some parts of the world. And if you look at the definition, you can see that there is an orientation towards the list above. (Salt please)

  • Availability class 1 (90%)
  • Availability class 2 Highly Reliable (99%)
  • Availability class 3 High Availability (99,9%)
  • Availability class 4 Fault Resilient (99,99%)
  • Availability class 5 Fault Tolerant (99,999%)
  • Availability class 5 Disaster Tolerant (99,999%)

Availability class 1 with 90% is removed from that list as 90% is considered conventional and with that outside the scope of that list. A system is considered "highly available" if there is a guaranteed downtime of less than an hour, which is also reflected in that list by 99,99% (~ 53 minutes).

5
  • Or maybe there even is an official definition of how much availability constitues "high availability", by the exact term. I have not heard of it, nor do I find anything useful about it on the internet.
    – Daniel
    May 19, 2015 at 20:57
  • 2
    I would argue that availability class 2 is normal today - that is what you can get with a normal computer. 1% downtime is more than 3 days per year.
    – TomTom
    May 20, 2015 at 6:18
  • 2
    @TomTom, while that would be great, it depends on what the company is willing to invest. They always want 100% availability, but in the end, it's the investment they are willing to make that defines the uptime. And if you have a customer that runs on hardware without service agreements (e.g. without an HP CarePack) a defective mainboard can become a real problem and easily give you an outage of two days if it's unavailable. That makes it difficult to achieve an availability of 99%. Bosses of smaller companies can be stubborn and consultation resistant to levels beyond our understanding. 😋
    – Daniel
    May 20, 2015 at 7:14
  • wrong. 99% HA requires no investment. Heck, a computer shop around the corner can replace whatever fails in a day and things do not fail that often. It is past 99.9% things get complicated. But 99% is "know how to order a replacement computer on any retail outlet".
    – TomTom
    May 20, 2015 at 7:54
  • @TomTom if you personally think that you can achieve 99% availability without any investment, so be it. But throwing a "wrong" at me for saying that it depends on the customer on what "normal" is, oversimpliyfies it. It's not always computer that fails, but also servers, services, printers, switches, etc. And what do you guarantee? High availability per server? Per Service? For all devices alltogether? Depends on the customer, depends on the service level agreement …
    – Daniel
    May 20, 2015 at 8:26
2

There is an article on wikipedia that describes high availability term. It is considered as one characteristic of a system. Also, you can find there already calculated percentages.

For example, availability percentage of five nines (99.999%) means max down time of 5.26 minutes per year.

1

Highly Available means (in just few words) you have some redundancy to continue you service operative in the event of a failure of some component.

The formula you wrote is to calculate the % of availability of a service. Most providers sign a SLA with clients, in order to penalize billing if this SLA is not accomplished.

2
  • 2
    My question was to determine if there is an precise definition of how much availability constitutes High Availability
    – Aditya K
    May 19, 2015 at 15:21
  • 2
    For me this is SLA (determine the %), you can have a system wihout HA with a 100% of availability and at the same time a system with HA with a 90% of availability. Also in SLA often is not only the % of availability agreed but also performance parameters (transactions done in less than x time). May 19, 2015 at 15:34

Not the answer you're looking for? Browse other questions tagged .