By Gustavo Herrera, Site Reliability Engineering Director, Orion Health
Twitter: @OrionHealth
As increasing numbers of organizations move to the cloud, software vendors pursuit of perfection is intensifying and nowhere is it more obvious than in their search for the elusive ‘five nines’. But is this costly goal worth the money for consumers?
The often-discussed concept of five nines (represented numerically as 99.999%) is a particularly appealing yet currently unrealistic goal for software providers. Not content with the traditional four nines, or 99.99%, of product availability/uptime (time the software is operational) a number of providers are starting to set stability guarantees at up to three decimal places.
The extra decimal point may seem unwarranted to people who would be happy with 99.9% uptime —if not 99%—but to those in the IT world the extra point demonstrates a costly path to perfection (or, if you’re cynical, a new marketing ploy).
But is the all the extra cost and effort worth an extra decimal point?
Well, many of the major cloud providers (Microsoft, Amazon, and Google) offer standard guarantees of uptime at 99.0% (minimum) which indicates that maybe those five consecutive numerals are not nearly as important as vendors think*.
The five nines do look good though—and not just because of the aesthetic quality of the repetition—as healthcare providers desire incredibly stable systems it is worth pursing as many nines as possible. However, if five nines are achievable it may not be that practical.
For instance, a 99.999% uptime equates to just 5 minutes and 15.6 seconds of service interruption (both planned and unplanned) a year, yet software updates and patches alone often take longer than five minutes. So, in a single update a provider can be knocked away from that elusive goal.
Then there is the issue of pricing: high availability results in high cost. Using the above example, providers need to work out how to make their system available while keeping them updated and upgraded at the same time. Does the system you use really need to be operational at such a high percentage (with just over five minutes of downtime a year) or could it have allowances to be down for ten minutes a year or twenty?
Providers really need to ask themselves: what is the worst outage length the organization could sustain without impacting patients. Once this has been determined, it could become apparent that the industry standard, guaranteed minimum of 99%, may be more than enough.
It’s easy to get swept up in the marketing and the high numbers but the reality is not every organization needs to have such high availability. Some organizations don’t use systems ‘after hours’ and others don’t necessarily use their systems with urgency, so a 99.999% availability could just leave them with unneeded expenditure. When looking for a cloud provider take note of the Service Level Agreements around uptime but also look into the support services provided, so that you make sure you don’t languish in the 0.001% downtime.
*It should be noted that all of the major cloud providers also offer the ability to purchase higher —and, in one case, lower—guarantees of uptime.
This article was originally published on Orion Health and is republished here with permission.