|By Toddy Mladenov||
|June 16, 2014 11:28 AM EDT||
Last week's Joyent outage brought us thinking how many IT teams make the effort what is the meaningful downtime that will not have significant impact on their business. In this post I will not discuss this particular outage although it is a yet another good example for improving the IT practices and processes but will concentrate on an important step in the Business Impact Analysis (BIA) that is a prerequisite for Disaster Recovery - namely the Cost of Downtime.
Very often because the lack of understanding of the overall IT application portfolio the cost of downtime is calculated using made up numbers or numbers for the whole company. Each application must be considered separately and the analysis must be done per application - there is no one size fits all. For now though let's take a simple example: company that has a revenue of $5o0M and 1500 employees and experiences outage that impacts 50% of it's employees and 30% of their revenue during the downtime.
The first thing you need to do in order to calculate the cost per hour of downtime is to calculate the labor cost per hour. Let's assume that the annual benefits per employee for our fictitious company are $75k and each employee works on average 1,920h per year (assuming 2 weeks vacation and 10 holidays per year). The hourly labor cost for the company will be:
1500 employees X $75K = $112.5M / 1920h = ~$58,600
Because we previously said that only 50% of the employees will be affected the cost per our of downtime will be:
$58,600 X 50% = $29,300
The next thing that you need to do is to calculate the revenue loss per hour of downtime. Another assumption that we need to make here is that our company generates revenue 5 days a week and it is closed only for the holidays, which means that the company generates revenue 2,000 a year. From here our calculation is:
$500M / 2,000h = $250K/h
and we need to multiply this by 30% assumed loss of revenue:
$250K * 30% = $75,000
The total loss to our company for one hour of downtime is determined by combining both numbers above:
$58,000 labor loss per our + $75,000 revenue loss per hour = $133,000
However we do not stop here! Our ultimate goal is to determine how much downtime can we afford for this application. And for that you need to some more financial calculations, which involve not only the revenue but also the profits unless you want to wipe out all the profits with this one downtime. Let's assume that the company's profits are 10% and management decided that the acceptable loss of profits from availability issues with this particular application is 0.1%:
$1B * 10% profits = $100M * 0.1% = $100,000
Dividing both numbers (acceptable loss and cost of downtime per hour) gives us the maximum tolerable downtime (MTD) for the application:
$100,000 / $133,000 = ~0.75 * 60 mins = ~45 min
If you are curious such application must have uptime of 99.991445% in order to satisfy the above requirements.
The overall lesson here is that determining the cost of downtime is not only application dependent but also requires a solid knowledge of the company financials and therefore good collaboration between IT and business owners.
Dec. 4, 2016 04:45 AM EST Reads: 4,938
Dec. 4, 2016 04:30 AM EST Reads: 1,543
Dec. 4, 2016 04:30 AM EST Reads: 820
Dec. 4, 2016 04:00 AM EST Reads: 4,940
Dec. 4, 2016 04:00 AM EST Reads: 6,219
Dec. 4, 2016 03:45 AM EST Reads: 1,544
Dec. 4, 2016 02:15 AM EST Reads: 855
Dec. 4, 2016 02:00 AM EST Reads: 3,782
Dec. 4, 2016 12:30 AM EST Reads: 1,783
Dec. 3, 2016 11:00 PM EST Reads: 4,154
Dec. 3, 2016 11:00 PM EST Reads: 960
Dec. 3, 2016 09:30 PM EST Reads: 1,611
Dec. 3, 2016 09:30 PM EST Reads: 1,773
Dec. 3, 2016 08:00 PM EST Reads: 3,958
In his session at 19th Cloud Expo, Claude Remillard, Principal Program Manager in Developer Division at Microsoft, contrasted how his team used config as code and immutable patterns for continuous delivery of microservices and apps to the cloud. He showed how the immutable patterns helps developers do away with most of the complexity of config as code-enabling scenarios such as rollback, zero downtime upgrades with far greater simplicity. He also demoed building immutable pipelines in the cloud ...
Dec. 3, 2016 08:00 PM EST Reads: 1,748