Class Discussion: AWS ‘crash’

In light of the recent AWS outage of its S3 service, a class discussion was raised on cloud storage, and the pros and cons of it.

Most of the information sourced for this discussion came from these websites:

WIRED: The Amazon s3 outage is what happens when one site hosts too much of the internet.

TechRepublic.com: Should the latest AWS outage scare you away from public cloud?

TechRepublic.com: 3 public cloud myths highlighted by the Amazon s3 outage.

AWS has an uptime of 99.99%. The sites state that this is equivalent to approximately 1hr downtime. However, we wanted to calculate it to determine a more accurate figure.This implies  the equation: 365*24*60*0.0001, which is days/year * hours/day * minutes/hour * downtime in decimal notation. The answer found from the calculation was 52.56 minutes.

This downtime would likely be required for maintenance, upgrades, new installations, etc. which makes logical sense for downtime to be required in this manner. In order to make this figure seem more relevant, we compared it to another service.

Next, we considered the private service known as TALOS. TALOS guarantees 100% uptime for business days. We approximated this to be 5 days per week, with each day being 8 hours long. Then we factored in 3 weeks of annual leave, and public holidays. This gave us a result of business days equating to 265 days per year, with 8 hours per day. This is 2120 hours. If we divide that by the number of hours in 1 year, we get a percentage uptime of 24.20%

One of the implications for TALOS’ low uptime is that potentially becomes cheaper to run this service not on the cloud. Whilst product costs for the necessary parts may range near $100,000 for a service that has a 3-5 year lifetime, it may be costing near $100,000 per year to run the service from the cloud.

For some businesses, the cloud is an incredibly accessible, and easy to utilize service. So what can they do to secure themselves against an incident like the S3 outage? There are a  couple of ideas that can be instigated, but they have pros and cons that need to be taken into consideration. These fixings are:

Multi-cloud services

  • Consider a business deciding to use both AWS and Azure cloud systems. Business needs a professional for both systems to ensure that any problem that arises in either system can be fixed. However, if one of the professionals is absent, then the other professional is unable to assist in place of the absent professional, as their knowledge is exclusive to their system and not the other one.
  • Hence, the business requires to have double the staff and professionals to cover any problems that arise.
  • Business has double the requirements and double the risk.

Two availability zones

  • Consider a business that decides two store their data in two different availability zones. They are then paying for both servers, which results in a very expensive cost for using cloud services.

Ultimately these fixings are more of insurances or back-up plans for a business. However, due to the increased expenditure required, these aren’t necessarily going to be beneficial to smaller business. A large business, consider one that deals with stock exchange, needs to be implementing some form of insurance or cover, as an outage from a server with all their data would cause an incredible amount of loss for them.

Ultimately Cloud computing is fantastic for start-up businesses, but is not necessarily the best option for certain business. Like any service, the pros and cons need to be weighed up before determining the course of action.

Advertisements