Companies across the world will recall February 28th 2017 as a bad day. It was a day when their patience was tested beyond anything most executives can remember in years. It was the day that the Amazon Cloud went down.
The Amazon Web Services (AWS) Simple Storage Service (S3) service now accounts for an enormous amount of Internet traffic. There are many major players in Cloud services, such as Microsoft, Google, and IBM, but AWS is the biggest. In fact, AWS is so big that their cloud revenue is about twice the total of those three other companies.
Forrester Research predicts that the cloud marketplace will be worth $236bn by 2020 – that’s up from around $95bn in 2016. So it is clear that this is an important market. It’s growing fast and it has changed the nature of IT services and IT outsourcing. Companies are now keen to buy end to end services that can be delivered over the cloud, rather than requiring infrastructure and software on site.
However, the AWS service failure in February shows that some companies have become complacent about the cloud. It was a simple human error by an engineer at AWS that took the service down for over 4 hours and affected millions of people. Why were so many companies taken by surprise and why did nobody have a backup plan?
This feature in Database magazine asks all these questions and suggests a few pointers for executives who want to mitigate against a similar disaster in future.
In general, cloud services are more reliable than setting up infrastructure inside your own office. Companies like Amazon invest enormous amounts to ensure that network and power outages are avoided. However, as we have seen, problems can happen and if your service needs to be available to customers 100% of the time, how can you offer your own guarantee when you can never be 100% sure that your own infrastructure providers will deliver?
In my opinion, the three best tips from the Database feature are:
1. Monitor services yourself – don’t just rely on your supplier. You might notice service issues before the cloud supplier ever tells you something is wrong.
2. Have an alternative provider ready – it’s expensive to duplicate, but if you really cannot afford any outage then mirroring may be required.
3. Communicate – make sure that you tell customers you have a problem so they can come back later or otherwise delay their interaction with you. By saying nothing, customers will expect a normal level of service.
IT services has been transformed by cloud based services and because it generally improves the reliability of services executives have often been blinded into believing these services can always be relied on. Mostly they can, but it’s worth remembering that even the biggest and best planned networks can fail. What would that mean for your business?