Cloud Deployment

The Latest

28 March 2021: AWS has a history of periodically lowering the costs of storage. But even with this typical behaviour, its recent announcement of an elastic storage option that shaves 47% off current service prices is impressive. Or is it?

The first thing to realise is that the touted savings are not apples for apples. AWS’s new storage offering is cheaper because it resides in a single-zone, rather than being replicated across multiple zones. In short, the storage has a higher risk of being unavailable, or even being lost by an outright failure. 

Why it’s Important

AWS has not hidden this difference. It makes it clear that the lower cost comes from less redundancy. Yet this architectural nuance may be overlooked when looking at ways to optimise Cloud costs.

One of the major benefits of moving to Platform-as-a-Service offerings is the increased resilience and availability of the architecture. Cloud vendors, including AWS, do suffer periodic failures within zones. Examples include the AWS Sydney outage in early 2020 and the Sydney outage in 2016 which impacted banking and e-commerce services.  

But it is important to note that even though some of Australia’s top companies were effectively taken offline by the 2016 outage, others just sailed on as if little had happened. The difference is how these companies had leveraged the redundancies available within Cloud platforms. Those that saw little impact to operations when the AWS Sydney went down had selected redundancies in all aspects of their solutions.

Who’s impacted

  • Cloud architects
  • Cloud cost/contract specialists
  • Applications architects
  • Procurement leads

What’s Next?

The lesson from previous Australian AWS outages is that organisations need to carefully match the risk of specific application downtime. This new announcement shows that significant savings (in this case 47%) are possible by accepting a greater risk profile. However, while this may be attractive from a pure cost optimisation/procurement perspective, it also needs to be tempered with an analysis of the worst case scenario, such as multiple banks being unable to process credit card payments in supermarkets for an extended period.

Related IBRS Advisory

  1. VENDORiQ: AWS second data centre in Australia
  2. Post COVID-19: Four new BCP considerations
  3. Running IT-as-a-Service Part 55: IBRS Infrastructure Maturity Model