For many start-ups Amazon web services offers a way to grow their infrastructure quickly and cost effectively, at least in the early stages of the business. For many, as the business grows, Amazon web services can become a bit clunky and expensive.
Amazon web services or AWS is a cloud-computing platform that provides customers with a wide selection of cloud services. AWS first debuted in 2006 as a way to allow the use of online services by client-side applications and websites via various protocols like HTTP, REST or SOAP. Services are billed based on usage.
According to Amazon, EC2 is more than a virtualized hosting service. It is a fractional system and network administrator rolled into one. EC2 allows you to build your network on an “as needed” basis, one box at a time. This eliminates the expense of new hardware, reduces power and network topology costs, reconciles vendor differences and eliminates network storage issues.
EC2 isn’t all Sunshine and Roses
EC2 is a workable solution but it has some serious performance and reliability limitations that can come into play as your needs increase. Most concerning is Amazon web services whole zone failure pattern.
AWS employs multiple geographically dispersed server locations they call “availability regions.” While they are co-located, they are isolated in terms of networking and power, issues with the regional zone system include:
• Virtual hardware is not as robust as real hardware.
The average lifetime of a virtual server on EC2 is only about 200 days. After this mark, the chances it will be “retired” rise dramatically. Unfortunately, Amazon’s process is unpredictable. You may be notified of the shutdown ten days before, or after the box fails. Hardware failure in a virtualized environment is not a major problem, but the notification process can be problematic. You need to be aware of this issue and plan accordingly.
• High failure rate, requires you to spread your data geographically across several zones.
The experience with AWS has been trending towards regional failures as opposed to single machine issues. You’ll need to build redundancy into your infrastructure. This means planning for failure by duplicating data across zones. If a zone goes down and your master data is only located in that zone, you may not have time for a backup and you may lose your information.
• Multi-zone failures have been reported, to address this contingency you must spread data multi-regionally. This can significantly increase your cost.
In 2011 a major multi-zone failure occurred that was nicknamed “cloudpocalypse.” AWS region-wide instability seems to frequently have the same root cause.
The takeaway from all of this is the while AWS does offer some advantages for start-ups, the overall experience for many companies is mixed at best. As a business owner make sure to do your due diligence before moving your IT infrastructure. Cloud services are the right fit for many businesses that are transitioning online, just make sure to do your homework to avoid any unpleasant surprises!