If your business uses Amazon’s AWS S3 cloud service for website or application hosting, you may be offline today.
As reported by numerous technical journals, a portion of AWS is experiencing “high error rates”, which seem to be concentrated in portions of the Amazon network along the eastern United States.
Websites, stock trading services, financial services, medical research databases — the works — are all seeing partial- to complete outages of web applications and services. Sites are loading slowly — if at all — and rich media streams are at a standstill. In some cases, website pages are loading, but without images, style sheets, or other presentation elements, the result being browsers downloading pages that resemble a 1995-era, stark white page background filled with generic Times New Roman text.
The irony of a service believed to be “always available” was best expressed in an article by Tech Crunch:
“IsItDownRightNow.com also appears to be down as a result of the outage.” (Source listed below)
Contrary to belief, outages of AWS and similar cloud service providers happen regularly, and when they do, the issues can be widespread. This is because so much of the scale-able technology that powers cloud networks is dependent on components that function separately from one another, distributed across a wide network that may span thousands of miles, and with a requirement that all resources are working flawlessly. Any failure in a component of the cloud network can subsequently slow down, interfere, or outright bring down other components.
No technology platform is perfect, and no amount of redundancy can protect against the type of service failure that AWS is experiencing today. The fallacy of an “unstoppable, always-accessible cloud” is one reason Canvas Host has not partnered with a larger cloud provider. When compared to traditional dedicated, clustered-server, or private (smaller) cloud services, we believe the same level of performance can be achieved on a smaller scale, at the same or better cost, and with greater reliability, than by going with a large, cloud service like AWS.