This week, major companies that use Amazon.com’s data services learned a hard lesson about how Amazon.com’s complexity and market dominance make it difficult for them to back up their data with other providers. Experts and analysts told Reuters.
Amazon claimed that Tuesday’s outage was caused by “an impairment of several networks devices” in its Amazon Web Services Virginia data center region. The outage temporarily disrupted streaming platforms Netflix Inc. and Disney+, trading app Robinhood Markets Inc, and Amazon’s own ecommerce site, which heavily uses AWS.
A spokesperson for Amazon told Reuters Wednesday that the issues were resolved.
AWS’s “US-EAST-1” network problem caused massive damage in one region, which AWS calls “US”. This shows how difficult it can be for companies to spread cloud computing.
According to research firm IDC Amazon holds 24.1% of the global market. Rivals such as Microsoft Corp, Alphabet’s Google Inc, and Oracle Corp are trying hard to convince AWS customers to use a portion of their cloud computing services, often as a backup.
However, it is not easy to create a complex online service that can be moved from one provider to the other in an emergency, according Naveen Chhabra (a senior analyst at research firm Forrester). AWS is not a single “cloud”, but a collection of hundreds of services. These services range from basic building blocks like storage and computing power to more advanced services like high-speed database access and artificial intelligence training.
Chhabra stated that a website could use many of these services, each of them necessary for it to function. It is difficult to backup another cloud provider as some services are proprietary to AWS, while others work in a different way at another provider.
It’s like asking, “Can I put an SUV body onto a sedan chassis?” If everything is exactly the same and lines up, maybe. Chhabra stated that there is no guarantee.
AWS charges higher “egress fees” for data that is sent to its cloud. This makes it difficult for businesses to diversify.
Matthew Prince, chief executive at Cloudflare Inc, stated that this “amplifyes issues like this (outage),” Matthew Prince said. “A more resilient cloud is one in which egress fees can be eliminated and customers can be multicloud. This would increase customers’ faith in the cloud, I believe.
DEPENDENCIES IN A REGION
Angelique Medina (head of product market at Cisco Systems Inc’s ThousandEyes) said that AWS has critical “dependencies” within its services. These dependencies can cause one service to fail while another fails. Because AWS’s complex services often rely on its more basic services, this is why. A basic function such as networking can lead to problems for other services that depend on it.
AWS stated that Tuesday’s incident was delayed by the outage.
Medina stated that AWS may also have critical services clustered within its US-EAST-1 area, where another outage last summer also had a widespread impact.
Medina stated that “that’s where most of their critical dependencies were located historically.” “They’ve changed over time.”
Chhabra, a Forrester analyst, stated that Amazon has done a lot to make its services resilient. Amazon doesn’t do what it says it does for its customers: It doesn’t build applications that can withstand outages by tapping multiple providers or locations.
This can be very beneficial.
