Users must architect cloud-native applications to reduce the impact of cloud outages
As many were reminded amid major data centre outages due to Europe’s massive July heatwave, deploying apps in the cloud doesn’t remove the risk of critical failures.
Given these events, Uptime Institute’s newly released Intelligence Report – “Public cloud costs versus resiliency” – contains timely and compelling research for the industry.
The report quantifies the costs, resiliency, and outage compensation of different cloud application architectures.
To reduce the impact of more frequent and widespread cloud outages, users must architect cloud-native applications that are resilient to the failure of providers’ virtual machines, availability zones and regions. In the event of a service interruption, outage compensation from cloud providers is highly unlikely to cover the cloud costs relative to the application, let alone the business losses. This report quantifies the costs of several common architectures for building resiliency for a stateless cloud application and determines the level of protection and compensation for each application.
Simplicity and flexibility are significant drivers of value for using a public cloud. However, building applications in the public cloud that are resilient, performant, and compliant is more complex. Unlike in privately owned and colocation data centres, cloud users have no visibility or control over the data centre — a cloud provider supplies the data centre, server and middleware or software.
Providers offer a range of optional tools and services for users to architect resilient applications, including geographically distributed locations and load-balancing services. The onus is on cloud users to use these tools and services to design and implement applications that meet their organization’s resiliency requirements. Cloud providers’ service level agreements (SLAs) and published reference designs offer some assurances of performance. But these assurances are focused on their services, not their users’ applications, and cloud SLAs provide minimal compensation in the event of an outage.
In the event of a service interruption, there are multiple design architectures that can provide resiliency to users. Each architecture varies in terms of SLA guarantees, recovery time, level of protection and cost. Determining which architecture best meets resiliency and cost requirements is not trivial. How much does an increase in cloud resiliency cost?
This report details several public cloud architectures that balance application traffic across virtual machines (VMs), availability zones and regions, and provides indicative costs and resiliency metrics for these architectures. This is the first in a series of reports from Uptime Institute Intelligence analysis cloud service resiliency.