How to protect your workloads from AWS outages
Michelle Yakura - December 09, 2021
In full swing of the holiday shopping and travel season, AWS yesterday experienced an outage in its US-EAST-1 region. It brought down websites and applications galore, including those of Disney+, the Associated Press, Amazon Prime, and Southwest and Delta Airlines. As cloud computing adoption continues to accelerate with the COVID-19 pandemic, enterprises are increasingly turning to public cloud, with AWS leading the pack. But with events like yesterday’s widespread AWS outage, how can enterprises protect themselves?
Besides outages, multi-cloud design patterns also enable enterprises to mitigate other risks, like when cloud vendors unexpectedly hike up their prices, retire services, or change their service policies. It’s an important approach to preventing vendor lock-in in an enterprise cloud strategy.
According to the Flexera 2021 State of the Cloud Report, enterprises are overwhelmingly embracing multi-cloud, with 92% of enterprise respondents reporting having a multi-cloud strategy. A vast majority of them, 82%, are choosing the path of hybrid cloud, combining both public and private cloud infrastructure.
Amid all the hype surrounding AWS, Azure and GCP, the widespread adoption of hybrid cloud underscores the enduring value that enterprises derive from private cloud deployments. Just last month, the OpenInfra Foundation announced a surge in the adoption of OpenStack, with more than 25 million cores of compute deployed in production worldwide. Primarily used for private clouds, OpenStack is the most widely deployed open source cloud software in the world, with 80% of clouds already in production.
Besides for telecommunications use cases like Network Functions Virtualization, OpenStack is increasingly being deployed in industries like financial services, the public sector, and retail. Non-telco companies like Walmart Labs and Workday now figure among the organizations that have deployed more than 1 million cores of OpenStack.
So just because you’re going cloud native doesn’t mean you should exclusively turn to public clouds. Conversely, a hybrid cloud strategy that includes containerized workloads on private clouds makes sense for many enterprises today.
What you need is an architecture in which interactions with each provider have been abstracted so that you can use a single API to interact with your workloads, no matter what provider hosts them. That way, should you need to suddenly move from one provider to another, your modus operandi doesn't change; only the destination does.
To sum it up: it’s important to adopt a multi-cloud strategy to protect yourself from cloud-specific risks, but without unified operations and monitoring, it’s too easy to get mired in unnecessary complexity and chaos, especially as you scale.
Mitigating Risk with Multi-Cloud
The answer lies in multi-cloud. One of the main drivers of multi-cloud adoption is risk mitigation. By distributing application components across multiple clouds, enterprises can achieve redundancy to minimize the impact of unforeseen disruptions and carry on with business as usual. If there’s a problem in one cloud, simply use a load balancer or service mesh to divert your workloads to another cloud.Besides outages, multi-cloud design patterns also enable enterprises to mitigate other risks, like when cloud vendors unexpectedly hike up their prices, retire services, or change their service policies. It’s an important approach to preventing vendor lock-in in an enterprise cloud strategy.
According to the Flexera 2021 State of the Cloud Report, enterprises are overwhelmingly embracing multi-cloud, with 92% of enterprise respondents reporting having a multi-cloud strategy. A vast majority of them, 82%, are choosing the path of hybrid cloud, combining both public and private cloud infrastructure.
Amid all the hype surrounding AWS, Azure and GCP, the widespread adoption of hybrid cloud underscores the enduring value that enterprises derive from private cloud deployments. Just last month, the OpenInfra Foundation announced a surge in the adoption of OpenStack, with more than 25 million cores of compute deployed in production worldwide. Primarily used for private clouds, OpenStack is the most widely deployed open source cloud software in the world, with 80% of clouds already in production.
Why Choose Private Clouds?
So why do enterprises continue to choose private clouds? According to the last OpenStack user survey full report, published in 2018, the top drivers of OpenStack adoption are increasing operational efficiency, accelerating the ability to innovate, and avoiding vendor lock-in. Other key reasons to deploy private clouds include cost savings, integration with on-prem data, and geographic availability. See a full list in our blog, Why the World Needs Private Clouds.Besides for telecommunications use cases like Network Functions Virtualization, OpenStack is increasingly being deployed in industries like financial services, the public sector, and retail. Non-telco companies like Walmart Labs and Workday now figure among the organizations that have deployed more than 1 million cores of OpenStack.
OpenStack and Kubernetes Together
When Kubernetes first entered the scene, some saw it as signaling the demise of OpenStack. To the contrary, OpenStack and Kubernetes are complementary technologies that are increasingly implemented together in multi-cloud deployments. According to the 2021 OpenStack User Survey, 68% of OpenStack users are also running Kubernetes to manage their applications. In particular, Kubernetes adoption is fueling increasing deployment of OpenStack bare metal clouds.So just because you’re going cloud native doesn’t mean you should exclusively turn to public clouds. Conversely, a hybrid cloud strategy that includes containerized workloads on private clouds makes sense for many enterprises today.
Achieving True Multi-Cloud
You can’t just deploy workloads on multiple clouds and leave it at that. A successful multi-cloud strategy requires unified management and monitoring and true workload portability. Otherwise, in your quest to mitigate risks like AWS outages, you needlessly introduce new problems like operational complexity and unmanageable sprawl. Effective multi-cloud implementations can simplify operations with a single pane of glass to streamline monitoring across clusters and a consistent UX across cloud providers to accelerate onboarding and ease lifecycle management.What you need is an architecture in which interactions with each provider have been abstracted so that you can use a single API to interact with your workloads, no matter what provider hosts them. That way, should you need to suddenly move from one provider to another, your modus operandi doesn't change; only the destination does.
To sum it up: it’s important to adopt a multi-cloud strategy to protect yourself from cloud-specific risks, but without unified operations and monitoring, it’s too easy to get mired in unnecessary complexity and chaos, especially as you scale.
