Rethinking redundancy in the data center

Many public cloud providers usually inevitably encounter catastrophic interruptions in their daily operations, and IT managers need to learn from the failures and lessons of the cloud platform and apply them to their internal infrastructure.

Cloud platforms (especially large public cloud platforms) have multiple levels of redundancy, but none of them can withstand unexpected downtime. Cloud computing platforms, like data centers, will encounter problems that cause downtime.

Data center redundancy

Hardware failures and software errors deployed within an organization are inevitable, but public clouds provide IT managers with a way to deal with these problems. The public cloud has availability sets, redundant data centers, availability zones, and cloud zones, enabling managers to better plan business operations in the event of an outage. These strategies are important, but administrators must design reliable applications to take advantage of public cloud redundancy.

The resilience of the organization’s IT assets depends more on the application rather than the infrastructure, which is different from the traditional way of thinking. Over the past two decades, administrators have maintained resilience through the backup, replication, and other infrastructure-centric technologies. But for most public cloud platforms, this strategy is no longer effective.

In recent years, cloud computing providers have suffered more losses than they can afford. When cloud services provided by cloud computing providers are disrupted, industry vendors and users seem to be learning how to design local cloud redundancy. Managers hope to apply these lessons to traditional virtualized data center redundancy.

Rethinking redundancy in the data center

For most traditional organizations, some basic services are designed to prevent downtime. Service designs such as network time protocol and network routing are usually highly redundant. However, just because the system should be redundant does not mean that its configuration is fully prepared to take advantage of natural redundancy.

Some core services do not have highly available options. IT departments almost always face technical debt that must be dealt with, which requires support for legacy systems that are not behaving well. For example, some legacy application authentication systems can only exist on the server, which limits the redundant functions of the data center.

Managers should not put all eggs in one basket. For most traditional deployments, it is best to use redundant hardware inside the data center. Managers can take this strategy further by using redundant virtualized clusters that do not share systems, similar to cloud computing providers that can provide multiple availability zones. This can make applications rely on a higher level of redundancy in the data center, but it is only valuable if the business needs to guarantee that level of redundancy protection.

By implementing these strategies in parallel between the cloud platform and the data center, it may be helpful to use public cloud provider concepts (such as availability zones) when using similar traditional infrastructure. This terminology makes it easier for developers to understand infrastructure concepts because they may be familiar with similar cloud concepts.

Original Article Source from