Fault Tolerance and high availability application architecture based on AWS cloud services

Fault Tolerance and high availability application architecture based on AWS cloud services

AWS provides reliable, fault-tolerant, and highly available system infrastructure and services built on the cloud. Special attention from developers is required. These features have been designed for AWS services, but must be clearly and accurately used.

Amazon EC2 provides modules to build the target infrastructure. They are not fault-tolerant. Hard disks, power supplies, and even racks can be dropped. The combination of features in this article can achieve fault tolerance and high availability.

1. Server Load balancer is an effective way to increase system availability. If the server Load balancer Instance becomes invalid, it can be replaced seamlessly without affecting the operation of other instances. Elastic Load Balancing can be used to balance the load of instances in multiple zones in a region.

2. AZS are specially designed to isolate different geographical locations from other AZs faults. By placing Amazon EC2 instances in multiple AZs, you can prevent spof of applications. It is very important to run independent applications on multiple AZS, whether in the same physical area or between different physical areas. If a zone becomes invalid, applications in other zones can continue to run. When designing a similar system, you must have a good understanding of Zone dependencies.

3. Elastic IP addresses can be mapped to multiple instances in a physical area. They are associated with AWS accounts, but not with the lifecycle of a special instance or an instance. Elastic IP addresses can be quickly remapped to another running instance, or a new replacement instance can be started to implement failover between the host and zone. Reserved instances can help ensure that this capability can be implemented in another region.

Without proper backup, replication, or re-creation of data, valuable data should not be stored only on one storage instance. The off-instance persistent storage volume provided by EBS is more persistent than the on-instance Storage. The EBS volume is automatically copied within an availability zone. To further enhance durability, you can create timely snapshots in S3, which can be copied to multiple zones. The EBS volume is bound to a specific AZ, while the snapshot is bound to a specific physical region. Snapshot can be used to create a snapshot in any zone in the same physical zone. This is an effective way to handle disk faults or other host-level problems and AZ-related problems. Snapshots are incremental, so it is best to keep the most recent snapshots.

This article permanently updates the link address: