Dell APEX Block storage for AWS best practices include:
- Multi AZ deployments to mitigate workload downtime if an AZ becomes unavailable. If availability is not critical, the deployment can be in a single AZ.
- Use the Amazon EC2 i3en instance family having NVMe SSDs to optimize for low latency and high I/O performance.
- Because Amazon instance store disks are ephemeral and their data is deleted when the system is shut down (not including reboots), plan accordingly by using system backup procedures before shutting down the system.
- Use the Amazon EC2 c5n instance family when deploying PowerFlex with EBS storage to provide more flexibility for frequent storage instances shutdown without needing to backup the system first, for example, if the system is brought down every weekend.
- Ensure that you understand the scope of AWS resources and their availability. For example, the Amazon instance store (NVMe storage) is ephemeral but remains persistent if an instance reboots and is protected by PowerFlex SDS RAID 1 mesh in which one SDS instance is shut down at a time. When using EBS storage, EBS volumes are associated with a specific AZ. If that AZ becomes unavailable, the storage is also unavailable. Deploying PowerFlex across AZs protects from this situation. S3 is a regional resource and backups to S3 are accessible if the region protects application workloads across regions or uses a replication strategy at the database, storage, or backup (such as DDVE) levels.
- Deploy the PowerFlex system in a private subnet with a private address.