Sizing EC2 for Oracle Fusion Middleware

Question

I am trying to understand what size instances I would need to run the Oracle Fusion Middleware products within AWS. Currently, they sit on premise and they have massive boxes that are not overly taxed. The issue is that the way they have them configured on VMware; they have a huge heap file that is designated at startup, which makes the box look like it is using more resources than it actually is.

Does anyone have insight into how this software might run in AWS and how to properly size the target EC2?

Answer

Hello,
not sure what heap file refers to (VMFS? JVM? Other?), so I go for a more generic answer.
You could follow two different approaches:

1) Size your instances 1:1 to your on-prem machines (in terms of CPU, RAM and storage) and down-size later, when you have real consumption data. AWS makes changing your EC2 shape very easy and many AWS customers use this approach.
You could use our [Amazon EC2 instance recommendations](https://docs.aws.amazon.com/migrationhub/latest/ug/ec2-recommendations.html) in combination with Discovery Connector for this.

2) Try to right-size from the beginning using load metrics from your current system. You might still want to down- or upsize your EC2 instance later, though.
For load metrics, you could get the information on average CPU load, used RAM, I/O throughput, IOPS from your monitoring system or from the operating system (e.g. "sar" on Linux). Use this data to see whether your system is under-utilized (e.g., average CPU load being much smaller than provisioned CPU) and a smaller EC2 shape makes sense.
See also our documentation chapter [Right Size Using Performance Data](https://docs.aws.amazon.com/whitepapers/latest/cost-optimization-right-sizing/tips-for-right-sizing-your-workloads.html#right-size-using-performance-data).

Sizing EC2 for Oracle Fusion Middleware

Relevant content