cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Recommended AWS EC2 Instance Types and Storage Types

eric_eiswerth
Dynatrace Advisor
Dynatrace Advisor

When deploying Dynatrace Managed to AWS, what are the recommended EC2 instance types (general purpose, compute optimized, memory optimized... https://aws.amazon.com/ec2/instance-types/)?

What is the recommend storage volume type? EBS, HDD, or SSD?

This would be for a multi-node cluster handling less than 50 OneAgents, but would need to be scaled out eventually.

Thanks,

Eric

6 REPLIES 6

pahofmann
DynaMight Guru
DynaMight Guru

You can find the requirements for the different cluster node sizes in the documentation. It also includes information about IOPS for the disk choice. (Note that you need three nodes for high availability).

For < 50 hosts one small node, or three for high availability, would be more then enough. You should also check what your planned maximum number of hosts in the future is, as you can only scale out, to a max of 10 nodes, then you have to increase the node size.

Dynatrace Certified Master - Dynatrace Partner - 360Performance.net

krzysztof_szynt
Dynatrace Advisor
Dynatrace Advisor

Regarding instance types: general purpose family M4/M5 is the most flexible option (e.g. m4.2xlarge with 32 GB RAM), alternatively, memory optimized R4 is also good for lighter workloads (memory is often the biggest constraint - e.g. r4.2xlarge or even r4.xlarge). Note that unless the customer is buying reserved instances, it is easy to upgrade to bigger instance type later on.

Regarding disks, definitely SSDs and preferably separate EBS volumes for different kind of storage (Transaction, Long term, etc.)

Hey Krzysztof,

per documentation it is not possible to have different node sizes in the cluster. Is it possible though, to increase the size of all cluster machines to a bigger node size at the same time?

Dynatrace Certified Master - Dynatrace Partner - 360Performance.net

it is not recommended to have different sizes, but during transition period it is perfectly fine to upsize the cluster nodes in a rolling fashion. This way you can avoid the cluster downtime.

Note: the new memory settings will be applied only after next cluster update. Additional steps would be needed if the change is from bigger to smaller size.

Thanks, that is very important to know! It would be great if this information could be added to the documentation.

Dynatrace Certified Master - Dynatrace Partner - 360Performance.net

eric_eiswerth
Dynatrace Advisor
Dynatrace Advisor
Thank you @Patrick H

, and @Krzysztof S. Exactly what I was looking for.

Featured Posts