Forums

Articles
Create
cancel
Showing results for 
Search instead for 
Did you mean: 

m5 vs m4 instance types on AWS for Jira DC

Greg Warner (Amazon) March 26, 2018

Does anyone have any first hand experience using the newer M5 instance types on AWS? I don't foresee any problems migrating from M4 to M5.

I'm doing budget forecasting and we currently run Jira DC in a 4 node cluster of M4. I intend to reserve 4 M5 instances in the next year.

3 answers

0 votes
Greg Warner (Amazon) April 3, 2018

As a follow up to this. I have deployed M5 instance types and I am running my performance test scripts. Early indications are good.

David Yu
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
April 16, 2018

Hi Greg, did you also consider C5 instances as well? It seems like it has a good ratio of vCPU to memory that might benefit in our usage scenario where we get large occasional bursts of activity.

0 votes
Craig Castle-Mead
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 26, 2018

I’ve started cutting some m4s to m5s and new workloads appear to be going ok. There’s some documentation around ensuring your kernel supports enhanced network interfaces (ENI) and checking fstab is referencing labels not devices before changing to an m5, but even after confirming this was the case, after changing the class on two servers, they didn’t boot fully. Need to do some more testing as I’m sure there’s something I’m missing, but for now it wasn’t a straight change. 

Need to get some more time sorting this out as our m4 RIs are up and want to purchase m5s, so happy to report back. If you figure anything out about changing, would be great to hear as well. FYI - nodes have been Ubuntu 14 or 16 so far. 

 

 

CCM

Boris Berenberg - Atlas Authority
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 27, 2018

Why wouldn't you deploy a clean AMI and re-provision it using Puppet or the like instead of migrating your existing one?

Uhub Admin March 27, 2018

.

Craig Castle-Mead
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 27, 2018

Doing that where possible, but we still have some snowflakes.

0 votes
Boris Berenberg - Atlas Authority
Rising Star
Rising Star
Rising Stars are recognized for providing high-quality answers to other users. Rising Stars receive a certificate of achievement and are on the path to becoming Community Leaders.
March 26, 2018

What is your concern? That they don't have enough burn in as a hardware infra class?

Greg Warner (Amazon) April 3, 2018

The concern was only any intricacies of a new instance type. I have since deployed 4 EC2 M5 instances in DC configuration. I am moving to reserved pricing and it's a large commitment for an instance type that might not be widely used across Atlassian customers.

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events