AWS Auto Scaling Groups

My company is looking to enable auto scaling groups in AWS. Per our security team, we are not allowed to use only SNMP or WMI monitoring on instances in AWS (too many ports to open from our on prem SolarWinds environment). Are only option for monitoring in AWS is to use the SolarWinds agent. When the automation team spins up a server using Auto Scaling groups, I do not see the "server name" in the agent listings in SolarWinds. I only see ip-xx-xxx-xxx-xx.domain.com where xxx is the ip address of the instance. The server name is set as a tag in AWS.

My question to the forum is:

1.) how does your company handle auto scaling groups in AWS?

2.) is there a way in SolarWinds to get the tags listed in AWS?

3.) is there a way to automate populating custom properties using the API

4.) is there a way to automatically assign application monitoring using the API?

5.) What are your experiences with servers that get destroyed and rebuilt and how you monitoring those servers using SolarWinds?

Thanks,

Steve H

Find more posts tagged with

aws cloud

auto scaling groups

Accepted answers

All comments

hpstech

I've seen a thread on this topic and it remains a big challenge.

This topic outlines a post I recently wrote in some other thread that elaborates on the future of elastic monitoring - the heaviness of SolarWinds and other monitoring tools make it an epic challenge to monitor resources that get destroyed and re-created in the cloud based world we are headed into.

In around 15-20 years the whole WMI/SNMP monitoring thing will be long gone - but for now, we're stuck in a very heavy lifting predicament. The monitoring sphere has not adapted to the new world of dynamic resource destruction and creation.

mesverrum

I work as a consultant so I don't get to own any of the environments for more than a few weeks at a time but this is what I normally see:

1.) how does your company handle auto scaling groups in AWS? When orion scans a cloud instance it will see the new server and automatically pick it up as a Cloud VM, but the question is if you have alert logic in place for all cloud objects or not and if you need to dial in those alerts in any way? IIRC cloud objects do not currently allow for any Orion custom properties or individual thresholds but you can trigger a basic alert with logic like Cloud VM has cpu load above 90% for 30 minutes.

2.) is there a way in SolarWinds to get the tags listed in AWS? I dont believe they currently pull tags from AWS, but you could script something up to get the tags and present them one way or another in Orion if you feel fancy.

3.) is there a way to automate populating custom properties using the API - Yes, Managing Custom Properties · solarwinds/OrionSDK Wiki · GitHub

4.) is there a way to automatically assign application monitoring using the API? - Yes, OrionSDK/SAM.Application.ps1 at master · solarwinds/OrionSDK · GitHub

5.) What are your experiences with servers that get destroyed and rebuilt and how you monitoring those servers using SolarWinds? - Most of my clients fall into two buckets, if they can script they build their own functions to add in new nodes to Orion as soon as they are generated in whatever tools they use (vmware, AWS, Azure, GCP), if they can't script they only offer the business superficial monitoring of cloud objects until someone goes through the traditional/manual process of requesting them to add it as a node.

To some extent ephemeral servers that come and go (cattle) do not often "need" monitoring in the same way that traditional (pet) servers do. The application software should have all the logic in place to spin up new servers as resources become tight, and destroy them when not needed anymore or if they start spitting out errors. So from the monitoring perspective we really just need to set up a place to receive those error messages in whatever form they come and do our reporting/alerting based on that data. Then we set up application monitoring on the end result of those server clusters, such as running some kind of synthetic transaction and tracking the metrics in terms of success/failure/duration. You can get fancier if you need but it just depends on how deep into the devops weeds you would like to go. Traditional tools like Orion are still just stepping their toes in, but SolarWinds has aquired several cloud native monitoring tools lately and their APM product, for example, directly integrates with Orion to help bridge the gap.

Application Performance Monitor | SolarWinds

It's still a pretty new space so generally speaking I am still seeing a LOT of roll your own solutions to every challenge.