What’s the Hardest Part of Monitoring Across Multiple Customers?
Summary: Madison May initiated a discussion on the challenges faced by Managed Service Providers (MSPs) when monitoring across multiple clients, encouraging the sharing of real-world experiences to improve practices. john reid mentioned the importance of platform setup for security and customer isolation, suggesting the need for features that ensure resources are not mixed between clients while maintaining scalability. Justin Lanoue highlighted issues with insufficient filtering capabilities, lack of granular permissions, and challenges in alert rule customization, while also noting the need for better integration with Single Sign-On (SSO) and improvements in Dependent Alert Mapping. Jared Meidal responded by discussing their experience with customizing vCenter modules to support customer segmentation effectively.
Every MSP knows that monitoring one environment is very different from monitoring many. As customer counts grow, the challenges change, and not always in the ways we expect.
Let's kick off the MSP Connection by learning from each other’s real-world experiences running monitoring at scale.
What’s the hardest part of monitoring across multiple customers today?
Here are a few prompts to spark ideas. Feel free to respond to as few, or as many, of these prompts that resonate with you. This is a casual space where there is no wrong way to express your experiences.
What becomes more difficult as you add more customers?
Is it alert noise, onboarding consistency, RCA quality, or something else?
Which part of monitoring takes more time or effort than it should?
Where do things tend to break down first as you scale?
What’s one challenge you’re still actively trying to improve?
You don’t need a perfect or polished answer. Short, honest insights are more than enough.
The goal of this community is to share what’s working, what’s hard, and what MSPs don’t always get to talk about openly-without the noise of anything else.
This space is also for asking the right questions that drive the shared value of this community, not to prescribe it. If you ever find yourself thinking that a different question would more accurately express your challenges in this community, let's hear it!
Looking forward to learning from and with each other as we get started.
Justin Lanoue
·7 months agoOne hurdle for us in an MSP sense is not having the ability to filter on instances / instance groups enough.
For example, we host clients in our hosted vCenter and separate clients into folders with an identifier tag.
vCenter
- Virtual Machines
- - Client A (Instance Group)
- - - Client A VM
- - Client B (Instance Group)
- - - Client B VM
The vCenter is under our tenant in LogicMonitor and we do have logic to utilize the tag to put the ticket under the client in our PSA which is fine.
But there would be no easy way for me to have an alert rule for a specific Client VM because alert rules do not support instance properties or instance groups to reference, just resource properties and instance names.
There are also not enough granular permissions. I may want someone in our company to be able to run active discovery for something or add a custom property but not have full manage permissions which allows them to add/delete resources and everything else.
A 3rd thing while I'm here would be the ability to white label a client's own SSO integration without them needing their own portal.
Lastly, we provide circuits to clients so we may get an alert on our device and an alert on the client device/collector going down. There doesn't seem to be an easy way to associate them together with topology mappings. I would love for dependant alert mapping to be able to associate like an interface/bgp/ospf going down with a resource and suppress / child them automatically but that is currently limited as per the documentation.
Note: Dependent Alert Mapping is currently limited to resources and does not extend to instances. For example, a down interface on which other devices are dependent for connectivity will not trigger Dependent Alert Mapping.
Jared Meidal
·7 months agoI have needed to depend on customizing our vCenter modules to group by auto.resource_pool properties, or filter by them, to segment our customers and align them with how the datacenter team operates. Have you found this beneficial @Justin Lanoue ?
Justin Lanoue
·7 months ago · EditedWe do a similar thing but with VM tags.
It was annoying when the vCenter modules moved from 1 property per tag to a comma separated list, but I just reverse engineered the old module logic.
john reid
·8 months agoIts mission critical that our platform is set up well. Ensuring security and isolation / separation of the customers in all ways is vital. Tenant ID helps, alot but EVERY element needs to have easy and quick separation. The easiest way to do this is to have a different platform setup for every customer, but this doesnt scale. So for us at the moment its down to best practice and templates and structure. Some new feature which would make it impossible to mix resources / users with different tenant ID would be very useful.