Sergey Nivens - Fotolia
The tools needed to monitor cloud performance aren't one size fits all. Instead, the monitoring options revolve around two architectural considerations: public vs. private clouds and which type of cloud service you need to monitor.
When seeking to monitor performance in your cloud infrastructure, understanding the architecture differences can help clarify where potential limitations occur.
The difference between public and private clouds boils down to responsibility. With public clouds, a business will partner with a public cloud service provider to host and own the responsibility of managing much of the underlying infrastructure.
Private clouds, on the other hand, are designed and built either on premises or in a service provider colocation space. In either case, the business manages the data center infrastructure using cloud management and orchestration software.
The biggest difference between public and private cloud monitoring is the level of flexibility one has when selecting monitoring tools. When working with a private cloud, your options for monitoring are endless, as you have full control over the underlying infrastructure on which the cloud platform is built. Thus, you can install any type of performance monitoring tools you choose.
While working in public clouds, however, you're far more restricted and can only use the tools the service provider offers or allows.
Monitoring SaaS, PaaS and IaaS platforms
Most businesses find public clouds are more cost-effective. The next decision to make is what level of service you want your public cloud service provider to offer. These service levels commonly fall into one of three categories: SaaS, PaaS and IaaS.
SaaS, PaaS and IaaS -- Oh my!
This primer explains how SaaS, PaaS and IaaS architectures differ and provides useful information to help your organization determine which cloud service is the right choice.
SaaS monitoring. When investigating how to monitor SaaS cloud performance, you'll immediately notice that your options and methods are limited compared to other cloud models. This is because the service provider controls the cloud from the infrastructure all the way up to the application.
Monitoring SaaS clouds boils down to the total experience from the end user to the SaaS cloud. Thus, monitoring the end user's path from the ISP to the public cloud is of the utmost importance. This includes monitoring cloud service latency, domain name system speed and accessibility, and content delivery network responsiveness from an end-user perspective.
Additionally, security tools -- such as secure web gateways, client or cloud-based firewalls, and data loss prevention tools -- can affect SaaS performance and must be monitored, as well.
PaaS monitoring. Stepping up from a SaaS model to a PaaS architecture gives customer administrators a bit more visibility into the cloud. PaaS platforms let administrators monitor the path from the end user to the cloud, as well as to services within the cloud.
Administrators can monitor and create baselines for application workload metrics, such as CPU, memory, minimum/maximum service requests and service response times. This data can then be used to keep an eye on performance and notify administrators when a metric is trending above or below normal performance baseline levels.
IaaS monitoring. With IaaS, you gain all the visibility of SaaS and PaaS models, along with added visibility down to the server OS level. In most cases with IaaS, your performance visibility from the end user to the application is identical to that of an on-premises deployment. This means you gain visibility into the performance of the OS -- including the memory, CPU and storage capacity allocated for each component. Performance monitoring takes on a new role in this situation, as IT teams can better right-size servers to meet cloud performance expectations.
An IaaS shortcoming, however, is the visibility IaaS provides into a cloud-provisioned network is somewhat deceiving. The reason for this is the cloud network visible to IaaS customers is a virtualized overlay of a much larger cloud network managed by the provider. Thus, problems within the underlying network can cause performance issues -- yet they are hidden from view. That's why it's important to remember that even with the utmost visibility provided by public IaaS clouds, it's still not the level of visibility you gain when managing your own on-premises data center network.
Consider performance monitoring when choosing a cloud model
As you can see, the level of performance monitoring visibility changes depending on which cloud model you use. Within private clouds, performance monitoring capabilities are virtually limitless. Yet, once you begin handing over the responsibility of managing infrastructure components to a third-party cloud service provider, performance visibility diminishes. Thus, the level of performance monitoring required by a business should be a key factor in what type of cloud architecture it chooses.
Discover strategies for WAN and SD-WAN performance monitoring
How to monitor performance in automated and virtual networks