Embrace an app-centric approach to monitoring in the cloud
Cloud computing minimizes the need for IT teams to worry about infrastructure hardware -- which means their monitoring strategies can, and should, hone in on apps and the user experience.
While IT monitoring tools have improved, especially with expanded capabilities for predictive analytics and machine learning, their core functionality remains the same. The biggest monitoring challenge IT teams face today is to keep pace with modern and distributed application designs.
As applications evolve, so, too, do the platforms on which enterprises develop and run them. The cloud has become a prominent environment for enterprise application development and hosting -- and also one that requires IT teams to adjust their monitoring strategies.
From infrastructure to applications
In many cases, the tools an enterprise already has in place can apply to monitoring in the cloud -- but with limited effect.
With both SaaS and PaaS, admins no longer need to monitor the underlying infrastructure, as the cloud provider is responsible for that task. This means, for IT teams, the hypervisor and hardware suddenly fall out of scope -- a significant transition, given how many operational resources enterprises traditionally put toward infrastructure monitoring in the data center.
However, this also serves as a wake-up call for what admins really need to focus on when monitoring in the cloud: applications.
A cloud monitoring strategy for SaaS vs. PaaS
Because infrastructure has historically driven many IT monitoring strategies, metrics around CPU and memory usage were a main focus. These statistics, however, did not represent the user experience with applications. While these infrastructure metrics still have value, monitoring must be a top-down effort, starting with the application -- and cloud migration is an ideal time to make that shift.
When it starts at the application layer, a cloud monitoring strategy shouldn't be all that different for SaaS vs. PaaS. The biggest difference is that SaaS applications typically have vendor-provided monitoring dashboards, which might have limited functionality. For example, sometimes these tools lack metrics for usage trends and response time. To rely solely on a SaaS provider to offer these metrics is to put all your eggs in a single basket.
Instead, supplement these SaaS dashboards with third-party cloud app monitoring tools, such as SolarWinds' offering for Office 365. These tools provide an impartial view that focuses more on the end-user experience than the technology behind it.
With PaaS, an enterprise still needs to monitor the application stack but also go a layer deeper. More traditional OS-level monitoring tools will work -- with a catch. Log aggregation and events are key metrics to monitor, but performance metrics in the cloud have the same limitations as they do in virtualization: You don't see the underlying hardware. Instead, you see which resources the hypervisor allocates.
Data that identifies memory or CPU bottlenecks still has value for monitoring in the cloud, as admins can use it to properly right-size their virtual machines and workloads. While cloud monitoring tools won't go as deep into the infrastructure as they would for an on-premises deployment, it's how IT uses that data that matters. Since the application of this data is different, the way in which admins view it should also be different: focus more on event trends than specific events or spikes. Event-driven issues -- such as an application component crash or server failure -- still occur and require a reaction, but the cloud becomes more of a right-sizing concern, where trending data is key.
Several tools, such as Amazon CloudWatch and Azure Monitor, give a detailed view of cloud environments, but remember the separation of duties -- third-party options offer an outside view of the PaaS environment.
Lastly, as more applications go to PaaS or SaaS, network connections between the data center and the cloud become more critical. Ensure monitoring tools have the ability to monitor these connections -- though, a "single-pane-of-glass view" remains largely elusive.
Editor's note: This article is in a three-part series on adjusting IT operations tasks for the cloud. Stay tuned for expert advice on patching and security.
AWS monitoring best practices extend beyond CloudWatch
Compare Grafana vs. Datadog for IT monitoring
Learn how New Relic works, and when to use it for IT monitoring