The promise of speed, agility, and continuous innovation lures enterprises to cloud native platforms.
But to realize these benefits, cloud native enterprises need cloud observability, an emerging discipline that helps monitor and optimize the performance of cloud-based applications and infrastructure.
How does this work? Cloud observability helps CloudOps engineers and site reliability engineers (SREs) identify issues, triage those issues, then assess root cause to resolve them. As described in our earlier blog, it exposes all the interrelated elements and microservices within a cloud workflow, and helps assess how those microservices communicate with one another. Implemented well, cloud observability helps customers remediate issues faster, resulting in higher customer satisfaction and retention. In addition, cloud observability can reduce costs compared with traditional monitoring solutions such as application performance management (APM) because it retains just the most relevant metrics, traces, and logs.
In these ways, cloud observability, enabled by products such as Chronosphere, helps enterprises ensure that cloud agility does not come at the expense of stability or customer satisfaction. This blog, the second in our series, explores use cases for cloud observability. The third and final blog will examine architectural approaches.
The use cases fall into two categories: optimizing the software lifecycle, and optimizing infrastructure. In both cases, cloud observability simplifies the work of developers and CloudOps teams by extracting those few telling signals from lots of noise.
Optimize the software lifecycle
Many cloud native enterprises differentiate themselves by building custom software to engage customers and streamline operations. They employ teams of developers and CloudOps engineers that design, develop, test, deploy, and monitor cloud-native applications. To maintain a competitive advantage, these teams need a software lifecycle that supports agile application updates while maintaining quality standards.
Cloud observability helps enterprises achieve these objectives by enabling CloudOps engineers and developers to pinpoint what they need to fix or enhance. They can identify, triage, and assess the root cause of application issues in a fast, precise way.
Cloud observability helps pinpoint software bugs or feature gaps that enterprises need to fix or enhance
Suppose a cloud application fails to log in certain customers, or emails erroneous receipts to customers. Cloud observability tools alert the on-call CloudOps engineer of those microservice failures, for example by sending notifications via Slack. They triage the issue, starting with an assessment of the business impact as measured by the number of affected customers. They also notice that the alert fired shortly after the deployment of a new version of production code. Surmising this caused the issue, they roll back to the earlier version. Normal services resume. Phew!
Now the CloudOps engineer has time to team up with the developer and assess the root cause. They branch the problematic version of application code to their lab environment, which runs on the GitHub development platform. They trace all the microservices in the application—inspecting metrics for each—and identify the root cause, which in this case is a buggy microservice. They fix the bug, test the revised code branch, approve it, and merge this new release back into production.
Optimize cloud infrastructure
Cloud native enterprises also seek to optimize the infrastructure that underlies their applications. Their CloudOps engineers configure containerized cloud resources; monitor and control their capacity utilization; then monitor and respond to performance or cost issues. Both CloudOps engineers and SREs must keep a close eye on infrastructure elements such as containers, compute clusters, network resources, and data stores. They need a granular and yet comprehensible view across one or more clouds to meet enterprise requirements for speed and uptime.
As with the software lifecycle, cloud observability helps enterprises achieve their infrastructure objectives by pinpointing issues and how to fix them. They can identify, triage, and assess the root cause of infrastructure issues to ensure myriad elements support one another.
Cloud observability helps pinpoint infrastructure issues that enterprises need to resolve
Suppose a cloud application slows down without warning. A cloud observability tool alerts the CloudOps engineer, who triages the issue by measuring how many customers are using that application and determining which clusters or other elements are involved.
Now the CloudOps engineer and SRE assess the root cause. They use the cloud observability tool to trace the full workflow of microservice tasks that infrastructure elements perform to support that application. They find that one Amazon EC2 compute cluster is over utilized, thanks to a malfunctioning load balancer that flooded it with traffic. Armed with this intelligence, the CloudOps engineer or SRE resolves the issue by spinning up another compute cluster to support the load, and notifying AWS of the malfunctioning load balancer.
Cloud observability also helps optimize infrastructure by tuning resources. Armed with new intelligence about their environment, the CloudOps engineer and SRE can tune application performance by re-configuring containers, adjusting load balancer settings, or changing the bandwidth of network connections. They also can reduce costs by identifying and shutting down abandoned instances of virtual servers that slowly consume CPU cycles.
Eyes on the horizon
Cloud observability also improves planning and design. When CloudOps engineers and developers know which web pages confused customers in the last application, they can design a more intuitive application for the next product offering. When CloudOps engineers and SREs know which infrastructure microservices failed last time, they can design a more fault-tolerant architecture for their geographic expansion. Cloud observability enables them to minimize risks at the outset, so they can build and deploy effective cloud environments in less time than they would otherwise need.
By optimizing the software lifecycle and infrastructure, cloud observability might just enable enterprises to realize the benefits that first drew them to the cloud: speed, agility, and continuous innovation.
Our next blog will examine what this looks like in a typical enterprise data architecture.
Kevin Petrie is Vice President of Research with Eckerson Group, a global research and consulting firm that focuses solely on data analytics.