Two key takeaways from last month’s DockerCon: containers have gone mainstream and system administrators are paying close attention. Long a haven for developers, this year’s DockerCon in Austin hosted more system administrators than I’d ever seen, and they all seemed to be asking one question: We get it that containers are here to stay; how do we operationalize them?
Fortunately, VMware has an easy answer: pretty much the same way you have put any other application into production when you are using our cloud management platform. VMware vCenter and vSphere Integrated Containers (“VIC”) enable system administrators to manage an increasingly diverse application ecosystem, including microservices, cloud-native applications and traditional, monolithic ones, across public and private clouds, all from a common platform. This greatly eases the administrative burden because it means you can manage containers using the same, familiar tools you use to manage your VMs.
The challenge of how to monitor containers is a bit more complex. As you might suspect, I’m optimistic that VMware can make that also as simple as possible, but the truth is monitoring microservices is somewhat different and the ability to customize your monitoring solution will be part of the answer for some time to come.
vRealize Operations allows for customization of the dashboarding, alerting and reporting functions to easily and effectively monitor containerized services in your enterprise. My approach here is less of a “how to” than a “why”—why is monitoring microservices different than other application types, and how does this affect system administrators?
Containers are changing the role of the system administrator, and that role is changing because how we define user communities is changing. It’s no longer as simple as being the SysAdmin for SQL Server or Microsoft Exchange. Microservices are among the cloud-era trends that redefine the composition of users along the lines of who uses a particular service—and it’s not just one app in relative perpetuity, it’s a service that crosses organizational borders and often for a short time (most microservices live for just days). Many of these services are exposed to customers, now bringing the system admin into the customer experience.
Microservices, with their short life spans and distributed nature, require more than just standard alerting; they require asking different questions. By creating custom dashboards, alerts and reports that consider the behavior of your microservices, you can derive a meaningful picture of application health; it’s a more holistic approach, one that challenges us all to think beyond individual components and even performance metrics. What is the health of the underlying application?
vRealize Operations offers an operationally efficient way to effectively monitor microservices and can easily be customized to derive meaningful data from containerized instances.
vROps lets you manage applications, infrastructure, hypervisors and yes, containers, all from one platform. vSphere Integrated Containers lets you easily manage VMs and containers leveraging vCenter, so you don’t have to learn new tools. Clearly as the enterprise becomes more complex, we need to ensure that management domains don’t also grow in number or complexity. VMware keeps it simple by viewing containers as another activity abstracted from physical infrastructure, yet allowing you the tools to gain the insight you or your business users may need.
VIC integrates natively with Docker to manage the monitoring and alerting of Docker containers. With microservices you can be less concerned with the container itself and more with the health of what it depends on; the application is broken into many different pieces so you are less concerned with infrastructure and more with app-level metrics (which are offered in vCOPs).
vCenter organizes all your containers in a folder; vROps can then monitor the affected resources as a pool or you can customize a dashboard that just dials in on what you need to know. With a custom dashboard a particular system administrator can say: “I only want information on cloud-native-apps#4 and #5.”
Best practice: make sure your alerts identify issues before they impact the business, not after. Microservices behave differently than traditional, monolithic applications; vROps uses machine learning to quickly understand the application’s behavior, but it may know more than it can tell you if you haven’t set up your alerting system with that context in mind. The key takeaway here is that before you configure alerts, define what triggers an alert. The mindset of monitoring symptoms calls for re-examination in a container environment; microservices will likely present different symptoms because they use and impact resources differently.
vROps alerts can be built and customized without programming or scripting languages, making it easy for even the newest on the team and driving opex savings in the bargain. Linking alerts with recommended actions and workflows drives further efficiency through automation.
vROps employs widgets for easy custom dashboarding. An intuitive GUI allows you to draw what you want to see, then paint in the details, such as “Cloud-Native App #4.” This can be helpful for the granularity needed when monitoring a service that may only live for days. The “mashup” chart is an important option for those managing containers: it uses a special widget that brings together disparate pieces of information for a resource, showing a health chart, an anomaly count graph and metric graphs for key performance indicators. Widgets can also help with the sharing of information from one dashboard to another, which helps with troubleshooting.
This rather mundane reality—the need to easily see the activity in your domain—may enjoy a more important role in the containers era. As mentioned, because user communities are changing the role of the system administrator, we may see the need to report to business owners and executives shift as well. Custom reports can be generated quickly, shared (yes, even printed out) and consumed easily well beyond the borders of IT. Again, the “why” of custom reporting when it comes to containers may be as simple as; new business owners are asking new questions for different reasons (for more on “how” see the customization guide link at the bottom of this post).
As cloud matures and microservices become just how we do business, the role of the system administrator may once again settle down to a job description that holds for the eternity of maybe a year. And today’s custom dashboard may be a regular feature in a future release. But I’m not holding my breath. We are in a transition that will last a while, and customizing the monitoring environment for this dynamic ecosystem is imperative to effective operations for containers. Using a single management console robust enough to handle it this complexity makes your job a little simpler.