“No responsibility without visibility!” may not be the rallying cry that “No taxation without representation!” was, but it still deserves its own revolution.
Delivering on digital transformation with consistently usable applications in collaboration with a cloud provider is frustrating if you cannot clearly see what is going on. Identifying the root cause of user experience problems is difficult and time consuming. Getting the problem fixed can result in excessive back-and-forth on who has primary responsibility. The revolution calls for visibility from the user’s perspective, starting with their device and extending through the network to the application backend systems, using cloud monitoring and performance management tools.
Here are the top five challenges of monitoring public or hybrid cloud solutions.
1) You cannot determine whose fault it is
Cloud providers often talk about shared responsibility, but that should not mean that each party is only aware of their own portion. Many monitoring tools are IT-centric and measure the health of infrastructure components such as networks, servers, CPU utilization, memory, and storage. Lack of visibility into what is going on within the cloud leads to finger pointing, delayed problem resolutions, and unhappy users.
For example, we were having intermittent issues with a Microsoft SharePoint site, with some users reporting very slow browsing times. Basic cloud monitoring reported that all systems were up and running. The initial support contact with our multi-tenant hosting provider led to a troubleshooting session and a proclaimed resolution. However, the next day we experienced similar issues.
2) You cannot see what the user is experiencing
Delivering usable applications and customer satisfaction is all about what the user experiences. It does not matter if the cloud provider says that everything is up and running if the users are experiencing slow responses and are not happy. Since users only care about their own application experience, it makes sense to build cloud monitoring and SLAs around their viewpoint. Unfortunately, many cloud SLAs stop at the cloud’s edge, provide few details about what is going on inside, and do not measure the health of the user’s device, leaving IT with significant gaps and blind spots.
Using advanced cloud monitoring tools, we analyzed the issues from the end-user perspective. We quickly identified that some users were having issues from multiple locations (left-hand graph of Figure 1), and that the poor performance was not directly related to our user volume (right-hand graph).
Figure 1 – Cloud monitoring details by location and by user volume
3) You do not have consistent views across a hybrid environment
With so many components involved in cloud-based applications, partial visibility of the hybrid network and its many moving parts is unacceptable. Network latency changes frequently across different paths, virtual machines spin up and down, containers pop into existence for a brief time, and workloads are continually on the move. In addition, users have a wide variety of devices, many of them personal and outside the direct control of IT. Performance and experience issues could arise from any of these components, and only with high-resolution cloud monitoring and end-to-end transaction data can problems be quickly identified, diagnosed, and resolved.
In our example, we were able to clearly document that the performance slowdown was being caused in the backend and was creating in a 2-3x increase in average user wait times, well beyond our SLA. The additional visibility helped the provider quickly resolve the actual problem, one which was affecting multiple tenants on their systems.
4) Your tools do not talk to each other
Legacy monitoring systems operate in isolation, tracking only their own small part of the overall system. Modern cloud monitoring tools follow each and every transaction from the user and their device and map it all the way through each intermediate network and module to the backend systems. Equipped with this detailed information, identifying the root causes of user complaints, triaging which ones are the highest priority, and quickly resolving the issues becomes the everyday practice in IT.
5) You cannot measure the business impact of a problem
Binary up or down of services is not enough to build a solid and sustainable SLA with your cloud provider. With so much riding on your digital services, SLAs need to be closely linked to your business processes and objectives to ensure that the right parties are held accountable. Advanced cloud monitoring systems enable you to build SLAs with your providers, and your internal departments, that are directly linked to your business. Monitor activity and get automatic notification of unacceptable conditions by user, department, device type, or geography. Analyze the performance of each application, not just the infrastructure it runs on. Compare performance before and after upgrades, enhancements, or feature roll outs to see if they are delivering the expected improvements.
The cloud monitoring revolution is here
Without having to overthrow anyone, modern cloud monitoring is enabling a tremendous revolution in the viability of shared IT/cloud operations. With resources freed by these productivity improvements, IT can do more to proactively help the business. Cloud monitoring at this level of detail also enables you to track usage patterns, identify the most popular features and applications, and more efficiently allocate resources for upgrades and enhancements. Vive la revolution!
About the Author
Gayle Levin is director of solutions marketing at Riverbed Technologies. Previously, she held product marketing and campaign roles at VMware, Oracle, and Splunk as well as several startups. Her interests lie in the impact of technology on the way we think and work today
