Enterprise architecture has battled complexity for years using a divide-and-conquer approach where individual domains have been isolated so that complexity can be more effectively managed. But the promise of cloud is that users and workloads can exist anywhere across a pool of fungible resources, which fundamentally breaks down this containment strategy.
For security and operational workflows to extend end-to-end from across all places in the network (data center, campus, branch and public cloud), monitoring and visibility tools must also reach end-to-end.
For anyone who has run major infrastructure, chances are they have experienced the paralyzing feeling of flying blind. When infrastructure extends beyond even moderate amounts of sophistication and complexity, it becomes nearly impossible to understand what is going on using common tools and processes.
For example, when something is broken, operators want to move to a dashboard. The issue isn’t that there is not a dashboard; the issue is that there are too many. In that moment, I would find myself looking at a dozen dashboards split across different tabs. The task of correlating issues became a manual one where I had to effectively do event correlation visually.
In a multicloud world, this dynamic gets worse. There will be dashboards for individual islands of resources—underlay and overlay, physical and virtual, private and public.
Blindness doesn’t just happen when there is not enough light—you are just as blind when there is too much to see.
If one primary objective of multicloud is simplification, then monitoring and visibility have to be made easier. There cannot be too many diverse ways of getting access to the information that resides in and around the network.
There has to be a central place where data can be accessed in a way that supports the various functions that must occur when managing infrastructure. You need basic operations analytics, state-driven orchestration for dynamic workload management, intent-based monitoring and alarms for helping flag issues to the appropriate people and data-driven capacity planning to understand where resources need to be turned up or spun down.
Of course, each of these tasks provides a different context for which management needs to happen. Depending on the activity, the data or response to that data might change. In fact, in many cases, the problem is not having enough data—it’s gleaning meaningful insight from that data.
In a highly-automated multicloud architecture, operators need to collect data, but that is really in support of gathering information. If you cannot take data and turn it into actionable information, the data is useless because you may drown.
This means that one of the most important requirements for end-to-end visibility is transforming raw data coming from a diverse set of resources into a format that is immediately usable by the operator. For some tasks, this means rationalizing events for easier visualization, and for others, it means packaging data for processing so that a system can automatically execute some remediation action.
And this all has to happen across the data center, campus, branch and public cloud.
Data Leads to Automation
Of course, if the objective is agility, then insight needs to lead to action in a way that does not always require a human in the middle to act as the intermediary.
Visibility has to be developed on a platform capable of analyzing diverse data and then translating insight into automated action. This can include things like automated event correlation driving intent-based remediation on the back end, or more orchestration-related activities like resource utilization being used to make dynamic workload decisions in private and public cloud.
But for any of this to be possible, the underlying platform has to support technology like machine learning and artificial intelligence to help derive information from data, and ultimately convert that information into concrete actions within the infrastructure.
Having played key roles on both the user and developer side of major infrastructure, I am particularly excited about Juniper Network® AppFormix® platform. It strikes a unique balance between improving visibility across multi-domain infrastructure and providing a development platform for more sophisticated operations.
Having watched as customers experience AppFormix for the first time, it seems clear that they come for the insight. And knowing what we are doing to build action into insight, I am certain they will stay for the automation and DevOps capabilities.