One key challenge facing IT in the evolving hybrid cloud arena is seeing what the heck is going on! Fortunately, IT service delivery folks have some options to create new observation posts in the rapidly expanding frontiers of their hybrid infrastructures. Here are some things to consider.
IT needs to embrace available tools to extend visibility into every part of the hybrid service delivery environment.
Virtual switches can provide a wealth of information about what those switches are doing, especially if they are deployed in a software-defined network (SDN) built around OpenFlow. They feed flow information out into network monitoring tools and keep systems on the outside apprised of what is going on among the virtual servers inside the host systems, even when the traffic among them never exits the host. Just about any virtualized server infrastructure -- VMware, Cisco or instances of Open vSwitch -- is going to have virtual switches in place. Moreover, virtual switches are used in external cloud environments as well; therefore, these can be your first vantage points to observe network and application performance in those situations.
Virtual appliances beyond the switch can do even more for you -- again, either internally or in public or private cloud environments. A variety of network appliances exist these days, ranging from load balancers and application delivery controllers to application acceleration, encryption or cloud bridge appliances. The migration of vital network functions to virtual appliances heralds the more general shift to network functions virtualization, which advocates that all core network functions be ported to commodity hardware and virtual systems.
Transaction monitoring systems focus squarely on the problem of application performance end to end. They map out every major step in a transaction's life as well as how much time is needed to complete each step. They also track over time how those spans for current transactions and users deviate from the norm. Yet these systems need a proper vantage point.
To obtain the full picture, they need to be placed between users and back-end systems (in-house or external) and between the various components in a service-oriented architecture. Otherwise, they need ways to get data from these vantage points if they are not directly placed there themselves. (This can lead to the creation of overlay monitoring networks, tracking physical and virtual components, and has provided an early SDN application example: Big Switch's Big Tap.) Transaction monitoring systems include BlueStripe's FactFinder, ManageEngine's Application Manager, SolarWinds' Web Application Monitor and associated components, and cloud-based platforms such as Boundary and Monitis.
Cloud management stacks such as OpenStack or CloudStack can serve as another clearinghouse for information from systems and components in the virtualized data center, tracking not just servers and networks but storage as well. In traditional tiered-application architectures (such as application and database or database and storage) these systems can make it easier to spot performance problems between tiers. Alternatively, they can simply spot system load and behavior anomalies in any architecture.
The current and coming IT environment is dramatically different than it was just 10 years ago, so IT's approach to monitoring performance has to expand to encompass the virtual environment and the intermingling of internal and external services. The good news? Tools and techniques to satisfy those demands are available today. IT needs to embrace them to extend visibility into every part of the hybrid service delivery environment.