How to Monitor Software-Defined Everything
The days of racking and stacking specialized network, compute, and storage to launch a new service or application are coming to an end. The market has embraced the concept of Software-Defined Everything (SDx) as a vehicle for greater business agility and innovation.
In fact, SDxCentral predicts the SDN market alone may grow as high as $35 billion by 2018. SDN-related spend may account for up to 40% of total networking in that time frame.
It’s clear that we need to adopt a model where trouble tickets are no longer part of the provisioning process. A recent article on the OpenStack web site illustrated the problem by pointing to one large enterprise that required 117 trouble tickets in order to provision a virtual machine. The net impact: it took six weeks for a VM to be delivered to the requester. That turnaround is unacceptable by today’s standards.
Virtualization of the entire application infrastructure stack enables elastic computing, driven by real-time application needs. It eliminates much of the manual process that slows business today. Self-service provisioning makes the IT organization more responsive to the business. It addresses the needs of end users who want it fast, want it now, and want it mobile.
Rapid, on-demand infrastructure requires a dynamic performance monitoring platform that provides insight at scale. The performance monitoring platform must:
- Understand the relationship between physical and virtual infrastructure
- Integrate with provisioning systems and controllers to share real-time device provisioning knowledge (systems can no longer rely solely on polling)
- Scale on demand with the virtualized resources; it must react to – and become part of – the system to be provisioned
- Provide service-context awareness; it’s not just about devices, it’s about the user and the applications that need to leverage specific sets of network services to meet their unique objectives
- Allow users to correlate all data across a slice of their infrastructure at any point in time for troubleshooting and capacity planning
What might performance monitoring look like in a software-defined everything infrastructure? Watch this whiteboard video to learn more.