UCS and UCS Express
I’ve been watching with some interest how reporting on UCS Express has played out in the wake of our latest Borderless Networks initiative. The original Integrated Services Router launch was the first I participated in when I joined Cisco back in 2004, so it’s fun to have things come full circle.
Because I think history is a great way to get a perspective on the present, here’s a quick look at how we introduced the ISR in 2004:
Cisco had already seen an opportunity to help reduce both physical IT clutter of multiple appliances as well as the support and management burden for small and branch offices, where on-site IT staff is often limited. We were able to offer wire-speed embedded services because of the way we separated service processing from traffic flow. And because all of these services were network-based at their core, we were able to roll much of the service management into the device manager, thereby eliminating several points of management—and a fair amount of troubleshooting.
This overall design philosophy of network-centric infrastructure convergence and management simplification should sound remarkably familiar if you know a bit about UCS. In a sense, the ISR with UCS Express and UCS itself have a symbiotic evolutionary path. The ISR’s rapid adoption proved it was operationally feasible to consolidate multiple services within a single box. The introduction of UCS—and subsequently, other forms of “unified computing” from some of our peers—built on this realization, so that the notion of integrating compute and networking no longer seems unrealistic, either. It follows quite naturally that the next wave of branch consolidation would be to put a server inside a router and have it remotely managed as an extension of the data center.
UCS Express is intentionally lightweight for now: it eliminates the need for standalone or dedicated print and file servers, while also providing local caching of DC-based apps for short-term application survivability in the event of link interruption. It can be managed remotely via an onboard management controller, similarly to the UCS C-Series Rack-Mount Servers (in standalone mode).There are some erroneous reports in the media that it’s managed with UCS Manager, which is not true at present—but we do have a long-term vision of increasingly centralizing UCS-E management and scaling it via abstraction.
As remote control increases, WAN speeds accelerate, and processing power continues to drop in cost, it will make increasing sense to host larger apps locally on the ISR as backup copies of those hosted in the data center. This will help make VDI implementations more reliable while maintaining data integrity and centralized control of the infrastructure.
Since we first released the ISR, the IT world has undergone some important shifts—such as the renewed supremacy of the data center as well as a host of new regulatory compliance concerns–with a correspondingly stronger interest in making branch IT ever simpler and lower-touch, with as much centralized, remote management as possible. So Cisco’s come up with a vision for “core” and “edge” computing, with our strength in end-to-end networking making it all possible.
Is this wildly improbable, or do you think it makes some sense?