Cisco M6 and Intersight

Diclaimer: You can take this with a grain of salt as I still have yet to see everything released and try it out myself. All of the comments below are all my own opinion.


Recently I sat in a briefing to get a glimpse of what the new UCS M6 series looks like. I was expecting the usual debriefing, but to my surprise there were a number of significant changes that were coming down the pipe. Intersight was the central focus of the conversation with a seemingly all knowing platform that could control your infrastructure, virtual, cloud and storage environments in one place. It almost felt like a rehash of UCS Director for anyone that remembers that, but in a more mandatory view with UCS. The solid stable UCS model that I had come to know and count on over the past 12 years is now taking a big jump and into unchartered territory.

The following are some of the most notable items that came to mind with this new venture into new territory.
1. Centralized management in a SAAS  model called Cisco Insight. The Local UCS manager is eventually going away?

2. New Chassis design. These new Chassis apparently will be part of a new distributed model opposed to the converged model UCS has long held. I will explain later. The idea is that the Fabric Interconnects would be phasing out eventually all connections would go directly to the Chassis.

Here are my concerns:

1. Cisco insight. A lot of features (monitoring, firmware mgmt, and automation) are in store with this platform, but the idea that something outside of the datacenter is controlling and managing the bedrock of my infrastructure concerns me. To be somewhat fair they mentioned a model with an on-premise footprint, but having all of that control in a cisco managed cloud definitely had my mind racing.

There are so many scenarios that run through my mind.

a. What happens if there is an internet outage. Your doing firmware patching and ________ happens. Understanding how this new framework recovers is imperative. Blips to the outside internet are sometimes out of our control. There better be a damn good plan in dealing with failed states and way to ensure things will be functional even without a cloud/external connection.

b. With greater power and control in one place, comes greater responsibility. Moving all management for UCS globally into a single environment sounds like a bad idea from a security perspective. I don’t even want to think of the consequences if something happens there.

c. Customers will be at the mercy of an outside provider with Cisco running the Insight management.. I’ve been in situations before with Azure and other providers that have outages or a bad bug. Not having the control over your own environment sucks when things go wrong. I hope that is never the case, but as we have seen in other cloud provider disasters it’s possible.


2. Going to a model removing the Fabric interconnects and becoming more distributed. I am curious about the justifications here. UCS was popular because it converged and simplified your infrastructure and networking and I feel this is going the opposite way. What was wrong with the FI structure before. Why is it necessary to go a distributed model.

a. What happens to FCoE or Fiber Channel? The FI takes care of all of the complexities for you in one place, but now it will span out to each Chassis?

b. Going to be more costly from a connection standpoint


Again, I still have more to find out about the M6 line and the future management here so I don’t want to knock it before I give it a chance, but these are definitely the first gut questions that come to mind that I want to either quell with more information or deflect this new line as long as possible. You can be sure that I will be taking this for a spin in my lab. I will try to post any updates as I begin to understand this a bit more.