A while back, I got a chance to speak with Chris Wolf, CTO, Americas at VMware, for episode 11 of the Virtualization EUC Podcast.
We spent the majority of the conversation discussing what software-defined data centers actually are and what the future of infrastructure in the cloud will look like. We also touched on a few other topics, including the vCloud Suite: Horizon, View, and Mobility. As usual, it’s a pretty frenetic conversation, simply because these podcasts are not predefined. While we have a general idea of what we will be speaking about, no questions are delineated ahead of time, so the answers can’t be easily scripted either.
Software-Defined Data Centers (Using vCloud Suite)
SDDCs are primarily about making life easier for businesses and developers (through agility, availability, and lowering costs); they’re about automating and integrating resources. Chris describes VMware’s vision and how it is challenging legacy models that are impeding agility and availability of resources for businesses. He also describes a new security model in which security is assigned by VM or container instead of being tied to a static IP address. His point is about having automation as a feature, not as a custom solution per customer. We talk about historical automation and its customization per environment, and how that cost can be extreme. VMware is trying to curb that and make it accessible simply as a part of its product.
View VDI is garnering increased interest from some sites due to the lowered cost of VSAN. However, VDI is all about being well connected and is not really a solution for every scenario, particularly mobility. Once the experience breaks down, the users stop working. Web applications or mobile applications will be the preferred method for most end users. A native experience will always be the best. This means that whatever works on the device the end customer is leveraging is what should be provided. This, of course, will be a challenge initially because the various devices can be so different.
We spend some time discussing the value of a single set of APIs and partner integration, with the end goal being that any third-party extensions can still be used. Chris makes the point that single-hypervisor, purpose-built management running on an x86 with good automation is more reliable and comes at a lower cost.
We also talk about the added customer costs that complexity in a data center can accrue by the end of the day. “Complexity is great for profits: just not yours,” seems to be the overarching message.
You can follow Chris Wolf at any of the following locations: