In my first blog about my experiences at both VMWorlds 2014, I talked about the differences in focus between the US and Europe when deploying VMware in the data center. In this blog, I want to talk about how virtualization changes data center design for tier 1, business critical applications.
For those who took the time to go through the scenarios in VAPP2947, we demonstrated cloud bursting, planned migration and disaster recovery (DR) in the cloud. The underlying infrastructure consists of a combination of traditional data storage, VSAN storage and NSX. You probably realize the current features in VMware products require a very structured process involving several manual configuration steps. So I was excited to read Duncan Epping’s blog about what is coming for vSphere and VSAN, and I particularly focused on the part about vMotion support across LAN and WAN distances.
Why is that important to me?
While today the practical use cases for vMotion across data centers is very limited, the use of vMotion to improve workload agility can eliminate the current error prone manual steps when moving a VM between hosts managed in different vCenters. A common promise of server virtualization as part of the Software Defined Datacenter (SDDC) is speed and agility of deployment and operation of the data center infrastructure.
While we frequently mention speed and agility in the same sentence, they mean very different things.
Speed is the time to provision and deploy while agility is the ability to scale and apply different policies to the infrastructure supporting a business application throughout its lifecycle. As much as we enjoy designing sophisticated infrastructure solutions, our jobs exist because of the business value of applications and its data. Achieving speed has been the low hanging fruit from server virtualization. Agility on the other hand is bit higher up the tree requiring more of a reach. It requires more integrated visibility and collaboration between the compute, storage and network silos and ideally, it is rooted at the application stack.
Within the last couple of years server virtualization has matured nicely. Now, we see more and more business critical applications (tier 1) being hosted on a hypervisor coexisting with the vast amount of tier 2 and tier 3 applications that were the low hanging fruit. But, tier 1 application virtualization drives new requirements for stability, performance and agility in order to guarantee adequate resources for these critical applications. And that drives the need for visibility across the virtual abstraction layers and a way to correlate the service abstractions with the physical infrastructure supporting them.
An example of how to deliver visibility across the physical network is Brocade’s integration with vRealize (previously VCOPS) for Fibre Channel (SAN) and VCS Fabric for IP networks. This integration with vRealize enables operators to pinpoint directly in the physical network which resources are being used by what VMs. For more details about this, take a look at the breakout session MGT2948-SPO - Effective Options For Reducing The Time To Perform Root Cause Analysis Of VMware FC SAN Environments I had the pleasure to co-present at VMworld Barcelona with Kenn Ross (Director of Product Management, Brocade). In this session you get a good overview of the integration with vRealize and examples of how identification of for example a slow draining device in the SAN is correlated to the dependent Storage Array ports, LUNs and VMs placed on these LUNs. The example also show how Brocade and EMC has worked together to deliver unified dashboard views in vRealize across SAN and storage Arrays for easier correlation and faster root causing of issues.
In my next blog I will talk more about how I see Enterprises transitioning towards the mega trend known as the “Third Platform” and its impact on data center design.