kitcolbert

Heterogeneous Cloud Management at VMworld Europe

October 9, 2012

In a recent blog post, Steve Herrod outlined how VMware products are embracing support for heterogeneous environments. VMware’s Enterprise Management products are no exception. Heterogeneity is important because while many of our customers have chosen a VMware-first policy, the reality is that their environments may not be completely vSphere: there are pockets of physical servers that have yet to be virtualized, developers run around IT’s back and go straight to Amazon Web Services (AWS) to test their latest changes, or the company makes an acquisition whose datacenters don’t run vSphere. So any management solution must be able to support heterogeneous environments. This is exactly what we’ve designed our Enterprise Management products to do.

In his keynote at VMworld Europe today, I joined Steve onstage to announce updated Enterprise Management products that extend the vCloud Suite. I’d like to take the opportunity to walk through some of the new management products and talk about how they enable heterogeneity.

Heterogeneity can take on many forms, but I will focus on two of them for this post: private vs public cloud and type of virtualization technology (or lack thereof, i.e. physical systems). In other words, VMware’s Enterprise Management tools not only must be best-of-breed for vSphere-based private cloud environments, but must also be able to manage public clouds as well as non-vSphere-based private clouds and physical systems. (Of course, our products support many other types of heterogeneity too, but here I’m just discussing the two I mentioned.)

The first step with any sort of cloud (public, private, vSphere-based or not), is to build it (duh!). Today Steve demoed how this can efficiently be done using vCloud. While vCloud Director focuses primarily on vSphere environments, with the addition of vCloud Connector you can include VMs running in vCloud Director-enabled public clouds. This is an example of how you can seamlessly manage VMs in both public and private clouds.

Once the cloud is up and running, you now want to set policies on the infrastructure and deploy workloads into it. VMware has two products to help with that: vCloud Automation Center (formerly DynamicOps, vCAC for short) and vFabric Application Director (or AppD).

vCAC is aimed at infrastructure admins and allows them to set up policies that govern infrastructure usage (e.g. creating new virtual datacenters or how workloads are provisioned, managed, and then retired). The DynamicOps technology began within Credit Suisse and was developed by a team that was used to managing very heterogeneous environments and thus recognized the importance of building an extensible system from the ground up. The infrastructure policies are specified in “infrastructure blueprints” that can be applied to any type of system (virtual, physical, etc). Infrastructure blueprints specify the usage of those systems: you can control the approval process (or not have one at all, if you trust your users ?), choose the deployment mechanism, automatically clean up after a certain time period, and much more. While vCAC has out-of-the-box support for many different environments, you can also extend the functionality yourself. The vCloud Automation Center Development Kit enables you to quickly model new systems and create modules that enable provisioning to and management of those systems. The most powerful aspect of vCAC is that it can plug into your existing deployment and management technologies, meaning you don’t need to change tools or processes to use it.

AppD focuses on application admins. It allows them to compose an “application blueprint” by combining an OS with their choice of middleware plus the actual application bits. (Yes, AppD also uses the term blueprint. It seems to be very popular!) The application blueprint concept is powerful because it is designed to work for any app, on any cloud, for the full lifecycle of the app. “Full lifecycle” means not just initial provision, but also updates as new versions or security patches are released. Today the application blueprint can be provisioned to Windows- or Linux-based VMs on either vSphere-based private and public clouds or AWS. This is accomplished through “deployment profiles,” which allow you to configure many aspects of how the application is deployed, including not only which cloud service to deploy to but also properties such as network settings, memory allocation, or permissions. You can even configure the deployment to automatically run unit tests or check for the latest security patches if you so desire. Support for AWS is new and is critical as we see more and more companies using AWS for test and dev. AppD allows you to have more control over your applications running in AWS by conveniently tracking your deployed applications and giving you the freedom to pull them off AWS at any time. To read more about AppD and VMware’s application management strategy, check out my colleague Preeti Somal’s blog.

The final piece of the puzzle is to manage the on-going operations of the cloud. This is where VMware vCenter Operations Suite (aka vC Ops) comes in. The vC Ops Suite contains a number of products, but the two important ones I’d like to focus on for this discussion are vCenter Operations Manager, which manages performance and capacity, and vCenter Configuration Manager (vCM), which manages configuration and compliance. Both have been built from the ground up to support heterogeneous environments. vC Ops Mgr has what we call a “data agnostic” server and analytics component that is built to be able to take in and analyze data from any data source. It uses “adapters” to connect to different data sources (e.g. vSphere, other hypervisors, public clouds, storage arrays, network switches, other management systems, etc) and pull in data. So in order to add support for another system, all you need to do is write an adapter. The data from all those adapters flows into the vC Ops server are automatically analyzed for performance problems and then can be visualized in the UI together.

vCM is very similar. It has “inspectors” that connect to remote systems and pull configuration information and track data change. In addition, there are parsers and XML definitions that allow you to specify how to model the configuration data that comes in from the remote systems. vCM takes all this data and automatically tracks changes and assesses compliance. You can easily view the results in summary screens that allow you to see change and compliance across all systems in your environment, regardless of what type of systems they are.

VMware is focused on ensuring that its Enterprise Management products support heterogeneous environments. We recognize the world has become more heterogeneous, and we’re talking about it at VMworld Europe this week. First, Ramin Sayar is presenting a spotlight session on multi-cloud management. There’s also a talk on Managing Operations in a Hybrid Cloud. In addition, we have a bunch of management demos in the VMware booth within the Solutions Exchange. So if you’re here in Barcelona, you should definitely check them out!

As mentioned toward the beginning of this post, our products support many other types of heterogeneity than the two mentioned here, such as heterogeneity of tools, processes, and best practices. The goal of all these efforts around heterogeneity is to achieve a “single pane of glass” and all the operational efficiency that comes along with it. As I’ve described above, the integration is at the data layer, which means that as a user, you can manage a workload in a vSphere VM in the same way that you manage a workload in an AWS VM or on a physical system. The products take care of shuttling the requests to the correct endpoint as appropriate. Moreover, there is consistency in how you manage: you can apply the same policy to a vSphere VM and an AWS VM, view compliance across all your workloads in both private and public clouds, or have a single screen showing the performance of your entire environment.

And this is really what we’re after: to improve how you do management. I started my previous blog post with this idea and I want to finish with it here. At VMware, we want to turn Enterprise Management sideways: to move away from the vertical silos of yesterday to the horizontal layers of the future. We’ve seen all the problems of the vertical silo approach, and virtualization and the cloud offer us an opportunity to change the way we manage. This is really powerful stuff and we’re excited about where it’s going!

Tags: , ,

kitcolbert

Kit Colbert

CTO, End-User Computing

Kit Colbert is the CTO, End-User Computing at VMware, driving technical strategy and advanced development of innovative solutions for VMware's desktop, social, and mobility products. A 10-year veteran at VMware, Kit previously held roles as the Chief Architect and Principal Engineer for Horizon Workspace and as the lead Management Architect for the vCenter Operations Suite ... More

Leave a Reply