As we kick off VMworld, there’s a lot of excitement about the software-defined datacenter and all the benefits that it brings. But the benefits of the software-defined datacenter can only be realized if you can manage it. Management here covers a broad range of disciplines, including knowing if there is a performance problem, having insight into your datacenter’s compliance status, understanding your costs, and enabling self-service provisioning for users while maintaining control. Management is critical to operating the software-defined datacenter, yet management of the software-defined datacenter is much different than in traditional systems.
Moving From Traditional Management to Software-Defined Datacenter Management
Traditional systems consist of dedicated infrastructure tied to specific applications, leading to tightly integrated vertical silos. Each of these silos has their own management tools and processes (and people!). As you move to the software-defined datacenter, those vertical silos are turned on their sides; infrastructure and applications – as well as end user computing services – are abstracted into loosely coupled horizontal layers. These horizontal layers enable you to run any application on any set of infrastructure services. Moreover, rather than a different set of tools for each silo, you have a single set of tools for each horizontal layer that can manage that layer regardless of the type of hardware, virtualization, cloud infrastructure software, or the application that resides in that layer.
These changes have a number of implications for management:
These changes necessitate a new way of doing management. Traditional IT management tools simply aren’t built for management of the software-defined datacenter.
Enter VMware’s Enterprise Management Tools
VMware has architected its management tools specifically for the software-defined datacenter, with a different set of tools for each of the three horizontal layers.
The first layer is the infrastructure layer. This layer manages all your hardware (servers, storage, network) plus virtualization and cloud infrastructure software in addition to any public clouds you’re using. The focus is on performance of the infrastructure, capacity analysis of the infrastructure hardware and of the VMs running on that hardware, and configuration and compliance of the hardware and software. The vCenter Operations Management Suite provides clear visibility into your infrastructure layer. This not only includes tight integration with vCenter Server, but also includes your physical servers, other hypervisors or cloud layers, and hybrid clouds. You can view all of these different parts of your environment through one common dashboard. Given the growing size of physical and virtual infrastructures, the amount of data is becoming massive. To handle the overload of data, we’ve created automated analytics that sort through the millions of metrics generated every few minutes to detect problems and generate alerts, help you troubleshoot by pinpointing the root cause of a problem, and to suggest how to fix the problem.
With the infrastructure humming along, we turn our sights to the application layer. This layer deals with everything “inside the VM”: the operating system, middleware, and, most importantly, the actual application. There are two questions for this layer. How do you provision these different components and how to monitor their performance and availability? Today the provisioning of various components inside the VM is handled in fairly adhoc or haphazard ways. There are many different approaches to OS image management, application versioning, patching, and so forth. Similarly, there are a variety of tools for monitoring the application, many of which are very specific to that type of application. The vFabric Application Management Suite provides a common mechanism for everything from application build and deployment to operational monitoring and DevOps. Rather than having many different ways to do all of these things for different operating systems, applications, and the like, the vFabric Application Management Suite standardizes on a single way, enabling you to do your work more efficiently and effectively.
The final set of management tools from VMware covers what’s called service governance and IT business management. It’s these tools that really enable you to run your datacenter as a service for your users. The first aspect of running as a service is to enable self-service. This means that users have the ability to provision applications, VMs, and other services themselves without an IT administrator in the loop. While we want to enable self-service (the “service” part of service governance), we also want to give you control (the “governance” part of service governance). This is where DynamicOps fits into the picture. It has advanced policy-based management that allows you to specify up-front rules by which your users must operate. For example, inexpensive applications might be provisioned without approval, but expensive applications might need approval from the user’s manager. These are the sorts of policies you can set. DynamicOps allows you to do this across all your pools of infrastructure (vSphere, non-vSphere, physical, hybrid cloud, etc).
The second aspect of running as a service is to provide IT business management. ITBM is all about understanding your costs, service levels, and vendor operations. This functionality is realized by VMware’s IT Business Management Suite. It complements service governance by understanding unit costs for the resources in your environments (servers, storage, electricity, etc) and by metering the usage of those resources as users self-provision applications and VMs. It can then aggregate those costs and charge them back to the line of business (or whoever you’d like! ?). Similarly, the ITBM suite can analyze service levels to ensure that you’re providing the right quality of service to your users.
This Week at VMworld
Steve Herrod was one of today’s VMworld keynote speakers and I was happy to see that enterprise management played a key role in his talk. Steve talked about the importance of being able to manage heterogeneous environments and how VMware is building support for these environments into our products. I discussed how vC Ops supports physical servers as well as other hypervisors and hybrid cloud, but Steve highlighted new support for heterogeneous environments in vCenter Server and vFabric Application Director.
One of the biggest benefits of the software-defined datacenter is the ability to automate tasks that used to be manual and to realize the operational and business benefits that implies. Steve spoke about how the software-defined datacenter can abstract datacenter services that today may take days to configure into software services that take only moments to configure. He also went over how you can control all this through policies, so that configuration can be fully automatic.
This is such an exciting time in our IT industry, and I hope to see you at VMworld this week!