For at least the last year, VMware has made it very clear that it views virtualization as a catalyst that will enable the delivery of a new management stack into the enterprise, and that VMware intends to be the vendor of that management stack. This is obviously in competition with the vast majority of the software vendors who are currently in the VMware ecosystem as well as in competition with existing systems management vendors.
Prior to the acquisition of some of the Ionix assets from VMware, the stack was very VMware vSphere specific and had a lot of holes. In fact the stack when layered on top of the existing VMware virtualization platform looked pretty much like the diagram below.
|EMC/Ionix Name||Original Source||Basic Functionality||Supported Platforms|
|Application Discovery Manager||nLayers (acquired 7/2006)||Application topology discovery via an appliance on a mirror port, with application fingerprinting||Everything accessible via TCP/IP, WMI, SSH, and SNMP|
|Application Stack Manager||FastScale (acquired 8/2009)||Just Enough OS (JEOS) technology that strips unneeded components from operating systems and applications||Windows Server and Linux Server based applications|
|Server Configuration Manager||ConfigureSoft (acquired 5/2009)||Template based configuration management system that tracks changes and ensures configuration compliance. Includes a CMDB.||VMware, HP-US, Solaris, Apple, Windows, AIX, Linux, Red Hat, Debian, Microsoft Active Directory, Microsoft BackOffice|
|Service Manager||Infra Enterprise (acquired 3/2008)||Web based incident, problem, change, configuration, and service management with a service catalog||Web based application with a back end relational database|
What Does This Mean to VMware?
Using the Virtualization Management model presented at the start of this article as a guide, this set of acquisitions bolsters VMware’s management stack in the following key areas:
- Application Discovery is an essential part of Service and Capacity Management. In a dynamic environment it is too little to late to batch load configurations into a CMDB on a periodic basis. In order for Infrastructure Performance Management and Applications Performance Management in a virtualized or cloud based infrastructure to be possible, continuous knowledge of the applications topology must be combined with and understanding of where the slowdown in performance is occurring. Of course Application Discovery Manager does not today integrate with AppSpeed or any of the other third party virtualization aware APM solutions. We will have to wait and see whether VMware is able to integrate ADM with AppSpeed in a timely manner, and more importantly make the ADM data available to APM vendors who address applications scenarios not covered by AppSpeed.
- Configuration Management is an essential part of ensuring the integrity of the virtualized system. A high percentage of the security, performance and availability issues that arise in an enterprise are related to configuration mistakes and configuration drift. Prior to this acquisition VMware had pre-announced a product, ConfigControl that was apparently going to be VMware vSphere focused, and based upon internal development. Now VMware owns a cross-platform configuration management and assurance solution that was formerly the flagship product of ConfigureSoft. This significantly bolsters the Configuration Management assets of VMware.
- IT as a service, built around internal and external clouds has been a vision of VMware for some time. With the acquisition of Service Manager, VMware now has the product assets to start to deliver upon this vision. Service Manager includes a self-service portal and the ability for enterprises to create a service catalog. This is therefore a huge lift to the provisioning and lifecyle management area of the VMware product line which was previously occupied by Lab Manager and LifeCyle Manager, offerings that were significant subsets of what VMware has just acquired.
- Finally, Application Stack Manager is a way to significantly shrink the sizes of the operating systems and middleware components in a guest, and therefore extend VMware’s current lead in memory efficiency and guest density. The benefits here are so fundamental that it would not be surprising to see VMware kill this as an independent product and to build its functionality into the core parts of vCenter Server that provision guests.
While this set of functionality does raise VMware’s game in the management software realm to a significant degree, it also raises a set of enormous challenge for VMware:
- VMware today sells a virtualization platform to the teams that provision and support operating system, server, network, and storage infrastructure in the enterprise. These management solutions are most appropriately sold to the same people that buy management solutions from IBM, HP, CA, BMC and the other hundreds of management software vendors. VMware has an enormous marketing and sales hill to climb in order to become effective at this layer of the organization.
- VMware has to decide if it is going to continue to develop, market and support these products in their current broad and hetorogeneous form. When these products belonged to Ionix, it was the intent of Ionix to support not just the physical and VMware environments, but also competing hypervisors like Microsoft Hyper-V.
- These products are not integrated with each other, nor are they particularly well integrated with the rest of VMware’s management stack. VMware desperately needs a management stack architecture with a common database schema for this product family, that also encompasses the rest of the VMware virtualization management offerings (like AppSpeed, CapacityIQ, Lifecycle Manager, and Lab Manager).
- VMware could use these new products to declare war upon its ecosystem of third party software vendors, or it could open them up and let the ecosystem add value to them to make them address constituencies and use cases that the base products do not. VMware has done as masterful job of this with its virtualization platform, doing so with a management stack is both more challenging and ultimately more rewarding.
- For more detail on this point, see the next section below, but by declaring itself as a management stack vendor, VMware has now positioned itself in a competitive manner with several of the largest software companies on the planet (CA, IBM/Tivoli, HP Services and Software, and BMC).
What Does This Mean to The Industry?
Prior to this set of acquisitions, management software vendors like CA, IBM, HP, BMC and hundreds of others could say to themselves and their customers, “VMware is building vSphere specific management solutions that have neither the depth nor the breadth of our offerings. We feel comfortable treating VMware as just another silo that our deep and enterprise aware offerings will manage”. This strategy on the part of the management software vendors has now been proven to have been an illusion. It is now clear that VMware has the intent and the product assets to use virtualization as a catalyst to inject itself as the vendor of a new management stack into the enterprise. Every existing vendor in the systems management, lifecyle management, service catalog, and performance management aspect of the management software business should now expect VMware to be a competitor by VMworld 2011. The only area that has been left relatively unscathed by these actions is security. The effect of this acquisition on the existing VMware virtualization management ecosystem will be covered in a subsequent post.
What Should Enterprise Customers Do?
The first thing that enterprise customers should do is recognize that VMware is probably right about one thing. The management of a dynamic IT infrstructure consiting of private and public clouds should not be done with legacy management software built around silos of hardware, operating systems and applications as has been done for years. This approach is simply too inflexible, brittle, and costly. Therefore enterprise customers should start to build a requirements list for how these next generation systems should be managed around the following concepts:
- Every new management system of any type that comes in the door from now on needs to be private and public cloud aware. It needs to be built for dynamic internal systems, and built for the commingling of applications across private and public clouds. Data collection approaches need to be “cloud friendly” with collection systems easily installed in public clouds, and architected to be able to communicate over public protocols like HTTPS back to a web service in the management system in the enterprise.
- The new management system should be built upon a CMDB that is kept up to date in the most efficient manner possible with continuous and dynamic discovery of the physical infrastructure, the virtual infrastructure, and application system topologies. Polling of some of the physical infrastructure is unavoidable, as much if it is not intelligent enough to push configuration changes to a management system. However, this approach for physical systems should be combined with virtualization aware approaches like integration with VMware configuration events, and integration with the VMware vmSafe API’s for the virtualization case.
- Support for multiple hypervisors needs to be firmly committed on the vendor’s roadmap for delivery by the end of this year. Microsoft Hyper-V is making gaining sufficient traction in enterprises so that it is highly likely to be a force to be reckoned with down the road. The whole reason that VMware is pursuing this management stack strategy is because VMware itself believes that the hypervisor will get commoditized at some point in time and that VMware is focusing upon the management stack is the next source of value to its customers.
- The management stack should be open so that third parties can easily use the data in it and add value to it. Many enterprises have had less than positive experiences with monolithic enterprise management systems that locked customers into release cycles and maintenance cycles from one vendor. No one vendor can keep up with the breadth and pace of innovation in this industry, and a management system needs to support independent asynchronous enhancement if it is going to have any chance of keeping up with this pace of innovation.
- The system should be role and policy based. Any new management system must be able to separate users into roles and govern their actions by policies assigned to roles. Many of the existing management systems for virtualization (including vCenter Server) are targeted at one role (the VMware admin) and they give this person the flexibility to do anything. An enterprise management system that encompasses the depth and breadth of functionality described in this article must deal with multiple constituencies and provide the appropriate role based controls for these constituencies.