By: Clive Longbottom, Head of Research, Quocirca
Published: 2nd September 2009
Copyright Quocirca © 2009
At VMworld in California this week, the topic de jour was, unsurprisingly, cloud computing. VMware wants to be the underpinning to internal and external clouds, using the capabilities of its growing management suite to service the needs of such a dynamic platform.
VMware does have a pretty impressive set of capabilities in this market, with its hypervisor, its ability to manage the provisioning and movement of images around a virtualised environment and so on. And, on the whole, the vendor does seem to understand a lot of the issues that cloud brings to the fore.
But, Quocirca was left with a few worries.
For example, VMware's chief executive, Paul Maritz has a concern himself. Putting an application in the cloud is one thing, but what if the cloud platform provider runs into trouble and you need to pull it out again? To Maritz, the analogy is like a line from the Eagles' song, Hotel California: "You can check in any time you like, but you can never leave".
But is this how the cloud will work? Is it really going to be a platform-as-a-service (PaaS) model, where users will still own the application and provision it to a virtualised hardware platform owned by someone else? Indeed, a number of chief executives from service providers stood up on stage and put forward this model repeatedly. When you look at the names involved—the likes of Savvis and Verizon, it is not surprising. These service providers come from the managed hosting environment—they have already built their models on providing environments on which people run their own software. But is the market ready to change? Quocirca believes so, and we see it as far more likely that the cloud will be essentially a functional platform model, where the user subscribes to the use of a set of functions, rather than to a set of hardware capabilities.
In this case, the application is not owned by the user at all—the cloud provider owns it. All the user is doing is subscribing to a set of functional services. This is the way that the salesforce.com approach works, along with the likes of Concur, Netsuite and others. The software becomes (relatively) immaterial—the way that it "does" things becomes far more key.
If it is a functional platform, then we also run in to issues as to how the "function" is provided. Are we still looking at an "application" per se, or are we now looking at a collection of services from which an aggregate application or process-focused stream of functions are provided, either from a single provider, or drawn from across a whole set of providers? It is far more likely that the cloud will go in this direction, with the likes of Google with Docs, Apps and Wave, Microsoft Azure with its Live services and other functional models "wrapping" functions up in a manner where users can call them at will to facilitate solutions to immediate problems. Sure, this will need some form of underpinning in the means of a process engine, and here players such as Cordys are taking steps to ensure they are first in line. Cordys has managed to place its Process Factory capability into the Google environment, enabling process flows to be visualised and managed, and for functions to be called and aggregated as required from within the Google cloud environment.
For VMware, this presents a bit of an issue. If the user has no control over the "application" itself, the market ceases to be one of "help me as a user to do virtualisation" to one of "help me as a service provider meet the needs of my users". True, cloud providers will need to provision and move functions around their own cloud to meet the needs of dynamic workload, but costs here become more of an issue, as the end result is not seen as being pure value-add in itself. And, if the application is actually an aggregation of smaller functional services, VMware has a different problem—which is perhaps why it has just acquired SpringSource.
SpringSource itself bought another company, Hyperic, which provides functionality for managing Java-based applications (SpringSource provides tools for writing applications in the Java environment). If Hyperic can be rolled in to the VMware management environment, then VMware gets closer to being able to manage a service-based environment. A combined vSphere, vCenter and Hyperic will provide a solid capability for service providers to manage applications in their virtual environment.
VMware's current ace in the hole is that the management of the virtual environment by the incumbent systems management vendors such as IBM Tivoli, CA, BMC and HP still leaves much to be desired. The majority use VMware's own capabilities to manage a VMware virtual platform through APIs. Therefore, VMware has currently got a solid market in being either the main management provider for virtualised environments, or at least the main gateway to it. But, this will change as Microsoft becomes more of a play in the market and 100 per cent VMware platforms become more hybridised with Hyper-V in certain areas. Indeed, a further issue is that VMware can't, as yet, manage Microsoft hypervisors or images.
The lack of capability to manage the physical environment, even at a basic level, is the biggest threat to VMware. Essentially, it is left to the incumbents. But these incumbents know their futures depend on managing the virtual environment, whether it be VMware, Microsoft, Citrix, Parallels or whoever. Rest assured that each physical management vendor is working hard on replication VMware's capabilities, and as they do, the need for VMware's management tools becomes less obvious.
As Maritz himself stated, VMware needs to move up and down the stack. The purchase of SpringSource moves VMware into the application arena. But the moving down in to the physical world was glossed over—as far as VMware is concerned, it is IBM's, CA's, BMC's or HP's domain. A partnership (or acquisition) of a company such as LANDesk or Numara could provide a quick solution to this issue.
Maritz knows that the hypervisor is fast becoming a commodity—he stated that VMware has to chase the cost curve down. The main focus is on how to make the VMware virtual world so much better an experience than anyone else's. This is admirable in itself, but as the physical management players improve their virtual capabilities, VMware runs the risk of being squeezed between the rock and the hard place.
Without taking on the physical world, could the sainted Maritz be facing a different Eagles song: Heartache Tonight ?
We automatically stop accepting comments 180 days after a post is published. If you would like to know more about this subject, please contact us and we'll try to help.
Published by: electronicdawn Ltd.