It has taken me a couple years for me to understand Infrastructure as a Service (IaaS) solutions like AWS and OpenStack. I’ve built infrastructures for Software as a Service (SaaS) environments. I’ve also been part of a larger team that built a community cloud for the service organization that I work at based on VMware vCloud/vDirector. I’ve traditionally worked in enterprises that have small to no in-house development teams. Most of my software customization experience is around packaged software products. Packaged software products lend themselves to single permanent workloads like an Exchange Server or a MySQL server servicing an in-house. They don’t surge. They are static until the server resources are upgraded.

So, a couple of years ago when I was looking at building a private cloud, I got introduced to cloud infrastructure products like Eucalyptus, Abiquo and Novell’s Cloud Manager. These solutions are pretty much based on an AWS model of cloud infrastructure which doesn’t lend itself to static workloads that my customers use. The advantage of a private cloud or any cloud is the ability to build applications that can be elastic or scale up or down in the amount of resources consumed. Basically, my enterprise customers are looking for a cloud solution that allows them to migrate their existing packaged product workloads. They are basically looking for virtualization + automation and chargeback.

They don’t care about application scaling because they don’t build applications. The OpenStack/AWS model makes no sense to these customers. They want to know how to do what they do today more efficiently not differently which is what the OpenStack model allows. vCloud allows these customers to do this type of migration without changing the underlying applications or worrying about getting charged for workloads that never scale down and capture use based savings. The stealth part of the vCloud solution is that customers get the API’s for development of “cloud aware” applications. So, when they are ready to move to an AWS model they have the capability baked in and they can extend their environment to vCloud partners. This is a pretty enticing sell to the enterprise that relies on packaged software.

OpenStack needs a cloud manager put in front of it. You can choose to build this interface or select a 3rd party solution from companies such as RightScale. The advantage of building your own is that you can get the exact feature set that meets your requirements. The disadvantage to building your own is that you need to know how to code. VMware has the complete stack that doesn’t require coding. Remember that the data center is run by network engineers and server administrators. They don’t program. Building a cloud manager is not a task for infrastructure teams.

I’m wondering how this will affect the race for cloud infrastructures. Will VMware’s mindshare in the data center help it achieve dominance in the cloud provider space or will developers lead the way with programming directly to AWS like API’s? If you run an enterpise data center what type of solution are you looking to provide to your customers?

VMware or OpenStack – Who will win the hearts of the data center?
Tagged on:                 

One thought on “VMware or OpenStack – Who will win the hearts of the data center?

Leave a Reply

%d bloggers like this: