For the last few weeks I have been evaluating Nimbula Director’s Beta, in the course of my experiments I had my confirmation that Director is a valuable piece of software and I started to think seriously to invest some more time on it.

However, Nimbula is pretty secretive about their road-map; even their Beta release (and relative documentation), while making an attempt to give you a taste of what the final product might be, deliberately omits features that would lead to the distinguishable traits of their product.

This is upsetting me because I would gladly see how Nimbula compares with the soon-to-be-incumbent OpenStack; I do not conceal my soft-spot for the former but, at the same time, I recognize the role the latter might play in the coming months in the Enterprise Cloud landscape.

The comparison is by no means accidental: while OpenStack is self declaring as a “a collection of open source technologies delivering a massively scalable cloud operating system”, the author (well, me) believes OpenStack might become the subject of indiscussed love of the big boys (Microsoft ahead of all) or, at least, a supported, integrated environment: look at what is actually happening with Hadoop in the BI landscape.

My rationale is further reinforced by the fact that Nova (OpenStack’s Cloud backbone) and Nimbula Director share a lot in common: they both focus on:

• extreme scalability
• private Cloud implementation with secure “Federation” towards public or remote Clouds (usual suspects here are AWS and RackSpace)
• run on COTS infrastructure (shared-nothing computing, low-cost networking)
• Interoperability with various virtualization technologies

It’s clear that Nimbula (supposedly) and OpenStack provide the eventual adopter a low threshold economic and technological cycle, reflected by low capital start and use of well-known Open Source technologies (and a lot of custom “glue”, mainly in Python for my pleasure).

So, while OpenStack is publicly attracting support of some relevant IT actors, my vision is totally obfuscated for what regards Nimbula: I said it’s a secretive company. So, while I don’t doubt they will come up with some strategic partnership at the launch (planned before June 2011), I don’t know the bride’s groom!!

I think that Nimbula might be a serious contender, being them a commercial company, by tying application stacks in their Cloud offering. VMWare, Amazon, RightScale: they all integrate entire deployments and application-related contexts. There is where IT is heading in the years to come.
Perhaps a little feature of Nimbula, the “Launch Plan”, predates this concept: Launch Plans are “specification describing how a suite of images will be launched and the parameters that control their placement and execution”; In the Beta Launch Plans do not implement all features, I suppose I have to wait the RC before having more insight.

Time will tell, Nimbula is and remains a product to watch out for in the next months, if momentum is built we will certainly see it.

For the curious of you, some characteristics of Nimbula:

• Private Cloud Computing stack running on X86 hardware
• Distributed storage (planned for final release)
• Virtual networking (layer 2) providing isolation and infrastructure services (e.g. DHCP. NAT, Vlan support, Routing, etc..)
• Customizable virtual images (KVM and Xen images are currently supported, VMware in the future?)
• Virtio devices support
• RESTful API and Web-based console implementing hierarchical object-based (i.e. “container-based” as Nimbula documentation defines it) policy management system providing:
  • Image and instance management
  • Launch Plans
  • Billing containers associated to Accounts (i.e. abstraction of organizational units or “Customers” as they are called)
  • Role-based policy management (users within Accounts) implementing delegation and object-oriented actions (on Containers)
  • Federation to external Cloud (the Beta supports only Amazon AWS, more are said to be added in the final release)
  • Instance bundling tools (KVM and Amazon instances in the Beta)