Private-cloud pioneer Eucalyptus is furthering its partner-centric growth strategy by teaming up with Red Hat (s rht). It’s a mutually beneficial alliance, as Eucalyptus gets access to companies running Red Hat Enterprise Virtualization and the Deltacloud API, and Red Hat gets an IaaS capability to strengthen its case against rival VMware (s vmw) and its vCloud tools. Possibly more important than the actual announcement, though, is the timing: Eucalyptus is no longer the only software on the market that can be used to build internal clouds, and CEO Marten Mickos knows his company must act to fend off the these encroachments onto its turf.
As cloud computing industry watchers know, Eucalyptus was the first house on the block now populated by the likes of Cloud.com, Nimbula, Abiquo, OpenStack, VMware (s vmw) and others. VMware might have the biggest home, but it was Eucalyptus’s house where everyone hung out – at least until OpenStack moved in. The open source project, led by Rackspace (s rax), hijacked a good number of developers and highlighted a rift between Eucalyptus and its once-model customer NASA. According to reports, NASA found the Eucalyptus platform too inflexible in terms of scale and hackability, due to its “open core” nature.
Mickos thinks reports of this split were overblown. He told me NASA once made a statement that “we don’t think … was correct or relevant,” and the press ran with it. In fact, he pointed out, USAspending.gov is still hosted atop Eucalyptus within NASA’s Nebula cloud platform that serves government agencies. NASA has many cloud needs, he says, and the work it’s doing with OpenStack addresses just one of them. Furthermore, Mickos said Eucalyptus hasn’t found the limits of its scalability in its own testing, although he acknowledges it’s now up to the company to prove that.
Regarding OpenStack, it’s worth noting that, according to Mickos, the project already uses some core Eucalyptus technologies, and his company plans to contribute even more going forward. The two are somewhat strange bedfellows, considering Eucalyptus is currently focused on building a customer base for its free, open-source software, which is exactly what OpenStack intends to do when its platform is ready for production. One notable difference might be their target audiences: Eucalyptus targets enterprise customers, whereas OpenStack is built, at least, with service providers in mind.
Then there is the open core issue. Mickos reiterated to me that the vast majority of the Eucalyptus platform is open source, and that free, open-source deployments probably outnumber the open core Enterprise Edition, for which users must pay, “by a factor of a thousand.” Most Enterprise Edition customers, he said, buy it because it supports the VMware hypervisor, which means those customers have already conceded to some degree of lock-in. For those customers running KVM (e.g., Red Hat) or Xen hypervisors, the free, open-source version will do the trick more often than not. Downloads of the free version have been doubling annually, with 15,000 last month, Mickos added.
Long-term, the goal is widespread adoption of the paid Enterprise Edition, but now, Mickos said, it’s just a matter of spreading Eucalyptus use via partnerships with vendors such as Red Hat (s rht), Dell (s dell), HP (s hpq) and Canonical. As Mickos explained:
To be able to compete against giants in the industry, like VMware, you need to have a strong revenue model of your own in the open source business. Secondly, you must have a clear differentiation for customers, so they know what they’re getting when they pay for it.
He thinks Eucalyptus has both, and time will prove this strategy effective. Mickos knows a thing or two about building successful open-source companies, having led MySQL (s orcl) to web database ubiquity before Sun Microsystems bought it for $1 billion.
Getting back to Red Hat, Mickos calls it a “first step,” noting, “there’s much more that we can do with Red Hat and hope to do there.” What that means is unclear, although an OEM deal or even an acquisition doesn’t seem out of the question. As mentioned above, Red Hat needs a self-service IaaS component to round out its cloud portfolio, and Eucalyptus has proven itself in that regard. Assuming Eucalyptus doesn’t get absorbed into a large vendor’s cloud lineup (a possibility Mickos didn’t acknowledge when I asked the question), Mickos said, “As we see ourselves growing, I can very well see a scenario where there are pieces of technology that we would acquire.”
Either way, I think Eucalyptus needs to do something (subscription req’d). Its partner-centric focus might not fare well over the long run against whole-hog vendors such as VMware and Microsoft (s msft), or startups selling cloud-in-a-box software. Buying its way into a complete cloud vendor is one option, but there’s no shame in building the first real internal IaaS product and selling to a fellow open-source company like Red Hat, which will use it to power its fight against the growing number of hostile neighbors on the cloud computing block.
Related content from GigaOM Pro (sub req’d):
- Defining Internal Cloud Options: From Appistry to VMware
- Why OpenStack Has Its Work Cut Out
- For Open Cloud Computing, Look Inside Your Data Center