Both reasons are at best incidental, and here's why.
For VMware, buying Nicira makes sense. VMware gets dramatically improved network abstraction and improved independence from physical network architectures. The new stack makes it easier to deliver its central value proposition: an elastic, scalable, commoditized, self-healing cloud. Plus, Xsigo is not Nicira in a different package. Nicira is a pure network virtualization suite that focuses on core network abstraction, elasticity, and dynamic service delivery, all while harnessing the power of OpenFlow. It promises the "death of the VLAN" and encapsulated, dynamically provisioned virtual network architectures. Nicira is a platform for building networks.
Xsigo, on the other hand, is a proprietary I/O virtualization suite that focuses on unified I/O delivery over InfiniBand. Sure, Xsigo supplies much-needed abstraction for diverse interfaces like 10 Gbps Ethernet and 4 Gbps and 8 Gbps Fibre Channel, as well as other perks, like dynamic provisioning. But what it really does is deliver storage, network, and other more-abstract I/O types in a unified fashion, yielding top-of-rack savings and improved I/O delivery by reducing I/O interconnect equipment and enabling central I/O provisioning and automation. Good stuff, but it's not Nicira.
As to the second item--grabbing yet another promising technology to fuel its claim to being a serious hypervisor vendor--let's be brutally honest. Oracle is not even a contender in the hypervisor war, no matter how much marketing hype it throws at the claim. The rebranding work Oracle has done to try to cast itself as a "real" hypervisor vendor has been both frenzied and expensive, yet it's hardly a blip on the radar to VMware. Our latest InformationWeek Private Cloud Survey showed Oracle a good 40 points behind VMware in mindshare among respondents. Even Citrix is years ahead of Oracle in delivering a robust, feature-packed, market-tested hypervisor, and Citrix is hardly the market leader that VMware is.
Ever met anyone who runs an Oracle hypervisor in production for anything other than Oracle applications? Exactly.
Still, you don't get as big as Oracle by being stupid, and while I don't think Oracle bought Xsigo to keep up with the Joneses, I also don't think it's averse to conveying that impression. It also clearly hopes the purchase will convince those just starting virtualization projects--like small shops--that Oracle is an up-and-coming hypervisor vendor. That's attention its desperately needs to prop up a weak hypervisor product line.
But don't be fooled. Oracle bought Xsigo precisely because its gear is so I/O intensive. Oracle servers voraciously consume storage and network I/O. Its data engines are built to cluster and load-balance billions of transactions in dense, heavily utilized server farms--exactly the sort of environment that can benefit most from unified I/O delivery. And it's not like Oracle doesn't have inside sales teams poised to bundle Xsigo products with existing Oracle offerings to drive Xsigo sales.
But even if Xsigo flops in Oracle's existing customer base and generates no more publicity than it has already for Oracle's virtualization line, the buy is still a win in terms of overall market strategy. Despite a reputation for database that eclipses its numerous other competencies, Oracle has been frantically diversifying for years. In addition to its many acquisitions that provide deep synergy with relational databases, like enterprise search and business process management technologies, Oracle has been buying organizations with competencies in content management, voice recognition, event correlation, identity management, SIP communications, CRM, digital rights management, and a legion of business-focused services that run on top of its databases and drive core product sales synergistically.
So, while Oracle's Xsigo acquisition may not be what it first seems, it makes perfect sense. The company would have to lose three times over not to realize a benefit.