Red Hat ブログ
As I was catching up on email at an airport recently, one of my colleagues came up and described the "Monty Hall" problem and asked me what I'd do. Supposedly based on the TV show "Let's make a deal" (and hence named after the show's original host), the problem goes this way:
Suppose you're on a game show, and you're given the choice of three doors: Behind one door is a prize, say, a car; behind the others, a "non-prize," say, goats. You pick a door, say No. 1, and the host, who knows what's behind the doors, opens another door, say No. 3, which has a goat. He then says to you, "Do you want to pick door No. 2?" Is it to your advantage to switch your choice?
Now, while the non-intuitive "I will switch" is the correct answer (proven so by better brains at probability than me), one can't help but be left with the question "Really?" at the back of one's mind after hearing the correct answer. While I was running the problem again and again in my head, visualizing two closed doors and trying to convince myself that switching from one closed door to the other is the probability-increasing choice, an aha moment occurred - aren't customers doing the same thing i.e. looking at closed doors and trying to determine which offers the better option for their choice of infrastructure for the next-gen applications? Thankfully, unlike the Monty Hall problem, there is an open door that merits a re-look as organizations place their technology bets! Yes, am referring to the option to embark on an open journey to the private cloud.
The focus on datacenter optimization began with a focus on improving the utilization of compute resources. Virtualization, the innovation that was the catalyst for this has since commoditized and made open via KVM which not-too-incidentially is one of the top performing hypervisors in the market (more on this below). Now, the entire cloud infrastructure focused industry is staring at an amazing possibility to re-invent the infrastructure used to provide the on-ramp to the next generation of cloud-applications. And that makes KVM the "prize" behind the "open" door. Here’s why...
KVM started as a better way to do open source virtualization. Rather than building a purpose-built hypervisor like other proprietary and open source virtualization solutions, KVM takes advantage of the fact that Linux is a great general purpose operating system and provides Linux the ability to also function as a hypervisor. The performance, hardware enablement, security and certifications built around Linux can now be inherited by the KVM hypervisor without additional work.
KVM was submitted to the Linux Kernel community in Dec. 2006 and was incorporated into the Linux kernel in Jan. 2007. This is a testament to the elegance of KVM, since incorporation into the kernel represents a commitment that as Linux continues to evolve, KVM will continue to be enabled and supported. KVM performance leads that of other hypervisors, proprietary or open source. As of May 13, 2013, on the SPECvirt sc2010 benchmark, KVM leads VMware and other hypervisors in two-socket and four-socket server configurations and holds the only eight socket server scores. The highest KVM score on the SPECvirt benchmark represents over 550 high performing VMs running on a single 8 socket server.(1) KVM also holds the only fully virtualized TPC-C benchmark scores, showing near bare metal performance on one of the most demanding industry standard enterprise benchmarks.
In other words, KVM should be the natural starting point for your cloud infrastructure. With Red Hat Enterprise Virtualization and CloudForms 2.0 (which includes ManageIQ technology), you can implement a KVM-based open private cloud that offers self-service, charge-back etc. And with Red Hat OpenStack (currently in Early Adopter stage), you can implement an open private cloud that is suited for next-gen workloads, be it mobile back-end as a service or horsepower to power growing big data analytics. As called out during April’s OpenStack Summit, 71% of OpenStack implementations are based on KVM.
If you are in the throes of deciding how to take some costs away from proprietary virtualization infrastructure and at the same time invest in the foundation for the future set of applications, we would like to engage with you and discuss the possibilities of an open private cloud. We promise there are no goats for you to worry about!
(1) SPEC®, SPECvirt™, and SPECvirt_sc® are trademarks or registered trademarks of the Standard Performance Evaluation Corp. (SPEC). Competitive numbers shown reflect results published on www.spec.org as of May 13, 2013. For the latest SPECvirt_sc2010 results visit www.spec.org/osg/virt_sc2010.