Why a Cloudlet Beats the Cloud for Mobile Apps
the article taken from http://lewisshepherd.wordpress.com
Posted on December 13, 2009 by lewisshepherd
Sure, you know cloud computing. You also know a bit about so-called “private clouds,” which enterprises and government agencies are exploring as an option to combine the power and scale of virtualized cloud architectures with security and control over data.But what do you know of Cloudlets? They may just be a key to the future of mobile computing.
That’s a possible conclusion from the results so far of a Microsoft Research family of projects called MAUI, short for Mobile Assistance Using Infrastructure. The MAUI approach is to enable a new class of CPU-intensive, and data-intensive, applications for mobile devices – but enable them in a new way. Today’s mobile devices can’t run such apps, at least not well. And if they stick to the cloud they may never do so.
I’ve just read a fundamental MAUI paper published last month in the IEEE’s Pervasive Computing journal: “The Case for VM-based Cloudlets in Mobile Computing” (November 2009, co-authored by MSR’s Paramvir Bahl along with colleagues from Carnegie Mellon University, AT&T Research, and Lancaster University).
The paper is worth reading just to get a reality check on the current cloud mania in popular technology; topics include: The Limits of Cloud Computing; Why Latency Hurts; WAN Latency is Unlikely to Improve; and Bandwidth-Induced Delays Also Hurt.
Illustration from the IEEE paper
And we can’t rely on the magic of cloud computing to deliver better resources. In short, it is unfortunately inescapable that ”long WAN latencies are a fundamental obstacle… Even trivial user-application interactions incur these delays in cloud computing.” And beyond those trivial apps, we already have mobile apps trying to deliver advanced, computation-intense capabilities like speech recognition, NLP, assisted vision, machine learning, and augmented reality. Those apps suffer because cloud computing isn’t powerful enough to support them seamlessly across a WAN like the public internet.
So MAUI researchers believe that “Rather than relying on a distant cloud, the resource poverty of a mobile device can be addressed by using a nearby resource-rich cloudlet.” In their paper, they write:
We put forth a vision of mobile computing that breaks free of this fundamental constraint. In this vision, mobile users seamlessly utilize nearby computers to obtain the resource benefits of cloud computing without incurring WAN delays and jitter. Rather than relying on a distant “cloud,” a mobile user instantiates a “cloudlet” on nearby infrastructure and uses it via a wireless LAN. Crisp interactive response for immersive applications that augment human cognition is then much easier to achieve because of the proximity of the cloudlet. - “The Case for VM-based Cloudlets in Mobile Computing.”Here’s a bit more detailed description of what the cloudlet looks like:
Cloudlets are decentralized and widely-dispersed Internet infrastructure whose compute cycles and storage resources can be leveraged by nearby mobile computers. A cloudlet can be viewed as a “data center in a box.” It is self-managing, requiring little more than power, Internet connectivity, and access control or setup. This simplicity of management corresponds to an appliance model of computing resources, and makes it trivial to deploy on a business premises such as a coffee shop or a doctor’s office. Internally, a cloudlet may be viewed as a cluster of multi-core computers, with gigabit internal connectivity and a high-bandwidth wireless LAN.There’s one last bit of important description from the IEEE paper about the cloudlet scenario,
The mobile device functions as a thin client, with all significant computation occurring in the nearby cloudlet. Physical proximity of the cloudlet is essential: the end-to-end response time of applications executing in the cloudlet needs to be fast (few milliseconds) and predictable. If no cloudlet is available nearby, the mobile device can gracefully degrade to a fallback mode that involves a distant cloud or, in the worst case, solely its own resources. Full functionality and performance can return later, when a nearby cloudlet is discovered.”That sounds a lot like typical WiFi scenarios, doesn’t it? Mobile devices and their users have already become attuned to the hotspot paradigm – it might be interesting for that to be extended to hotspots of dense computational resources which light up your handheld or goggles and transform them from normal devices to immersively augmented supercomputers.
There’s much more work to be done, on technical issues like dynamic VM synthesis and on figuring out the business model for cloudlets. As the paper’s authors muse, “Is deployment driven bottom-up by business owners installing cloudlets for the benefit of their customers, much as they install comfortable furniture today? Or is it driven top-down by service providers who share profits with the retail businesses on whose premises cloudlets are deployed?” These are fascinating questions which have puzzled ISP’s and wireless providers for a decade now.
I’ll certainly follow up on the cloudlet progress in Microsoft Research in future posts.
