Is The Cloud Good for Scientists?
There’s a great, although somewhat scientifically-centric story on the future of cloud computing that’s worth reading.
Essentially, the story asks: “Is cloud computing a useful computational platform for the computational problems you and your institution need to address?” In other words, why is computing on the cloud better than maintaining internal servers, hardware and all the resources needed to support them?
Right off the bat, the story points out a great reason: “Through the use of pre-configured virtual machines, scientists avoid any of the technical issues associated with getting an application to run — they just use it.”
But a good point to consider is that, since cloud computing is on the Web, the first and most obvious question to ask is whether the Internet connection used by your cloud computing service provides enough bandwidth to handle your app demands. And does it add too much latency? This author says that the cloud was designed as a “latency-tolerant network,” and may not be suitable for certain users – such as scientific and HPC computing. Interestingly, he points out that “many HPC algorithms and applications… bottleneck and die” even after just a few microseconds.
But for many other types of users, enterprises from large corporations to small and mid-sized web-based businesses, cloud providers offer plenty of bandwidth – even considering the presence of virtual machines. VMs can add to latency issues because of the time-sharing nature of their underlying hardware. This is so even though users may never have a clue of a VM failure running in a cloud environment. Some cloud computing providers take snapshots of the VMS as they run jobs, and if the hardware fails for one VM, then all can be restarted from that snapshot.
But while the author of this article may have mixed feelings about the cloud for the scientific community, he underscores, however, that “one strong appeal of cloud computing” is the simplicity of contracting with an Internet provider to access a lot of free-floating computational resources. It’s an “astounding bang for the buck” platform that needs very little capital expenditure, he notes.
Further he gives a boost to the Amazon Elastic Compute Cloud (EC2) as one example of a cloud computing service that is available to the scientific community, as users can specify VMS that can run on unused hardware resources within their datacenters as those resources become available. This way, scientists only pay for those resources they actually use.
That’s a key benefit of the cloud for all users – not just scientists, I might add.
In the end, after an analysis of the cloud for scientific computing, he adds: “The future looks bright, although there are some big clouds out there! Happy “cloud” computing.”
I, as the owner of a cloud-computing service provider who takes great care to provide proper bandwidth, security, cost-savings and other benefits to my customers, couldn’t agree more.
