LAMMPS in the cloud?

Hello all. The following is a solicitation from Dan Moore from GreenButton about the possibility of using LAMMPS “in the cloud” on a pay-per-compute-cycles basis. If you are not interested, please feel free to read no further and delete now. Even though we typically discourage use of this e-mail list for commercial purposes, I’m authorizing the posting of Dan’s message because I find the concept interesting and am curious if other LAMMPS users might find it interesting as well. This may be of some general interest to the LAMMPS users’ community, especially if cloud computing continues to grow.

If you’ve read thus far and are interested in the concept, I encourage you to read Dan’s message below and (optionally) respond to Dan (dan.moore@…3120…) and me (pscrozi@…3…) by e-mail (but don’t CC the LAMMPS user’s e-mail list in order to keep that spam down).

Thanks,

Paul

Greetings,

My name is Dan Moore and I work for GreenButton Ltd. I would like to solicit feedback from the LAMMPS user community about a new service proposition for LAMMPS users.

GreenButton specializes in helping organizations leverage the elasticity and scalability of cloud computing resources to bear upon processing-intense workloads. We do this by ‘cloud-enabling’ an application such that users can offload a processing job to the cloud, and have it completed and delivered back to them in a fraction of the time it would normally take…all with the simple push a button. This new service would allow LAMMPS users to simply pay for only the compute resources they need for a particular job, and saving you the cost & maintenance of running your own hardware clusters.

The challenges we see in maintaining on premise hardware is scaling the resources efficiently to match the need. Very often, the amount of compute resources needed averaged over the course of a year will be dramatically lower than the occasional need (a.k.a. “spikes”) for a large amount of resources for a given project. In that scenario, users are crippled by the limits of their resources, and typically need a lot more time to complete those jobs. GreenButton introduces a fundamental shift in resource planning strategy, such that organizations would no longer need to maintain expensive infrastructure hardware, but rather pay for only those resources that are needed in an on-demand basis. We also specialize in “job prediction” that allows a user to execute a job based on cost (“How much are you willing to spend?”) or time (“How fast do you want this job completed?”). This enables a much more efficient way of tailoring your compute resource needs to your budget or schedule constraints.

On a very high level, we aim to help our customers focus more on their core competencies, and less on managing the resources needed to support their competencies. In cases where IT investment is purely a supportive effort and not necessarily the primary focus, we’ve found enormous support for the GreenButton concept…primarily because we make the application less cost-prohibitive. More people and organizations could use LAMMPS for research without any up front IT costs…and still utilize existing IT resources in parallel with the GreenButton service. So, even though the technology could be construed as disruptive, the way in which we go about enabling applications does not disrupt your workflow nor devalue your existing resources.

I would very much appreciate getting a sense from this community as to whether or not your individual organization might be interested in this type of compute-on-demand service to lower your overall CAPEX and allow you to manage those costs based on what you’re actually using. You can reply to this thread or simply email me directly at dan.moore@…3120…. I look forward to hearing your feedback.

Respectfully Yours,

Dan Moore
GreenButton Ltd.
+1 206.295.9607 mobile
Dan.moore@…3120…
Skype: danmoore1969