I recently read an article in Desktop Engineering published November 1st 2013 by Kenneth Wong called The Zombie Computer Survival Guide. It discusses a phenomenon referred to by the author as “Zombie Desktops” where under powered desktop computers running complex CAE software simulation jobs become fully engaged in number crunching, to where they are almost at a standstill – “not quite dead but barely alive”.
It provided some very interesting statics around CAE software usage. According to the referenced survey of 1000 participants (performed by Desktop Engineering), 39% of the respondents reported solely running programs on workstations and only 11% reported running on clusters exclusively.
The article goes on to mention the disparity that 60% of those same participants report running complete system simulations and 58% report running large assemblies and models. 52% also reported running parts and small assemblies. The author notes:
“The disparity between the volume of simulation executed and the number of clusters involved suggests a significant segment of the participants may have accepted the zombie workstation syndrome as a fact of life.
Two easy methods will prevent the return of the zombies:
1. Set up an in-house cluster to tackle the CAE jobs and free up the workstations; or
2. Offload them to a remote cluster or on-demand cloud computing vendor.”
As a provider of “On-Demand” High Performance Computing resources we couldn’t agree more. We hear from end users that are forced to launch simulations overnight or on weekends because the simulations monopolize their workstation for hours if not days.
Because of this we are partnering with several independent software providers* to offer solutions to run complex CAE simulations via both an “On-demand” cloud based interactive sessions and as batch “Solver” options that leverage the resources of our large scale HPC cluster.
So whether you want to prevent your desktop from becoming a Zombie, finish your simulations faster, don’t want to wait for time on your in-house cluster or want to have the same powerful end-user experience on your underpowered home desktop as your workstation at work, we have a solution for you.
The article also makes a wonderful point of discussing the importance of different performance factors for High Performance Clusters like interconnect speed, size and type of resource options (High Memory, GPU Accelerated, etc…).
Nimbix’s solution has been built from the ground up specifically for High Performance Computing. It leverages bare metal servers (No virtual servers), the fastest interconnect currently available (56 Gig FDR Infiniband) along with different accelerators (GPUs, DSPs, Xeon Phi, FPGAs, etc…) and High Memory Multi-Core nodes.
Trying to deploy clusters on a typical virtualized commodity hosting provider systems is like trying to fit a square peg in a round hole. You may get the peg through the opening, but it will take a lot of work and the results are less the optimal. This means a lot of wasted time and money versus using a purpose built cloud HPC platform like Nimbix’s NACC.
Barriers to Setting up a High Performance Cluster
The article also explores different barriers to cluster adoption. The two largest being lack of budget and lack of skilled IT staff to manage the cluster. Several of the other barriers given could be grouped under these categories as well. For example: Cluster management software is too expensive (Lack of Budget) and Cluster is too complex to setup (Lack of skilled IT staff).
Nimbix’s solution helps address all of these barriers by providing the infrastructure and applications in an “On Demand” fashion that allows the users to pay for only what they need when they need it and still benefiting from the performance of a large cluster.
Additionally with Nimbix’s solution, the applications are already installed, configured and optimized. All the user has to do is submit the workload via an easy to use portal, API or command line interface for the solver or launch the application in an interactive session via the same portal.
They do not have to worry about setting up or managing the cluster, optimizing hardware resources, maintaining power, cooling and connectivity for the datacenter or any of the other headaches that go along with running a larger scale high performance cluster. Nor do they need to incur a lot of the same pain but on virtualized instances like on commodity infrastructure-as-a-service provider public clouds.
It was great to see an independent third party validating the same issues we are seeing and the needs we are building our offerings to solve. Please take a moment to check out our NACC solution at /https://platform.jarvice.com/landing/landing#tasks and if you do not see the application you are interested in listed please contact us at http://nimbix.net/contact/
(*It is important to note that each software provider has different licensing options, and requirements. However, we will work with you to find a solution that meets your needs.)