The Bright Computing booth was a bustling hub of happy activity at SC15. Solving cluster deployment issues “headache free” has brought Bright Computing to the forefront in the under-appreciated world of HPC and cloud provisioning as well as generated a global customer (and fan) base. In short, Bright Computing provides comprehensive software solutions for provisioning and managing HPC clusters, Hadoop clusters, and OpenStack private clouds inside the data center or in the cloud. It’s easy to request a live demo, just click here.

According to Earl Joseph, VP for High Performance Computing at IDC, the HPC community appears to have transitioned from the experimental to an applied phase for cloud computing. The expectation is that we will now see an accelerated consumption of cloud computing by the HPC, scientific, and academic communities.

Earl Joseph (VP of HPC at IDC)

Cloud computing is great because it gives instant access to hardware resources, but users still have to provide the software. Bright Computing solves the software pain-point by giving users the ability to provision new cloud instances without sacrificing the ability to provision local hardware resources. Fundamentally, this appears the reason for the excitement shown by HPC experts visiting the Bright Computing booth. With a highly agnostic approach, Bright Computing appears to be taking over the cluster deployment marketplace for those customers who need to stand up HPC and cloud solutions.

A large number of Customer, Partner, and Analyst testimonials demonstrates the ubiquitousness of the Bright Computing Platform adoption. (Click here for the marketing brochure.) Just follow the previous hyperlinks to get a sense of the large number of organizations that already use the Bright Computing software. The following videos should give a sense of the toolset and applications:

Infrastructure

Monitoring

HPC applications

Apache Spark

Apache Hadoop

Interested in giving Bright a try?

Contact Bright Computing to get a trial software key, that will work for a cluster containing two nodes (For example: a headnode and one compute, or setting up 2 headnodes in HA config so that your teams can try stuff out in say a VM or KVM in multiple environments separate from production):

Sign up to access the customer portal: https://www.brightcomputing.com/Customer-Create-Account
Download the ISO: http://www.brightcomputing.com/Download (See note below)
Download the cmgui to control the cluster: http://support.brightcomputing.com/cmgui/
Review the cluster manager manuals: http://support.brightcomputing.com/manuals

Here is a video showing how to set up a basic cluster:

Linux cluster planning checklist

FQDN
Location Address
Room
Rack
U Location in Rack

Bright Version
Base OS

NETWORK
Internal network(Provisioning/Management eth0) Head nodeIP address on internal network
Netmask
Gateway
Domain
External network (Login eth1) Head node IP address on external network
Netmask
Gateway
Domain name
Name Server1
Name Server2
IPMI/iLO/Drac/BMC
IP Address
Netmask
Gateway
Domain name
High Speed Interconnect OFED (Mellanox/Qlogic/Base)
IP Address
Netmask
Gateway
Domain name
Time servers
Graphical Head	Yes
Graphical Nodes	No
Ethernet Switches
Are the switches managed or unmanaged?
Accessible
In the case of blade chassis, are the switches internal to the chassis?
Have the switches been configured (IP address and SNMP enabled)?
IB Switches
Does the IB switch have an embedded subnet manager?
Nodes
List of MAC addresses: Is a list of MAC addresses for the nodes available? This information could optionally be used to identify nodes.
List of switch port mapping: Is a list of switch port to node mappings available? This information could optionally be used to identify nodes.
Name of the default node category	Default
Default software image	default-image,
Additional/Special kernel modules
Disk setup RAID setup, partitioning and filesystem layout (if non-standard, specify details below)
Mount Points: Add any addiotional mount point that should be mounted across the nodes in the form of (NFS server, path on server, local mount point) (nfsserver, /path/on/server, /local/mount/point)
Workload manager
Workload Management System	PBS Pro
Queues
Specify the workload management queues that should be created
Third-party applications
Please list the third-party applications that should be installed
High Availability
Disksetup: Disk setup that should be used for secondary head node
Heartbeat network: The dedicated heart beat network between the two head nodes for heartbeat ping. Normally a direct connection between the two head nodes.
Shared storage: Type of shared storage
GPUs
Install CUDA in software images
Xeon Phi
Install Xeon Phi packages in software images?
Configure Xeon Phi devices?
Rack View
Populate the rack information?
Custom Requirements
Please fill in any additional custom requirements that are needed.