Harness the potential of your HPC today
What customers say
“The integration of OKA Predict with Slurm placed a solid foundation for automation of job submission, ensuring efficient usage of the HPC cluster, which would affect in positive way not only the simulations of the end-users but also the power consumption.”
Eddy Taillefer – Okinawa Institute of Science and Technology
“OKA Core has allowed us to identify non-optimal uses of the cluster disrupting its operation.”
Matthieu Marquillie – HPC Lead at Université de Lille
“With HPC users becoming increasingly conscious about the cost and environmental impact of HPC systems, the OKA tools are increasingly being put forward to our customers to help address these challenges.”
Andrew Dean – Sales Director at OCF
And shape your HPC for tomorrow
The OKA Suite features the right tool for each of your clusters’ area of optimization
OKA Core is the backbone of the OKA Suite
and you can try it for free now
HPC resources rarely operate at their full potential
Resource waste
Resource utilization rates are suboptimal
Easily identify areas of resource optimization
Energy
HPC is a heavy energy consumer
Optimize energy consumption
Costs
HPC is very capital and cost-intensive
Get a hold onto operating costs
Inflexibility
On-prem HPC is inflexible
Shape the future of your infrastructure
End-user productivity
These inefficiencies affect time-to-results for end-users
Optimize usage of resources in real-time
OKA’s battle-tested components help you tackle the most beneficial areas of your clusters’ rooms for optimization
Identify abnormal user behaviors and job submission patterns
Improve cluster overall quality of service
Limit waste of compute resources
What resources are left unused, while requested by your users?
Identify states of congestion and contention in your clusters
Plan and motivate future evolutions of your clusters
Easy-to-use interface and consistent integrated components
Per-cluster views and user management
The right component for each and every optimization area
Advanced data filtering to get all the answers you need, at any level
Forecast and detection
Plan in advance the behavior of your clusters: either globally or at the level of a job. Plan your projects on to
low-activity periods; anticipate peaks of load… become proactive and anticipate problems.
Costs
How much will it cost to run a specific workload in the cloud? Project the necessary budget based on the
desired scenario.
Then, monitor and control your actual Cloud consumption on a day-to-day basis.
Physical view
Follow the evolution of your metrics server by server and visualize where physically the “hot spots” are
located.
