The business needs are changing, according to Wolfgang Dreyer. Big Data, HPC and Cloud are merging together in the industry. This means that the data resources are much more unified. It is necessary to optimize the analyzing process and to guarantee a certain level of services.
Adaptive Computing is working together with a lot of organisations in the software industry. One of them is OpenStack which Adaptive Computing integrates in its new product. Adaptive Computing is quoted by many research companies in the world, including IDC, Intersect360, and Gartner, as innovative owners and thinkers in the world, Wolfgang Dreyer said.
The new version of Moab, called Moab 8.0, was launched at ISC'14. Wolfgang Dreyer pointed out the key advantages of this version. He talked about the performance boost of Moab 8.0. It has a 2 to 3 times overall performance improvement. Users can perform 100K job submissions in a second and realize a higher throughput with the new product Nitro which comes on top of Moab.
Adaptive Computing is investing a lot in advanced status staging because with Big Data, users need to move the data around in their HPC environment, in different calculation centres, so data staging is a very important point, according to Wolfgang Dreyer. There is a multi-job work flow which helps users to have their computer resources much better utilized by not blocking the data staging time in the job time already.
In times like these where the power restrictions in the world are growing, the green factor in the advanced power management is one of the key functions that Adaptive's partners but also Adaptive's customers ask for. Adaptive Computing always has been a pioneer in getting green, already showing a lot of potential in this area before, but now the company introduces advanced functions like susoending parts of the cluster, hybernating or even a shut down of parts of the cluster to the software. The most advanced feature is clock frequency control of each single core in the CPU.
The usability of the software is one of the key points for the end user The HPC community is a community which is driven by innovation and reaching out to new user groups. New user groups are not so familiar with the software as traditional HPC users, so they demand a nice interface. Adaptive Computing also wants to make it user-friendly for the administrators who manage the big systems by offering them a graphical interface to see in one point where they may improve the utilisation of the cluster. The end users can submit through an end user portal jobs to the cluster.
For a company like Adaptive which is part of the middleware, it is very important to have partners who will communicate about the technology.
Wolfgang Dreyer showed the 3D topology for aware-scheduling on Cray systems. The big Cray machines have a very special network connection. It depends whether the job is run side by side on the CPU, or whether it is on different parts of the system. Adaptive Computing invented the 3D topology for aware-scheduling which guarantees that the user has predictable times for the run time of the job as the job is always bound together.
HP has launched the new Apollo systems - the Apollo 6000 and the Apollo 8000. Both series are run with Adaptive's add-on product Nitro, which is developed for high throughput computing. Adaptive provides scheduling and optimization for HP Helion selfservice HPC. Apollo, combined with Moab, even generate a greater energy saving.
Together with Bright Computing, Adaptive Computing delivers a whole middleware stack. Bright Computing provides the intelligent monitoring, the workload comes from Adaptive and the resource management is monitored by Bright Computing. This also brings enhanced green capabilities and automated health checks. The professional service offers an integrated solution for the customer.
Adaptive's website provides the opportunity to download the new products for test drives. Adaptive is also supporting users with test drives. Adaptive Computing is offering the users a chance to run HPC in a very smooth way, Wolfgang Dreyer concluded.