Kauac is a cluster design and system integrator company. Its main added value is design and deployment of large systems both in Linux and Microsoft Windows Server environments. Its systems are leading in price performance in the HPC and server markets. Kauac architecture is in compliance with the petascale and green era and with heterogeneous co-processors paradigm. A customer driven strategy was chosen to fit performance, reliability, scalability, usability, availability, confidence, security and budget demanded. Kauac is prepared to develop systems in critical scenarios in a worldwide business model, competing at the same level with the biggest and the best.
Kauac means light storm. It comes from an ancient Central America idiom. Kauac is the power of nature on the move. Kauac allows the commitment between the Skies and the Earth, the dark and the light, the giant and the little, the past and the present.
Supercomputers are about Power. The domain of this tool means autonomy in all strategic areas of any nation. Supercomputers systems are the main tool to support scientists to understand the big challenges of today. Supercomputers process applications like oil and gas exploration, genomics, biochemistry and science of materials research, molecular modelling, rockets launch calculations, vehicles, aircrafts and satellites engineer simulations, climatic and seismic simulations and visualizations, nuclear forces research, microchip simulations and development and other complex tasks. Supercomputers systems are the tools to provide the insight to understand new horizons. They allow the human being to cross the line of knowledge.
Flame is the Kauac´s supercomputer. Flame was designed to deliver in any size a massive computing power to anyone, anywhere, anytime. Despite the complexity of the system, Flame was designed with a concept to be powerful and easy in mind. Its design recalls the same concept of a single motherboard scaled to the extreme of the technology possibilities.
Flame uses a building block subsystems architecture that provides a system to be easy to deploy, to administrate, to program, to use and to grow. From the end user point of view, each subsystem appears to be just a regular resource on a standard desktop computer. Users can remotely access the system with well-known and comfortable Linux, Microsoft Windows or Java interfaces.
Flame has a complete built in security subsystem. The main concept behind this subsystem is that, IF it is violated there will be no critical damages or consequences. This design works on the concept that a system may be violated but NOT its content. This subsystem works with strong cryptography in hardware processing in real time. It covers user authentication process with biometrics and external devices support, online transactions, individual and group access, application execution, collaboration with email, VoIP and video instant messengers and data access on the storage subsystem all in 2K or 4K cryptography. All data, applications, preferences that belongs to a user or a group are handled only by the user or group. Even system administrators have no access to content data of any user or group.
Availability and scalability
Thousands of resources for thousands of users. Flame may be configured to achieve very large and huge systems. Flame can scale up from a single node with a few core processors and scale out to several SMP nodes with hundreds of thousands of processors and co-processors of different types. Flame can run from a single application that use massively all resources simultaneously to thousands of users and applications sharing the resources at the same time. All subsystems may have different levels of redundancy and virtualization to stay on line anytime.
Flame is a Petascale supercomputer. Its design can deliver several PETAFLOPS of computing power; PetaBytes of memory RAM and process hundreds of PetaBytes of mass storage. Flame has a very high standard load balance processing environment and also was designed to archive the maximum power each resource can deliver. That means it was made to cross the edge of computing power and boost it to a next level of results. Flame balance concept extracts from the machine the best efficiency from the resources working altogether.
Flame has a heterogeneous co-processors ecosystem. Through its different subsystems, Flame integrates, in a single environment, standard super scalar CPUs processors with GPU vector and stream multi thread processors, programmable processors (FPGAs), graphics processors, among others. This flexibility offers to the users the opportunity to choose between several programming paradigms, type of applications and price performance ratios. This approach provides a way to solve any challenge adapting applications to get the best results.
A Flame system uses the state of the art components available in the world today. All recourses are carefully chosen and combined together to reach the best benchmark on the market. Kauac only works with the first class and economic stable component manufacturers. That guarantees a system with a commitment to quality, performance, upgradability and durability.
|Processors:||Intel Xeon Scalable families.|
|Vector processors:||Intel Xeon Phi family, Nvidia Tesla family.|
|Maximum processing x86 cores in SMP:||over one thousand|
|Maximum system RAM in SMP:||96TB (terabytes).|
|Maximum processing x86 cores in PGAS/MPI:||over 100 million.|
|Maximum system RAM in PGAS/MPI:||7EB (Exabytes).|
|Maximum Vector cards in PGAS/MPI:||over 80 thousand.|
|Visualization Subsystem||Same specs as compute subsystem, but with Nvidia Quadro|
|Maximum Quadro cards in PGAS/MPI:||over 80 thousand.|
|Maximum # remote viz users:||over a million with GPU
|Render terabytes of data in real time.|
|Tightly integrated with storage subsystem.|
|File system:||IEEL - Intel Enterprise Edition for Lustre.|
|Maximum file system size:||512 PB (petabytes).|
|Maximum single file size:||2.5 PB (petabytes).|
|Maximum file system aggregate IOPS:||over 100 million (average read and write 4K).|
|Maximum simultaneous users:||100 thousand.|
|Maximum aggregate data throughput:||2.5 TB per second (terabytes).|
|Maximum SSD cache size:||96 TB (terabytes).|
|Fabrics support:||Infiniband HDR or Intel OPA|
|# node connections:||4x 200Gbps IB or 100Gbps OPA ports|
|Aggregated link:||1Tbps with RDMA support.|
|Management network:||10 Gbps Ethernet dedicated network.|
|Metropolitan Long-haul RDMA 40Gbps ports are available
to up 100 Km links. Long-haul links may use existent
dark fiber infrastructure with latency of 5 μs/Km.
|Recommended Linux distributions:||Red Hat, SUSE, Ubuntu, CentOS.|
|System management:||Bright Cluster Manager.|
|Compilers:||Intel Studio and PDI suites.|
|Debuggers:||Allinea and Total View.|
|Parallel APIs and libraries:||OpenMP, OpenACC, MPI, PGAS/APGAS.|
|Graphics programing API: VSG Open Inventor.|
Any FOSS mathematical, parallel, virtualization,
cloud and server platforms, APIs and libs that
run on above Linux distributions win run on Flame.
Kauac works in an extreme outsourcing business model. This collaboration environment guarantees the best professionals and carefully chooses companies allocated to specific tasks. Kauac operation network counts with more than 200 companies and organizations spread around the world with more than 3000 professionals working on projects just on demand. On this network we find suppliers, value added resellers, commercial channels, distributors, wholesalers, manufactures, system integrators, international logistics, banks, economists, lawyers, warranty and maintenance networks, consultants, technical support companies, training, universities, research centers, and much more. Kauac has the same huge structure of any of the big iron IT companies, but without its costs. It works with top efficiency and skill and delivers products and services in the state of the art.
Supercomputers systems may require in some occasions large financial resources to turnover. Kauac´s operational workflow uses international irrevocable letter of credit to trade. With a LC applied by the customers to acquire a system, all parties of the supply chain, including the customer, are confident that the process will be completed at cost on time. Kauac works only with first class banks ensuring all parties a successful transaction. With this approach, Kauac trades with a very low turnover and can deliver large systems.
Kauac has a business concept with innovation similar to its technical cluster design. The extreme outsourcing results in a very low operational overhead to face the global market. A strategy that cuts cost and offers quality and productivity that complies with high standard requirements to maintain the top position. Kauac’s light governance approach easily adapts the international tendencies with minimum impact and provides an aggressive time to market. Kauac is ready to face challenges and even predatory competition making business in a worldwide scale with price, quality, performance and reliability at the same level of the biggest and the best.
Kauac is aware of the nature and contents of its customer’s application and provide a security relationship model with its partners. This method provides a close trust confidence between all parties involved in the business. The agreements and transactions are made in strong cryptography environment and only minimum disclosure between parties' identity is allowed. Resulting from this, only the customer knows the big picture. No partner, even Kauac, if the customer so desires, has a view of the complete scenario.
Kauac works with numerous manufacture partners to integrate and deploy the final solution that is designed as demanded. Kauac´s network integrators have locations worldwide with facilities throughout North and South Americas, Europe and Asia. They have manufacturing capacities of multiple thousands of systems every month. Whether you are in the world and no matter the size of the solution you demand, Kauac can meet your needs and ensure the highest ISO 9001 manufacturing standards.
Kauac provides a global warranty coverage from 1 to 3 years in parts and labor as well as onsite global warranty coverage. These guaranties that the systems are fully supported throughout its lifecycle, anywhere the systems are deployed in the world. Our global onsite warranty attends the high standards of 24x7x365 help desk support with 4 hour onsite response. This onsite maintenance network covers several countries in North and South America, Europe, Middle East and Asia.