WekaIO together with Destiny Corporation have published a reference design that is capable of delivering 5X runtime and wall clock time improvements for SAS analytics long running jobs, while also providing among the best storage economics.
The solution is available on-premises as well as on Amazon Web Services (AWS). Most of the Fortune Global 1000 companies rely on SAS analytics to power their businesses and transform data into intelligence.
Although SAS solutions are architected to take optimal advantage of multi-core CPUs, their performance and scale are often constrained by access to volumes of data and storage I/O responsiveness.
Even today’s leading on-premises and cloud storage solutions cannot keep up with modern CPU and server capabilities. As a result, SAS Grid and SAS Viya application servers remain vastly underutilized when operating across large data sets as they wait for data to be served from the storage system.
“Digital transformation is forcing organizations to modernize their IT operations, but inefficient application logic and performance limitations due to long-running jobs, processes, and pipelines negatively impacts the potential for business growth.
“Modernization is critical to overcoming challenges such as the better utilization, scalability, and stability of application servers, which fall over when trying to keep up with the storage performance and capacity demands.
“Our SAS analytics reference design solves these challenges and transforms data analytics for a more agile environment,” said Shailesh Manjrekar, WekaIO’s head of AI and strategic alliances.
“As a trusted advisor to our clients, we always look for new methods and technologies to help our clients improve their analytics work. Destiny chose to partner with WekaIO because we like the disruptive capabilities of this technology. They elegantly solve our clients’ problems with no change needed to the SAS code,” said Dana Rafiee, managing director at Destiny Corporation.
The reference design comprises SAS application servers configured either with SAS Foundation, SAS Grid, or SAS Viya software. These servers connect over Infiniband or 100Gbit Ethernet to a storage system running the Weka File System (WekaFS), the world’s fastest parallel file system, delivering enhanced performance at any scale while offering enterprise readiness.
The shared storage system starts at eight storage nodes with local NVMe drives, either from our leading OEM partners (on-premises) or Amazon Elastic Compute Cloud (Amazon EC2) instances (cloud), and can scale to hundreds of nodes, depending on the throughput requirements.
In addition, WekaFS seamlessly integrates object storage for the best economics on multi-petabyte data sets. The starting reference design can scale from one SAS client running 32 concurrent sessions to 8 clients running 256 concurrent sessions, WekaFS measures up to 106GB/sec of mixed Read/Write bandwidth, while only taking 320 SAS cores to achieve this performance.
The reference design is tuned to ensure the SAS application cores are not waiting for storage I/O and are fully utilized, without wasting any expensive compute cycles.
This reference design can be deployed either on-premises, on AWS, or as a hybrid cloud to get similar results. WekaFS matches the parallelism of the SAS Grid compute environment to enable the highest data loads and ETL (extract, transform, and load) processing while providing enterprise data management capabilities to SAS workloads.
“Using this reference design, organizations will be able to consolidate more users and workloads with the same SAS licensing in place, delivering exceptional economic value.
“Not only will this improve time to market and business outcomes, but it will eliminate racks of server infrastructure for a significantly improved total cost of ownership. The reference design leverages object storage data lakes to extend the file namespace and provide the best economics leveraging flash and HDD media,” added Manjrekar.
“Data-driven applications using SAS analytics are increasing in volume and complexity. With Weka and Destiny Corporation providing a reference design, customers have access to a validated solution that enables predictive and cognitive analytics in financial services (FSI), risk management, IoT, and healthcare markets,” said Amita Potnis, research director at IDC, Enterprise Infrastructure Practice.