Graphistry and Pavilion Partner to Accelerate Graph Analytics Using RAPIDS and NVIDIA GPUs – HPCwire

SAN JOSE, Calif., April 8, 2021 Pavilion Data Systems, a leader in data storage, announced today its partnership with Graphistry to support large-scale, multi-GPU, and NVIDIA Magnum IO GPUDirect technologies to supercharge how data analysts investigate event and entity data.

Pavilion and Graphistry will showcase their joint capabilities at NVIDIA GTC in session # SS33029 Visually Investigating Patterns in Logs at Scale with Graphistry, RAPIDS, and Pavilion using the most performant, dense, scalable, and flexible data storage platform in the universe.

Attendees will learn how to make big log volumes (including multi-petabytes) easily explorable by simple low-coding of GPU and graph interfaces. The session will walk attendees through how Graphistry has integrated friendly data UI tools like Jupyter notebooks, Streamlit dashboards, and Graphistry visual analytics with the NVIDIA RAPIDS GPU ecosystem (Dask_cuDF, GPU Direct storage, DGX A100) and with the Pavilion HyperParallel Data Platform. The combined result is attendees will learn how to adapt their existing familiar workflows to a style where they can now see and explore their entire dataset.

The presenters will also provide recommendations to optimize log collection and GPU storage optimization while guiding attendees regarding how they can run and fork the open-source reference architecture providing attendees with new insights that will be shared with the RAPIDS accelerated data science community.

There are just a few visualization tools that take full advantage of NVIDIA GPUs with Magnum IO GPUDirect to deliver real-time graph analytics across petabytes of incoming and stored data. Graphistrys software and pipeline technology, combined with NVIDIA GPUs and Magnum IO GPUDirect along with Pavilions universally unmatched storage, enable the next generation of interactive visualization for incident response, SIEM optimization, and threat hunting across market segments, including financial services, federal agencies, and data science.

Historically, teams have been excited to analyze logs at speed-of-thought by trying GPU acceleration, but bigger-than-memory datasets have always been difficult to connect to, said Leo Meyerovich, CEO and founder of Graphistry. Now that we can connect GPUs to Pavilions HyperParallel Data Platform, we can automatically keep the GPUs running at high rates on large data sets without waiting on the storage tier, so analysts can focus more on investigating and less on massaging queries.

The Pavilion HyperParallel Data Platform can hold over 2 petabytes of usable capacity in 4 rack units, making it an ideal fit for data-intensive applications like Graphistry, said Costa Hasapopoulos, Field Chief Technology Officer of Pavilion. With Graphistry bringing automated queries across Splunk, Elasticsearch, Apache Spark, and CSVs, delivering the highest fidelity at scale requires NVIDIA GPUs and Magnum IO GPUDirect to enable analysts to visualize key relationships, patterns, and anomalies in real-time.

Pavilion is also presenting three other sessions at GTC21, including with partner OmniSci, as the two companies jointly showcase the power of Accelerated Analytics within a Data Science Platform. It is GTC # SS33294.

Attendees looking to understand how to virtualize, share and aggregate GPUs will find this session of particular interest: GTC # SS33026 and associated blog How High-Performance NVMe-oF Storage Accelerates CPU & GPU-Powered Virtualized Environments Demonstrated by Pavilion.

Attendees wanting to understand the competitive advantage of a multi-controller, parallel architecture will find this session quite informative: GTC # SS33030 and associated blog How a multi-controller storage architecture shatters expectations for modern applications.

About Pavilion

Pavilion shatters customer expectations and resulting organizational outcomes by revolutionizing data processing for modern AI/ML, HPC, Analytics, Enterprise Edge, and other data-driven applications. The Pavilion HyperParallel Data Platform, powered by Pavilion HyperOS, delivers unmatched performance and density, ultra-low latency, unlimited scalability, and flexibility, providing customers unprecedented choice and control. Learn why Fortune 500 companies and federal government agencies choose Pavilion. Visit http://www.pavilion.io

About Graphistry Inc.

As a spinout of UC Berkeleys Parallel Computing Lab, Graphistry, Inc. is supercharging how organizations interact with their data. Powered by NVIDIA GPUs, Graphistrys investigation platform creates visual fast paths through complex investigations by transforming diverse data into interactive graphs that visually answer complex questions like correlations, scope, progression, root cause, patterns, and outliers. Ideally suited for investigations in areas like cybersecurity incident response, threat hunting, retail anti-fraud, and anti-money laundering, Graphistry enables analysts to find hidden connections, pivot on the fly easily, and share their findings with other teams. Graphistry, Incs investors include Bloomberg Beta, Nvidia, In-Q-Tel, Greylock Partners, and angel investors, including Prof. Kurt Keutzer (UC Berkeley, ex-CTO Synopsys), Andy Chou (founder & ex-CTO Coverity), and Patrick OMalley (CFO Seagate). The companys headquarters are in San Francisco, California, with a satellite team in Austin, Texas. More information can be found at http://www.graphistry.com.

Source: Pavilion Data Systems

Read more:
Graphistry and Pavilion Partner to Accelerate Graph Analytics Using RAPIDS and NVIDIA GPUs - HPCwire