The networking demands of big data analytics, flash-based and scale out storage, and hyper-converged compute solutions are driving a migration from legacy fibre channel to next generation IP-based storage networks. These next generation storage applications require an open, programmable, lossless and highly available IP storage networking solution in order to support their unique traffic patterns.
Arista's key benefits for Big Data acquisition and analysis include:
- Scale to Petabytes - Universal Cloud Network architecture, non-blocking designs, dense switching
- Lossless Transport - Deep Buffering with VOQ and DCB/PFC
- Leveraging the best-in-class scale and robustness of the Arista EOS
- Highly Available - Network designs, Self-healing Arista EOS
- Telemetry and Visibility through LANZ, VM Tracer, Hadoop Tracer
- Operational Expenses reduction via automated operations with CloudVision provisioning and change management
- Familiar GUI-based network administration tools for storage admins
- Return on Investment - Low power consumption, compact form factors
- Seamless migration to 25, 50, 100, 200, 400 and 800GbE for next generation IP Storage
The drive to develop and deliver innovative products and services in the future will be fueled increasingly by companies' ability to acquire and analyze vast amounts of structured and un-structured data.
Large and small enterprises are racing to acquire this capability by leveraging the vast computing power of the public cloud and by re-engineering their datacenters into private clouds.
Big Data provides a tremendous help to organizations making critical decisions that drive their businesses. At the same time, Big Data requires storing and efficiently processing terabytes or exabytes of unstructured data, which creates an extraordinary challenge for CTOs and Network Operators, requiring new and innovative technologies.
The Big Data framework is comprised of distributed file systems, databases, and data mining algorithms. For a successful implementation, Big Data deployments require a highly scalable, high-performance, and easy to manage interconnect, advanced turnkey management solutions, and familiar visualization tools.
With up to 24GB of dynamic buffer memory the Arista 7280R3 Universal Leaf platform complements the Arista 7800R3 Universal Spine to meet the IP/Ethernet storage challenge of massive east to west traffic by delivering a compelling IP Storage networking solution. This includes deterministic latency characteristics, any-to-any non-blocking host communication, open programmability via EOS and deep packet buffers capable of absorbing the largest of storage IO bursts and traffic patterns.
Arista Networks Storage networking solutions enable an organization to build clusters that keep up with the growth in data, minimize operational expenses, and deliver sophisticated data analytics.
- .Lossless Network for AI/ML/Storage/HPC with RDMA
- .Big Buffers for Big Data Applications
- .Deploying IP Storage Infrastructures
- .Scaling Deep Learning Performance on the NVIDIA DGX-1 Server with WekaIO Matrix
- .HPC Deployment Scenarios
- .Big Data Applications Design Guide
- .Arista Network Design Considerations for Hadoop Clusters
- .Big Data and Cloud Storage
- .Networking in the Hadoop Cluster
- .Hadoop Clusters Built on 10 Gigabit Ethernet: Intel & Arista
- .Rapid Automated Indication of Link-Loss (RAIL)
- .Big Data Needs Cloud Networking - Jayshree Ullal
- .IP Storage Next Generation Platforms White Paper