Frank Lee, Senior Director of Product Planning at Infortrend, describes how the company's latest scale-out NAS offerings can help any organisation to grow without incurring unnecessary cost

Data growth never stops. It impacts all companies around the world and shapes new requirements for business operation. IDC forecasts that 'the Global Datasphere' will grow from around 50 Zettabytes in 2020 to 175 Zettabytes by 2025. To deal with such tremendous quantities of data, enterprises will need to modernise their IT infrastructure to a future-proof, scalable and cost-effective solution.

Infortrend thoroughly researched the global data growth trends before entering the scale-out NAS market. What we found was that, compared to structured data, unstructured data has the highest growth rate imposing unbearable management burdens on legacy IT. Customers' feedback also revealed challenges of ever-increasing unstructured data volumes and issues with storage silos management. Our EonStor CS scale-out NAS (CS) has been designed to address these issues with its scalable capacity and linear-increasing performance. CS can keep up as the business grows, making it the optimal data storage and management solution.

EonStor CS is suitable for a wide range of data-intensive industries, such as Media and Entertainment (M&E), High-Performance Computing (HPC), general IT, Surveillance, and Backup/Recovery. It is ideally positioned to meet projects across key health care and public sector settings and we continue our drive with key public sector procurement groups.

For M&E and HPC latency-sensitive applications, CS 3000/4000 with scalability through a scale-out expansion offers an easy way of managing growing assets while reducing performance bottlenecks. Post-production studios with more than 10 workstations and growing business can use CS as a shared media storage to get the highest and the most stable performance for their NLE workflow performed on 4K/8K ultra high-resolution videos via Adobe Premiere Pro, DaVinci Resolve, or Apple Final Cut Pro.

As for HPC and real-time data analysis, CS delivers maximum system efficiency with hybrid HDD/SSD cluster configuration for flexible data allocation in high-capacity and high-performance pools.

Enterprises can deploy CS' 2000 model as an archiving solution for large-scale surveillance projects with 1000+ 3MP-resolution IP-cameras and requirement of at least 90 days video retention period. To get the complete surveillance solution, CS can be integrated with major VMS providers, such as Milestone, Genetic, and Digifort. For general IT, CS can serve as a centralised storage for backup of employee's personal work files and office file sharing. As a Veeam Ready Repository, CS can be as well deployed in IT environments applying virtualisation technologies.

A huge advantage of EonStor CS is that it offers both high performance and capacity by supporting online scale-out and scale-up expansion: overall, CS systems can be expanded to increase up to more than 100GBps cluster performance and 100PB capacity. For mission-critical applications, CS can individually scale performance and/or capacity without requiring service downtime.

Another important benefit is lowered management overhead. By unifying multiple nodes to form a single namespace system, users can access the system via standard protocol. With symmetric file system architecture, where all nodes understand the disk structures, there is no need for stand-alone metadata servers or client-side software, and this effectively reduces data management hassles and costs.

Infortrend cares about data protection so that users can rest assured their data is safely stored/operated and applications have no single point of failure. To simultaneously satisfy the demands of system growth for various applications and provide data integrity, CS supports three kinds of cluster scale-out policies: traditional scale-out policies, erasure code and replica, plus Infortrend's exclusive distributed mode.

By configuring distributed mode, users can optimise capacity utilisation and increase performance, thus, lowering overall system TCO. Thanks to supporting single and dual node clusters, system deployment can be started with only one or two nodes and then subsequently expanded with new nodes as storage capacity and performance demands grow. In distributed mode, a file is evenly distributed in each node, and only one copy of it is written to storage. Unlike erasure code, distributed mode does not require data parity calculation and unlike replica mode, it does not create duplicated copies of data. This allows ultra-high 90% capacity utilisation and increased performance - by up to 50% compared to other modes.

If a node failure is detected, CS initiates self-healing features to automatically rebuild node data and prevent data corruption. To prevent performance degradation or data loss when multiple disks fail on different nodes, EonStor CS has a unique system architecture with an embedded layer of Infortrend RAID protection on each node. Unlike traditional scale-out NAS, where the damaged data must be reconstructed by calculating corrupted parts from other nodes through the network, data recovery in EonStor CS architecture does not occupy the network bandwidth as it occurs inside the node via RAID mechanism, so it is faster and system performance is not affected.

Not all enterprise data is equally important and frequently accessed; sometimes storage optimisation is vital to provide faster services for time-sensitive applications and archiving processed or outdated data. CS supports automatic storage tiering between All-Flash and HDD node for allocating data to the appropriate tiers based on file access frequency or file status. Less-used data can be stored in high capacity all-HDD pools, with hot data in high performance SSD pools. In this way, users can meet the requirements of shorter response time and massive capacity with optimised costs. The all-flash CS supports up to 3,400/2,000 MB/s Read/Write speed per single node and 40Gb/s RDMA node-to-node connection to provide lightning-speed access to mission-critical workloads of such applications as HPC and M&E.

CS can be scaled-out by demand to up to 144 nodes. When more than one storage is deployed, a common problem is uneven distribution of client traffic causing bottlenecks and overworking of some nodes, while others remain unused. To avoid this, EonStor CS scale-out NAS integrates DNS Load Balancing which automatically and evenly distributes the workload across all storage nodes in the cluster. It greatly maximises access performance and system analysis capabilities for applications where access for multi-client traffic is provided, like media streaming services or multi-tenant HPC infrastructures.

To accommodate any network environment, CS supports various I/O card combinations of such network interfaces as 10GbE, 25GbE, 40GbE. And finally, the cluster can be easily set up by a non-IT expert in just 30 minutes with the intuitive cluster deployment wizard.

Compared to other vendors' solutions CS is more cost-effective, so for financially savvy enterprises, Infortrend will often be the first choice as a scale-out NAS solution provider. The unique feature of CS system is that it can start from one deployed node (all data will still be protected by Infortrend RAID) and it can be, on demand, scaled out to 144 nodes. Also, unlike many competitors, Infortrend offers open-drive policy and perpetual licensing for system management and data services.

We have defined the future trajectory of CS scale-out NAS both from hardware and software perspectives. Speaking about the hardware platform, CS will support the popular NVMe interface specification for SSD model, 100GbE network, and Intel's next-generation platform. On the software side, we are actively working on features such as auto-tiering and dedupe to improve overall TCO, HDFS support for data analytics, and cloud integration for public data backup.

For more information around the CS, please contact or visit the website below.
More info: