Recently, the international authoritative research institution Gartner officially released the "Hype Cycle for Storage and Data Protection Technologies, 2023 (hereinafter referred to as the "Report"), which is based on forward-looking insights into market applications and After years of intensive technical cultivation, Inspur Information has once again been selected as a sample vendor for file storage.
Gartner releases 2023 storage technology maturity curve
In order to help users track the maturity and future potential of data storage technology in a timely manner, Gartner releases a Hype Cycle (Hype Cycle) analysis report on technology, applications and industry trends every year. In the report, Gartner used the priority matrix and market hype to provide an in-depth analysis of more than 20 popular data storage technologies such as NVMe-oF, software-defined storage, object storage, distributed file storage, and container native storage, and also announced The Sample Vendor list is used to recommend representatives of manufacturers with mature products and unique value in this field. This report has extensive influence around the world.
What is a storage solution suitable for efficient processing of key cloud digital intelligence applications?
With the deep integration of cloud computing, big data, artificial intelligence technology and the real economy, the digital economy has become a new engine for social and economic development, and its prosperity and development will inevitably bring explosive growth of data. It is predicted that by 2025, the amount of data in China will increase to 48.6ZB, 80% of which will be unstructured data, and it will become the world's largest data circle. This will make distributed storage that is good at storing and processing unstructured data storage and applications welcome. Come big benefits.
The report points out that with the explosive growth of unstructured data and the increase in file-based workloads, distributed scale-out storage architecture has been brought to the forefront of IT infrastructure planning to build scalable and low-cost distribution for unstructured file data. A modern storage platform is a top priority for users. Gartner believes that distributed file storage will usher in four major opportunities. First, in the face of the exponential growth of unstructured data, users are increasingly replacing vertically scalable NAS device systems with distributed horizontally scalable file system products to gain the benefits of linear expansion of capacity and performance. Second, enterprises are preparing for the exponential growth of unstructured data and seeking better data insights and integration with cloud storage to achieve long-term data lifecycle management. Third, big data analysis, artificial intelligence (AI) and machine learning (ML) applications have become typical use cases of distributed storage. Fourth, data modeling and digital twins will continue to drive the rapid growth of distributed file storage.
For example, today, when large models are in full bloom, the storage system needs to cope with the challenges brought about by the large scale of data, diversified data types, varying file sizes, and frequent reading and writing for the training and inference of large generative AI models. The demand for storage systems with 100GB bandwidth and 1 million IOPS has become the norm. Enterprises need to build high-performance concurrent file storage capabilities to improve model performance, generalization capabilities, and application effects.
For another example, in real-life 3D applications, the production business includes many processes such as original image uploading, spatial processing, modeling, and slicing. The amount of data processed online in real time is huge. The uploading of original images and data modeling require many concurrent clients. operation, and a single client often requires a bandwidth of more than 20MB/s, which places extremely high requirements on the immediacy of high-concurrency data processing.
Finally, the virtualization/cloudization market has grown rapidly in recent years, and more and more core and critical applications are moving to the cloud. Higher IOPS and lower latency are required during data access to ensure that key businesses run without lags; cloud platform business Rapid growth requires a distributed horizontally scalable file system to provide flexible space, making system expansion more convenient.
High-performance distributed file storage supports the use of multiple IO paths to read/write data to multiple storage devices. It can also be horizontally expanded to accommodate PB-level data, and supports high bandwidth, naturally adapting to the storage requirements of key cloud digital intelligence applications.
Inspur Information's high-performance parallel file system is designed for key applications of cloud data intelligence
Thanks to forward-looking layout and technological innovation, Inspur Information is one of the first domestic manufacturers to deploy high-performance parallel file storage. It adheres to the product concept of "storage as a platform" and accurately identifies customer pain points to create data storage for key cloud digital intelligence applications. The platform AS15000G7 can meet users' more extreme needs for data storage performance, management, integration and efficiency in scenarios such as oil exploration, autonomous driving, and weather forecasting.
- Extreme performance is the core cornerstone for accelerating efficient processing of key cloud digital intelligence applications. AS15000G7 high-throughput parallel file storage system is based on GDS and RDMA technology to shorten the I/O path; through intelligent metadata management, it significantly improves data access and retrieval speed; with the help of unique intelligent network optimization technology, it can shorten the delay by more than 50% , the small file-level transmission delay is reduced to milliseconds.
- Ultimate fusion is an effective way to accommodate massive amounts of heterogeneous data. AS15000G7 creates the industry's first all-in-one ultimate convergence architecture, which uses storage methods of files, objects, big data and videos for parallel access, multi-protocol real-time mutual access and flat expansion of the system, and maintains consistent semantics and lossless performance during the data access process. , thereby achieving efficient sharing.
- Ultimate management is an important guarantee for the transparency and controllability of the entire data life cycle. AS15000G7 can be equipped with AIStation scheduling and InView data management platform at the same time to monitor and manage high-performance servers, networks, and storage devices, achieving transparent and controllable data operation processes, and HDD/SSD life and failure prediction functions are accurate within 14 days. The rate is higher than 95%.
- Extreme efficiency is a powerful weapon to improve the return on investment of users' IT equipment. Based on automated data layering and migration, AS15000G7 can be flexibly configured on demand through four models of flash memory, disks, tapes, and optical disks: performance, balanced, capacity, and high-density capacity. To improve user investment return ratio.
At present, Inspur Information's high-performance parallel file storage has won multiple communications centralized procurement bids, and its product maturity has been fully verified. It has also provided services to six major banks, TOP3 insurance companies, National Meteorology, National Industrial Internet Center, PetroChina, Tsinghua University, and Fudan University. , TOP10 hospitals and other major industry customers provide support and guarantee. In the future, Inspur Information Storage will continue to adhere to the product concept of "storage as a platform", delve into data storage product solutions, help customers build safe, reliable, economical and efficient data infrastructure, and fully release the value of data.