High availability:distributed storage and fault migration of data and data recovery mechanism ensure that INFINITY can guarantee data security and business continuity in the event of disk, network card and node failure, achieving high availability.
Erasure code and replicas:Supports two security policies: erasure code and replicas. Supports setting in directory. The erasure code can implement N+1~N+6 mode, and can maintain data integrity when up to 6 nodes are down at the same time.
Snapshot:Snapshots can be created online. The number of snapshots is up to 2048. The creation process is simple and the rollback time is short.
Access control:Supports two levels of access control for sharing and file systems, enabling complex access control networks.
Multiple backup mechanisms:support multiple data backup strategies, including real-time synchronization, timing synchronization, asymmetric synchronization, etc.
Fault domain:Different fault domains can be set for different cluster sizes, including data center/machine room/rack/chassis/host/disk, etc., to ensure that all faults in the entire fault domain do not affect data security and business continuity.
Unified storage:INFINITY supports three major protocols: file system, block storage and object storage, which can support applications such as database and Online use.
File system:It adopts a global unified namespace and is internally compatible with a variety of NAS protocols, including: CIFS/SMB, NFS, FTP, AFP, and WEBDAV.
Object storage:The object storage is a flat file structure, compatible with S3 and Swift interfaces, providing rich and convenient cloud storage functions and supporting metadata retrieval. Can support applications such as big data. Object storage can be used for media asset archiving, content trading, and application scenarios such as media big data.
Block storage:In combination with the iSCSI protocol, INFINITY can provide IPSAN services to support database-related services; INFINITY block storage services support virtualization engines such as KVM/XEN/VMware, which can be directly used as cloud platforms such as DATATOM HyHive/OpenStack/CloudStack. End storage for easy integration with cloud computing platforms for virtual machine and image management.
Shared-nothing architecture:INFINITY adopts the new Shared-nothing architecture to replace the functions of metadata nodes in traditional clusters with software, realizing software-defined storage.
On-demand expansion:support single-node configuration, support online expansion, up to 4096 nodes, capacity up to EB level.
Cluster removal:The nodes in the cluster can be detached into a single NAS node, which can flexibly meet your various needs.
Centralized maintenance:centralized maintenance across the WAN through the WEB, multiple operation and maintenance presentation modes, display and screening for bandwidth, node CPU, and running time, etc. Displaying the docking domain users: docking LDAP domain and Windows AD domain.
Scale-Out:The performance of the cluster is proportional to the scale. The more nodes, the higher the performance.
Global cache: INFINITY combines the isolated SSD caches of each node into a large global cache. All data will be processed in the global cache layer, which improves the cache hit rate and improves read and write performance.
Data tiering:INFINITY leverages the advantages of different media disks to store data hierarchically, storing more frequently read and write data in higher performance storage tiers, and improving business processing capabilities.
IO scheduling optimization:Optimize a large number of stripe IO scheduling processes for non-business scenarios, support stable ultra-high-definition video editing scenarios (4K/8K), cloud computing scenarios, database scenarios, etc.
Non-linear editing is a key part of video processing. For storage systems, it requires high stable read and write speed to avoid frame loss and blurred picture.
INFINITY's many innovations in technology make it fully meet the needs of large-scale non-linear networks in terms of stable read and write performance. INFINITY enhances the stable output of internal data through multiple technologies such as global cache, IO intelligent scheduling and data tiering. The 10 Gigabit and 2 Gigabit hardware network configurations provide stable read and write speed of up to 800 MB/s. Support multiple workstations for simultaneous nonlinear editing. INFINITY's wide range of flexible data security protection strategies, including replicas and erasure codes, protect your data security while balancing reliability and capacity.
With the popularization of high definition, more and more users plan to develop to UHD(4K/8K), but in 4K production, high speed, high stability and high scalability storage support 4K production is required. The production site is equipped with a 10G network card, and INFINITY provides high-end stable performance for the front end to support the current 4K production.
INFINITY uses M+N erasure code (N+1~N+6) and replicas security policy to protect data, while INFINITY optimizes for large files, each node can provide 800MB/S stability for front-end production sites. The bandwidth is mixed and the number of nodes is configured according to the actual service. The bandwidth and capacity of each node can be superimposed to support online expansion without affecting the services of the front-end.
The user’s media asset management system stores important materials accumulated over time. The security protection of the materials and the quick and accurate retrieval are important. INFINITY stores the erasure code, multi-replicas storage, WORM and version management, self-checking and self-repair through object storage. Designed to eliminate the risk of data loss in physical and logical errors and ensure long-term retention of media asset data.
INFINITY uses object storage, through the standard Amazon S3 object storage interface to store and call the media assets to achieve quasi-online media cloud platform construction, media asset data can be directly accessed, and seamlessly connect to public media through the standard S3 interface. The cloud can be used to put inactive or duplicate data on the public cloud based on the policy. The current application can quickly call back when it needs to access, and realize seamless connection with the future public media cloud. The maximum capacity of object storage that DATATOM has deployed in the project is 4.32PB so far.