BEIJING, April 2, 2022 /PRNewswire/ -- A new generation of information technology is accelerating the advent of the digital economy era, and data, as the information support of the digital economy era, data has become the fifth largest factor of production after land, labor, capital and technology, and is becoming an important force driving economic and social development.
Data is also an important digital asset of enterprises, has become an important engine to drive the digital transformation and upgrading of enterprises, data reliability, integrity and security, related to the viability and competitiveness of the industry and enterprises, business is always online, data is never lost, performance is endless, capacity is always sufficient, has become a new challenge in the era of wisdom. As the core carrier of data, the storage system pays more attention to important indicators such as reliability and performance, and Inspur Storage has launched the InRAID function through a series of technological innovations such as algorithm optimization and distributed hot spare space, breaking through the performance and reliability limitations of traditional RAID and providing a data base with higher reliability and better performance for digital transformation.
Capacity, performance, reliability is The main cause of RAIDIn the early stage of computer development, the data generated and processed is also increasing rapidly, and a single disk cannot meet the system's requirements for data storage capacity, read performance, security and reliability. RAID (Redundant Arrays of Independent Disks) technology came into being, which means that the array of independent disks with redundant capabilities, multiple independent disks are composed together to form a large disk system, so as to achieve better storage performance and higher reliability than a single disk, this redundancy capability is reflected in the failure of a disk where the data is stored, RAID technology can reconstruct and restore the data of the faulty disk to the hot spare, This is the way to keep your data safe. The reliability of RAID refers to the probability that a RAID will lose redundant data. For example, when RAID5 double disk fails, it is redundant loss, and before the failure of a single hard disk is completed, the probability of the failure of the second disk is the failure rate of RAID5. From an intuitive point of view, if the faster the single disk reconstruction speed, the lower the probability that the dual disks will fail at the same time, and the higher the reliability of the RAID.
The age of the digital economy, tradition RAID has new limitationsWhile the traditional RAID solves the problems of capacity, performance, and reliability, the idleness of its hot spare causes a waste of certain resources; in addition, when a disk fails, the array reconstructs the recovered data to the hot spare, and the write bottleneck of a single disk affects the reconstruction performance of the entire storage, resulting in increased failure efficiency and reduced reliability of the RAID. In addition, if the refactoring speed is too fast, it will have a greater impact on the performance of the host business, and the recomponing speed will have an impact on the storage reliability, how to balance the recomposition performance and the impact of refactoring on the service performance is also an important indicator to measure the performance and reliability of the storage system.
tidal wave InRAID, a raid-based and distributed innovationThrough years of technology accumulation, inspur has continuously innovated, proposed InRAID technology, improved performance through multi-threading technology and verification delayed write technology, increased performance by more than 50% compared with traditional RAID, increased reconstruction efficiency by nearly 20 times through distributed hot spare space and effective data reconstruction technology, and effectively controlled the impact on host performance through flow control algorithms, taking into account performance while improving the reliability of storage.
In InRAID technology, Inspur proposed a new type of space concept pack, and disk arrays are divided in packs. When storing data, according to certain rules, the iIO is assigned to the corresponding CPU core, a single RAID can process services concurrently with multiple CPU cores, and achieve physical space isolation through the pack, without the need to ensure data consistency through mutually exclusive synchronization technology, and improve the performance of the entire storage.
Improve performance by delaying check writesThe traditional RAID write process encounters two problems, one is that the read request needs to be completed to perform the XOR, and the other is that the IO is completed after the data disk and the check disk are successfully written.
Flowchart of traditional RAID processing
Inspur InRAID optimizes the above write process through APU technology (atomic parity update, atomic parity update), first using lazi-parity technology (check block delayed write) to delay the writing of the check block, and the data block can be returned to the upper layer successfully after the data block is written. Replace synchronous waiting with asynchronous process processing as much as possible, improve performance as much as possible, and save the integrity of the check block through APU.
InRAID data writing process
Limit optimization, refactoring performance improvement 20 timesInRAID technology breaks apart the fixed hot spare disks in the traditional technology into distributed hot spare space, which is regularly distributed throughout the disk array as the pack rotates. In this way, there will be no idle disks in the disk array, which effectively avoids the problem of waste of resources. When a failed disk occurs in the array, the data after multi-threaded reconstruction will also be concurrently written to the hot spare space of all the disks of the array, effectively avoiding the bottleneck problem of writing to a single disk, effectively improving the efficiency of reconstruction, and thus improving the reliability of the storage system.
Traditional RAID mode
InRAID mode
Based on the improvement of reconstruction performance through distributed hot spare space and multi-thread concurrency, InRAID uses active data reconstruction instead of full disk reconstruction, reducing the time and performance loss caused by invalid data reconstruction. At the same time, because only the valid data part needs to be reconstructed, the performance improvement is significant compared with the overall reconstruction.
At present, Inspur InRAID has been widely used in Inspur distributed, centralized and other storage products, and the mode is deployed to finance, communications, government, enterprise, medical, education and other industries to provide protection for customers' core applications.
Inspur Storage will continue to focus on the needs of industry scenarios, based on the storage strategy of "storage as a platform", continue to innovate, overcome data storage technology problems, create the ultimate technology, promote technological innovation with new storage methods, and bring a safe, reliable, cost-effective, easy-to-use and easy-to-manage storage platform to the industry.