DCS storage operating system becomes the "wisdom engine" of the wave high-end storage

"Carefree Tour" is "Zhuangzi" a book of the first article, the author uses the imaginative imagination to describe an ideal state: "By the Heaven and earth, Yu Six gas change, to swim infinity", that is, to adapt to the nature of all things, to grasp the changes in the environment, free from any constraints, freedom of activity, to achieve "carefree" realm. "Carefree Tour" is very representative of Zhuangzi's philosophy, because of its philosophical in the uncanny imagination, written Kiwi magnificent and praised.

Coincidentally, cloud computing, large data age, with the data volume, the number of applications and the increase in the number of users, enterprise storage systems are facing the expansion is not easy, inefficient, complex management and poor usability problems, become a puzzle of high-end storage development of a "rope." How to break the limit, the storage operating system as a high-end storage of "intelligent engine", the overall management of the entire storage system hardware and software resources, the realization of the data "carefree" flow, play a high-end storage equipment should function, performance, reliability, as the storage of the core display of technical strength.

The operating system is very familiar, is a kind of special system software, it is responsible for managing the entire computer system hardware and software resources, a good operating system is vital to the computer. Similarly, storage operating systems are the system software responsible for managing all storage resources in the storage system, which plays a decisive role in the reliability, function and performance of storage devices.

Many storage vendors have their own storage operating systems, such as NetApp's unified storage operating system data ONTAP,EMC Symmetrix's storage operating system enginuity, Huawei's virtualized storage operating system XVe, and so on. The storage operating system has a great impact on the market competitiveness of the storage system, for example, the data Ontap in the design of a unified storage design concept, through the WAFL integration of the file system and block device layer, while in the bottom raid technology, snapshot technology and other key points on the innovative design, so that data ONTAP is the industry's best storage operating system. With this advanced storage operating system, NetApp quickly launches its products and becomes one of the top three in the storage market.

DCS to build high-end storage core competitiveness

Today, the tide of the research of the DCS (Data control System) storage operating system, is the wave of high-end storage technology cornerstone of the storage system, all kinds of advanced software functions are running on the DCs storage operating system, the future of DCs will be further migrated to the wave of a full range of storage systems. The following is the architecture diagram of the DCS storage operating system.

Schematic diagram of DCS storage operating system

In the conventional software layer, DCS storage operating system is the platform of storage System configuration management, remote replication, snapshot, cloning, service quality management, monitoring and alarm, storage virtualization and other software subsystem, which provides rich value-added function for high-end storage.

DCS has the functions of unified storage resource management, Multi-Protocol host interface, global high speed sharing cache, intelligent storage resource pool, high speed interconnection communication and global resource scheduling.

And in the storage hardware layer, DCs can be compatible with a variety of platforms.

As a whole, DCs realized the real virtualization design, focusing on the stability, real-time, reliability, management and deployment of the optimization, to create a wave of high-end storage of the four core competitiveness.

No.1 real-time response, business efficiency improvement

The real-time performance of the storage system is the key, the IO data sent by the compute server must be processed in time, otherwise it will cause IO timeout, and then affect the front-end computing business. Therefore, for the storage system, the real-time performance of the storage operating system must be effectively protected. According to the characteristics of the storage system, the scheduling of DCs operating system is optimized to ensure the real-time performance of the task.

When optimizing the thread scheduling algorithm, DCS storage operating system is specially optimized for multiprocessor. In view of the storage system performance more or the need for hardware support, so the 2-way or 4-way controller thread scheduling optimization is particularly necessary. The research of multiprocessor real time task scheduling includes three aspects: dispatching strategy, scheduling algorithm and scheduling decision boundary of real time task. These three aspects will have a significant impact on the real-time and performance of the system, so the DCs storage operating system is optimized separately.

No.2 Anrupanshi, defending data assets

Reliability is the cornerstone of storage, only stable and reliable storage can provide users with continuous service to ensure the security of user data. DCS storage Operating system has done a lot of optimization and improvement in reliability, including the optimization of disk system, the optimization of RAID system and the optimization of caching system, etc. The reliability of the storage operating system is ensured from several aspects, which provides a reliable guarantee for the storage system.

On the disk subsystem, DCS uses multiple levels to increase the reliability of the disk subsystem. Firstly, the link fault decision algorithm is adopted to solve the reliability of the back-end link of the storage system. This algorithm can effectively deal with the flash-break problem of back-end link and ensure the continuity of the access to the back-end disk array. Second, the disk block/bad problem, the use of jumping disk scan and repair algorithm. The algorithm tests the disk-specific areas to restore the damaged disks through the checksum and maps to other areas of the disk. At the same time, the algorithm can dynamically adjust the frequency of the scan according to the load of the disk. Thirdly, the disk intelligent replacement algorithm is used to solve the problem of disk life. The algorithm determines whether to replace the disk by counting the number of bad blocks and the usage time, and when the requirement is met, the algorithm migrates the data of the old disk to the new disk without affecting the front-end business, and realizes the replacement of the disk.

No.3 global acceleration, data automatic migration

In order to provide high-performance storage system, DCS storage operating system in two aspects of high-performance related design, respectively, high-performance storage pool and high-performance cache. High-performance storage pools can consolidate hundreds of of hard drives into a fault-tolerant large storage pool, which is then provided to upper-level use. High-performance caching provides a large-capacity caching system, reducing the probability of accessing back-end storage, thereby improving the performance of compute node access storage.

The storage pool layer blocks the disk data, organizes it by specifying the number of member disks (blocks), load balancing, and discrete distribution principles, and can change the stripe block flexibly. When a single disk fails, the storage pool layer can intelligently compute the distribution of hot spare blocks in the pool, and refactor with multiple member disks to avoid the risk of data reliability resulting from the excessive raid reconfiguration time caused by the traditional single disk write performance bottleneck. On the basis of reliable distributed raid, the data is further block-virtualized, logical volumes are formed, and distributed and multiple-replica metadata organizations are used. At the same time, the storage pool layer also implements the tiered hybrid storage, which can build a hybrid storage pool based on SSD and common mechanical hard disk, and realize a higher performance storage pool through SSD hard disk.

The high performance cache is realized by the global shared cache, which can share the cache space by the way of the global shared cache, so as to improve the cache hit rate and improve the performance of the compute node access storage.

No.4 Intelligent management, releasing storage potential

In traditional storage system, when computing nodes need a part of storage space, it is often a bit large enough space in the back-end storage system. This space is often not used immediately, but it takes one months or even longer to use the space, but since this part of the storage space has been reserved, other compute nodes will not be able to take advantage of these deployed but idle storage capacity. In this case, in particular, there are multiple computing nodes connected to the unified storage situation, will be a lot of space in advance, resulting in a great waste of storage space.

In the storage pool layer of DCs storage operating system, a resource allocation algorithm based on automatic thin configuration is adopted. In the initial stage, the computing system will perceive a storage space which is the same as the traditional storage, but the space is virtual space, it does not occupy the actual storage resources, the storage resources are not immediately assigned to compute nodes, but are distributed gradually according to the actual usage, so as to realize the efficient use of space.

To swim infinity without being stuck

Word, storage systems, especially high-end storage, often serve the core business system of large enterprises, how to ensure that the key data is not lost, the whole year business uninterrupted, is the ultimate mission of high-end storage, the survival and development of enterprises is vital; DCs can set data management functions, usability and outstanding performance and integration, Help customers improve storage architecture efficiency, achieve efficient access to data and unrestricted development of business, to achieve the data "on demand, not lag in the" carefree realm.

Today, wave storage is booming, H1 market growth 149.3%,dcs has been applied to high-end storage systems on a large scale. Supported by core technologies such as DCS, the wave will make a difference in the high-end storage market, as data Ontap drives NetApp into the world's leading storage vendor.