In addition to Weibo, there is also WeChat
Please pay attention
WeChat public account
Shulou
2025-02-22 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > Internet Technology >
Share
Shulou(Shulou.com)06/03 Report--
Beijing, June 17, 2020-Beijing Zhiling Hainer Technology Co., Ltd. (hereinafter referred to as "SmartX"), the market leader in super-fusion, officially launched a high-performance, low-latency super-converged solution and SMTX Halo P series based on core software SMTX OS and Intel ®Aoteng ™persistent memory. For the first time in the industry, SmartX uses Intel Aoton persistent memory as App Direct Mode (Application Direct access Mode) as a core component in super-converged solutions. At the same time, through the end-to-end optimization of computing virtualization, storage network and storage media, the industry-leading high-performance and ultra-low latency server virtualized storage performance is achieved with limited computing resources. In the cluster test, only 3-node P-series all-in-one machines are needed to achieve 1.2 million IOPS, while the application latency is kept below 100us.
With its excellent product stability and characteristics, SmartX products have been used in the production environment by users in banking, securities, insurance, funds and other financial industries. However, when users want to apply hyperconvergence to core systems and replace bare metal servers and all-flash arrays, they still lack confidence in the latency caused by hyperconverged computing virtualization, network, and storage layers. To this end, SmartX launched a project codenamed Sailfish (swordfish, the fastest fish in the ocean) in 2019, which uses Intel Aoten persistent memory App Direct mode for cache acceleration and optimizes all aspects of SmartX superfusion software SMTX OS.
Intel Aoten persistent memory has added a new storage level between memory and SSD, with ultra-low access latency, ultra-high life and reliability, as well as persistent storage and byte access capabilities. In App Direct mode, specific applications can directly access independent persistent memory resources brought about by Intel Aoton persistent memory.
To take full advantage of proudly persistent memory, the Sailfish project optimizes end-to-end virtualization of computing to storage networks and storage media. Due to the deployment characteristics of the super-converged architecture, the above performance must be achieved with limited computing resources, which brings many challenges to the Sailfish project. Based on this, the SmartX R & D team has made the following major improvements to the system:
Storage layer
The Sailfish project takes full advantage of Intel Aoton persistent memory App Direct mode's low latency and persistent storage capabilities to preserve the most frequently accessed data in the cluster. At the same time, in order to ensure the reliability and high availability of the data, the cached data will also be saved to other nodes in the form of copies. Moreover, SMTX OS makes full use of the byte access feature of persistent memory to redesign the log to effectively solve the problem of log write magnification. And through the DMA way, the data copy task between memory and persistent memory is devolved to the hardware, which greatly improves the efficiency of memory copying, and does not generate additional CPU resource requirements while improving storage performance.
Computational virtualization layer
The storage virtualization function is devolved from the virtual machine to the storage software stack through SMTX ELF Boost mode, which avoids the additional performance overhead caused by the IO request passing through the virtual machine. At the same time, the memory copy on the IO path is avoided by sharing memory with the storage system by VM.
Storage network layer
Through the use of RDMA over Converged Ethernet (RoCE) for network IO acceleration, and the implementation of the network protocol, devolved to the network card hardware to complete, make full use of the acceleration function of the network card hardware, and greatly reduce the performance overhead of distributed storage in network communication.
After the above optimization, the IO performance of the virtual machine is improved by 3.2 times, while the latency is reduced by more than 85%.
At present, the technology used in the Sailfish project has been integrated into the latest SMTX Halo P series of all-in-one machines. The SMTX Halo P series of all-in-one machines can not only be competent for core application scenarios such as transactional databases and machine learning that require stringent IO performance, but also increase virtual machine density with higher performance, further simplify IT deployment and reduce the total cost of ownership. The SMTX Halo P series of all-in-one machines have been tested and deployed in the financial industry.
Compared with a mainstream high-end all-flash array in the market, the 8KB random read performance of the three-node SMTX Halo P series all-in-one computer is 1.2 million IOPS, which is 1.4 times that of the all-flash array, and the sequential read bandwidth is 25GB, which is 2.1times that of the all-flash array. At the same time, the application delay is only 100us, while the controller delay of the all-flash array is up to 500us. Not only that, the P-Series all-in-one machine still maintains the characteristics of simple and flexible super-fusion architecture, and provides a very competitive total cost of ownership compared with mid-and high-end arrays.
Zhang Kai, co-founder and CTO of SmartX, said: "SmartX is committed to creating the best super-converged infrastructure. In the Sailfish project, we provide best practices for Intel Aoton persistent memory super-converged deployment. The test data show that through the introduction of new storage technology, the performance of the superfusion system can be greatly improved and the practical deployment problems of key application scenarios can be solved. SmartX will continue to work with Intel and other industry professionals to explore the application of more cutting-edge technologies in super-fusion systems, leading the next stage of super-integration innovation. "
Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.
Views: 0
*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.
Continue with the installation of the previous hadoop.First, install zookooper1. Decompress zookoope
"Every 5-10 years, there's a rare product, a really special, very unusual product that's the most un
© 2024 shulou.com SLNews company. All rights reserved.