To improve transaction performance ,POLARDB What optimizations have been made at the kernel level ？
Take the answer 1：
Bypass Kernel PolarDB Born in 2015 year , because RDMA The emergence of high-speed networks , Make the network bandwidth close to the bus bandwidth .PolarDB Make bold assumptions , That is, the bottleneck of database will shift from network to software stack in the future . therefore PolarStore A lot of Bypass Kernel The design of the . The first is the new hardware NVME and RDMA Use , Get rid of IO User mode and kernel mode interaction during access .
See Alibaba cloud about PolarDB Official instructions given : https://help.aliyun.com/document_detail/426488.html
本文为[Alibaba cloud Q & A]所创，转载请带上原文链接，感谢