Heat removing method and device as well as heat adding method and device for CPUs

Abstract

Embodiments of the invention relate a heat removing method and device as well as a heat adding method and device for CPUs. The heat removing method for central processing units CPU is suitable for the servers with non-fully connected first CPU topologies, and comprises the following steps: determining a first CPU in a plurality of CPUs by a controller, wherein the first CPU is a CPU which has an error or which needs to be removed according to first indication information, and the first indication information is from the first CPU topology or a user interface; determining at least one second CPU, according with a preset condition with the first CPU, in the plurality of CPUs by the controller; sending second indication information to the first CPU topology by the controller; and after receiving the second indication information by the first CPU topology, removing the first CPU and the at least one second CPU so as to obtain a second CPU topology, and operating the second CPU topology. Through heating removing method for CPUs, the online removal of the CPUs can be realized, and in and after the removal of the CPUs, the systems can work normally, so that the user experience is enhanced.

Description

technical field [0001] The invention relates to multi-CPU interconnection technology, in particular to a CPU heat removal and heat addition method and device. Background technique [0002] With the rapid development of IT (Internet Technology, Internet technology) technology, the amount of data in various IT systems is increasing. It determines that the data and information it processes are the core business data and information of users, and are usually massive. Judging from the three most common types of applications in some key business areas: online transactions, business analysis, and databases, even if they are applied to an ordinary enterprise, the amount of data processed may be astonishing, not to mention that they are used in banks, When industries such as telecommunications and securities are running, they often face TB or PB-level data volumes. Such a large-scale data volume is also related to the production, operation and decision-making efficiency of commerci...

AI Technical Summary

Problems solved by technology

[0003] As a result, the multi-CPU interconnection technology is derived, that is, multiple CPUs are connected to each other through high-speed interconnection channels between CPUs (such as QPI (QuickPathInterconnect, QuickPathInterconnect) fast interconnection channels, etc.), so that multiple physical CPUs can pass through these high-speed The interconnection channels are connected to each other to form a resource-sharing server system. However, while the multi-CPU interconnection enhances the processing performance of a single server, it also brings some additional risks, because as long as there is any CPU in this multi-CPU interconnection system A fault may cause the entire system to hang up. To repair the CPU fault, you must power off the entire server system and then replace the CPU. This operation of powering off and replacing will inevitably cause system business interruption and seriously affect the continuity of the system. service hours

Images