A method and a device for scheduling Pod

Abstract

The invention discloses a method and a device for scheduling Pod. The method comprises the following steps of: creating a Pod at a master node of a Kubernetes system; and obtaining position information of a Worker node of the Alluxio storage system where the resource data for creating the Pod is located, and scheduling the Pod to be created on the Worker node of the Alluxio storage system corresponding to the position information. The method comprises the following steps of: creating a Pod at a master node of a Kubernetes system; acquiring position information of a Worker node where resource data for creating the Pod is located; the Pod application can directly use the memory data through the Worker node and then the Pod creation is dispatched to the Worker node, so that the limitation ofthe network transmission rate on the performance of the Aluxio system is avoided, and the performance improvement caused by the use of the Aluxio system is maximized.

Description

technical field [0001] The present application relates to the technical field of storage management, and in particular to a method and device for scheduling Pods. Background technique [0002] Alluxio is the world's first memory-centric virtual distributed storage system. It unifies data access methods and builds a bridge between the upper-level computing framework and the underlying storage system. Applications only need to connect to Alluxio to access data stored in any underlying storage system. In addition, Alluxio's memory-centric architecture enables data access speeds orders of magnitude faster than existing conventional solutions. [0003] The Alluxio system adopts the Master+Worker model, where the Master node is the management node used to communicate with each Worker node, and the Master will actively collect the data information loaded in the memory on the Worker node server. When the application needs to use data, it needs to establish a connection with the Ma...

AI Technical Summary

Problems solved by technology

However, since the Alluxio system is a distributed system, when there are multiple Worker nodes in Alluxio, the data will be distributed in the memory of each Worker node. Read from the memory of other Worker nodes through the network, so that the network transmission rate will become the limiting condition of the data reading rate. When the network rate is high, it will not affect the read and write performance of Alluxio, otherwise it will make the performance of Alluxio worse. unable to play

[0007] In view of the problems in the above-mentioned prior art when the network rate is low, the Pod application reads data from the memory of other Worker nodes through the network, how to improve the rate of reading and writing data of the Pod application in this case, so as to avoid the network transmission rate from The performance limitation of the Alluxio system, so as to maximize the performance improvement brought by the use of the Alluxio system, currently there is no related solution in the existing technology

Images