Inspur Completes the World’s Largest Single-Cluster Test Based on OpenStack Rocky
Inspur Will Contribute the Test Result to the OpenStack Community to Help Users Deepen Their Understanding of Rocky, and Provide Them with a Reference Case for Deploying Their OpenStack Private Clouds.
Inspur InCloud OpenStack5.6 (ICOS 5.6) has completed a test with a single cluster size of up to 500 nodes, which is currently the largest single-cluster test based on OpenStack Rocky in the world. InCloud OpenStack 5.6 is a commercial release of OpenStack that is optimized for extended functionality, usability, stability and performance.
In the 500-node test, the high-density and high-concurrency capability were verified. The concurrency of virtual machine creation can reach up to 2,000, while there were 20,000 idle virtual machines running in the environment. When the load of 20,000 VMs was added to 60%, the concurrency was still 1,000. The test result is leading in the industry.
At the same time, the test verified the high availability of large-scale clusters in extreme cases by simulating a power outage of a whole rack, which is a serious scenario in the real world. In an environment with a load of 180 virtual machines per rack, the evacuation was completed and access was restored within 5 minutes.
In terms of high performance, this test incorporated low-latency, low-jitter requirements of real-world application scenarios of customers, resulting in a packet loss of less than 10-7 in 48 hours and a multicast packet loss rate of 0.
Previously, Inspur InCloud OpenStack already possessed the capabilities for single-cluster design specifications with more than 1,000 nodes, and had successfully deployed large-scale industry clouds with a single-cluster size of 1,000+ nodes for radio and television customers. This test focuses more on the capabilities of ICOS 5.6 in terms of high availability, high performance, and high efficiency in a large-scale single-cluster cloud environment.
Moreover, Inspur ICOS 5.6 also performed well in deployment and delivery efficiency. It managed to deploy 500 servers and set up the base operating systems in 10 hours, build 323 OpenStack nodes in 5 hours and 77 distributed storage nodes in 3 hours, complete expansion and reconfiguration of 100 computing nodes in 1 hour, and finally deliver 500+ nodes in a single day.