OpenStack upgrade process Slow

Every six months, the OpenStack Cloud management platform releases a new version, focusing on how to improve the upgrade process at this week's OpenStack summit.

The open source cloud Platform version is frequently updated without an automated, multiple-system upgrade path, and the destructive human processes left by the user are cycled through different openstack elements, such as the Nova Technology application interface and the Keystone Identity Management application interface.

"There is absolutely no way to do this," said David Andrew, an independent analyst at the conference. ”

He believes that chef cookbooks and Puppet modules should be upgraded and standardized to simplify the upgrade process. At the moment, cookbooks needs both updates and detailed configuration instructions, and now it's just barely applied.

At the same time, industry has been slow to progress on this issue. Another speaker was absent from the meeting.

Experts say upgrades are usually the idea behind a software development process.

"OpenStack developers have a lot of basic things to do, so the community doesn't think about upgrades," Boris Renski, an executive vice president at OpenStack Engineering Services.

Dell's outstanding engineer Rob Hirschfeld called on the community to develop more OpenStack built-in upgrades. He estimates that 25%openstack customers still use the Essex version because of the upgrade barrier, which is now too old. 50% of businesses run on top of the newly released Folsom Platform, and 25% run on grizzly.

According to Hirschfeld, it is not possible to migrate directly from Essex to grizzly at this time.

There is no such thing as a silver bullet, but he suggests that users facing escalation difficulties can take an iterative approach, rather than a drastic shift.

OpenStack as a whole for only two years, the chairman of the OpenStack Foundation, Alan Clark, noted that we have made some improvements, such as Modularity and API versioning, which cut off the problem of cyclic dependencies.

"As the code is packaged and deployed, it needs to be stable for the enterprise," Clark said at the end.

(Responsible editor: The good of the Legacy)