問題描述
在 alpha 階段使用谷歌云功能進行生產 (Using a google cloud feature in alpha stage for production)
今天我讀到了 google 的“基於隊列工作負載的擴展”(鏈接到它的文檔) 它對我計劃開始的項目非常有用。但它說它處於 alpha 階段,不應該用於生產目的。所以我有點擔心它的結果。
- 一個處於 alpha 階段的功能通常需要多長時間才能成為穩定版本?
- google 之前會花時間切換到新的穩定版本嗎?從服務中撤出 alpha 版本?
參考解法
方法 1:
You can find more info about Google's launch‑stages on this page: https://cloud.google.com/terms/launch‑stages
Every product goes from alpha to beta to GA. But even beta‑stage products are generally not recommended to use in production.
Some examples of recent releases:
Cloud ml was released in alpha in march 2016, and it went to public beta at the end of september, which it still is at the moment.
Cloud Functions was released as alpha in february 2016, but is still in alpha.
I have no knowledge of any release of Queue‑based Autoscaling in beta unfortunately so I can't help you with that...
方法 2:
Serge's answer already provides the link to our launch stages doc, so that's good. Unfortunately it is hard to characterize exactly how long it will take to go between the stages, because it depends on a number of factors, including:
- The size and complexity of the overall product
- How much feedback we might need to get from alpha (generally about usability, correct functioning, etc.), and
- Similar feedback from beta (generally about usage characteristics at larger scale, and production readiness)
- Production dependencies (like for new hardware like GPUs or new CPUs, having enough of them distributed around the world to meet demand)
In most cases launch stages have a few months between them (Serge gave some examples, others would be recent launch of SQL Server VMs which were in Alpha and Beta for about 1.5m each).
As documented in our Terms of Service and SLA, we explicitly do not give any guarantees or promised timeframes of support for products that are not GA. Now, generally speaking we want customers who help us by testing to have a good experience, and so we try our best to provide notices and reasonable deprecation/migration periods before forcing transitions. But these launch stages are not meant for production use, generally. Some customers do it, but they are making their own predictions about how stable a pre‑GA product might be.
(by Agent Smith、Serge Hendrickx、Paul R. Nash)