テストのためのパワーとしては仰る通り一つで十分です。
しかし、私はロードバランサーを通した時の動作。
特にセッションに関して心配しています。
仮にセッションの引継ぎに問題が起きた際にテストサーバーが同じ環境でなければ
再現と特定、及び対処ができないのではないかと思っています。
![[削除済みユーザ]](https://secure.gravatar.com/avatar/695023b4efd563721d13716b1457bc73.png?d=identicon&mode=crop&r=PG&s=42)
評価
52
翻訳 / 英語
- 2014/05/22 10:40:48に投稿されました
I think that a single device has enough power for the testing purpose as you told me.
However, I have a concern on the performance through the load balancer.
Particularly, I have a concern on sessions.
I think it cannot respond for the regeneration, the troubleshooting, and the rectification when there is an issue on taking over a session if the test server is not in the same environment.
However, I have a concern on the performance through the load balancer.
Particularly, I have a concern on sessions.
I think it cannot respond for the regeneration, the troubleshooting, and the rectification when there is an issue on taking over a session if the test server is not in the same environment.
評価
57
翻訳 / 英語
- 2014/05/22 10:46:07に投稿されました
One power sauce would be enough as you said, if it's only for a test. But I'm talking about the movement through a load balancer.
Especially I worry about session.
When some problem occur during transferring session I strongly believe we can not specify what was wrong and how we solve the problem if we didn't test under exactly same condition.
Especially I worry about session.
When some problem occur during transferring session I strongly believe we can not specify what was wrong and how we solve the problem if we didn't test under exactly same condition.
★★★★☆ 4.0/1
評価
55
翻訳 / 英語
- 2014/05/22 10:48:29に投稿されました
As you say, only one machine is enough for test. But we are concerned about the result through a load balancer, especially about the session. We think that we cannot reproduce an action and identify a problem and work around it, if the test server is not under the same condition when a problem occurs in taking over a session.
評価
56
翻訳 / 英語
- 2014/05/22 11:10:00に投稿されました
As the power for testing, one is enough, as you suggest.
But I'm concerned about the transaction, especially the session in the correspondence through a load balancer.
When any problem occurs in taking over a session, unless the test servers are under a same environment, it's difficult to cope with the problem for the reproduction and specification, I'm afraid.
But I'm concerned about the transaction, especially the session in the correspondence through a load balancer.
When any problem occurs in taking over a session, unless the test servers are under a same environment, it's difficult to cope with the problem for the reproduction and specification, I'm afraid.
