Question
· May 24, 2021

Backup strategy 双机备份策略

HA backup strategy

we use Rose to do HA with cache2016.2, the database is placed on the hard disk array, which is essentially an instance, and the same array disk is mounted by rose switch, another mirror machine is made.

Is there any problem with this structure for disaster recovery and backup, I hope to give some advice. Thanks!

Chinese translation:

单位cache2016.2用rose做的双机,数据库放在阵列上,通过rose切换挂载同一个阵列磁盘,本质上是一个实例,另外做了个mirror机。

这样的结构在容灾备份上有什么问题没,希望给点建议。谢谢

Product version: Caché 2016.1
$ZV: 2016.2
Discussion (8)1
Log in or sign up to continue

Translation:

HA backup strategy
we use Rose to do HA with cache2016.2, the database is placed on the hard disk array, which is essentially an instance, and the same array disk is mounted by rose switch, another mirror machine is made.

Is there any problem with this structure for disaster recovery and backup, I hope to give some advice. Thanks!

@wenjie zhao 

If I'm interpreting your question (and its translation) correctly, it sounds like you will have multiple mirror members on the same disk array. Is this a mirror member in the same mirror or different?

If the same, I think part of the value of a mirror is that you can failover if a problem happens to a piece of hardware, for example. If both mirrors are running on one disk array, if that disk array has a problem, you can't really failover.

If a separate mirror, then you could have two separate mirrors failing over if your 1 disk array has a problem - if that's acceptable to you, I don't necessarily see a problem.

Realistically, it's hard to provide advice without knowing exactly the architecture you are considering and the types of problems you want to be prepared for. I'd refer back to my initial reply, which is that I'd think about mirroring as I would any other HA architecture and think about what will happen if single items fail - generally, I think you would want your HA to be able to work around a single piece failing.

Is this what you mean by Rose? It's not a technology I'm familiar with, but from this page and your description it sounds like an HA cluster with shared storage:

https://www.rosedata.com/index_en.php/Prodetail/index/proid/1

Do you have any specific questions about this setup?

Can you provide more detail on what kind of disaster events you are looking to handle? What is your desired behavior? Basically, I would suggest thinking about different disaster scenarios and how this environment would react.

Failover Strategies for High Availability

Here are the docs regarding different HA strategies including HA clustering and mirroring. The main point they highlight for a HA cluster with shared storage is that the disk then becomes a single point of failure. Adding a mirror instance on separate hardware provides an option if your RoseHA disk fails.

Thank you, Michael, a picture will help a lot.  It sounds like Rose is some kind of server clustering solution that allows two servers to share a disk.

@wenjie zhao, if you are using Caché 2016.1 today, I would suggest that you discuss this with your account manager, who is probably most familiar with your system.  Your account manager can bring in technical resources if necessary to help you.