Virtual combat: Three stretched Cluster of disaster tolerance design

Stretched cluster is a double-edged sword, will use such as flowing, with bad but by its limitations.

The traditional vsphere cluster means that all ESXI hosts within a cluster are in a physical room. Stretched cluster, as the name suggests, extends the concept of cluster to a location that is not physically constrained. A host can be a different physical room in the same data center, or even a data center in a different city. Sounds pretty cool, huh?

See more highlights of this column: http://www.bianceng.cnhttp://www.bianceng.cn/Servers/virtualization/

The use of stretched cluster requires careful evaluation and design. Here are three steps to help you sort out your ideas.

What exactly do you want?

The big concept of disaster tolerance is so vague that everyone may have a different understanding. So we need to have some understanding of the following nouns.

Disaster Recovery

After a data center application is accidentally interrupted, the application is restored in another data center. The aim is that when the infrastructure layer appears large to failure, the application can recover in a short time and minimize the impact.

Disaster avoidance

Usually means taking proactive measures to avoid possible application interruptions. For example, the planned engine room power, hurricane, earthquake warning and so on. Before a disaster occurs, switch the application from a potentially affected datacenter to another secure data center. There may be a short period of service interruption during the switching process.

Downtime avoidance

Unlike disaster avoidance, downtime avoidance pay more attention to the interruption of service and the loss of data. This is also vmotion and storage vmotion to the point of their prowess.

Active Site Balancing

If a data center or cluster is specifically built for disaster tolerance, it must be ensured that there is sufficient free resources in the event of a disaster. It is likely that a large amount of resources will be idle in 99.99% of the time. We do not want disasters to happen, but we also hope to make more effective use of resources. What to do?

Active Site balancing refers to the efficient use of all resources in all data centers to avoid unused resources.

The following figure compares the site Recovery Manager and stretched cluster based on the above four aspects. See this link in the text

If you are clear that downtime avoidance and active Site balancing are your most important areas of concern, stretched cluster may be better able to meet requirements than SRM.