[Virtualization practice] one of the design methods for disaster recovery-Alibaba Cloud Developer Community

A clear idea is needed in disaster recovery design, which can help us consider the overall situation and take details into consideration. It is necessary to take commercial demand as the leading factor, rather than talking about the specific functions of a product at the beginning. I summarized the following three steps:

A learn more about business needs

the preceding figure lists some Business Parameters. From this article .

We focus on several elements:

RTO ( recovery time objective) : applications must be restored within this period of time after a disaster.

Theoretically, the disaster recovery solution supports the smaller the RTO and RPO, the better. However, we must not simply pursue the minimum value, resulting in unnecessary high costs, which is called OverEngineering. A good architect should consider from the customer's point of view and provide solutions to meet the needs.

When communicating with customers, be sure to break the casserole and ask to the end, how do RTO and RPO values come from? Most of the time, you will find that no one can make it clear. This needs to start from the application. For example, some applications have implemented high availability, such as MSCluster and LVS. Infrastructure that support this application do not need to overconsider disaster tolerance. Most of the time, Hypervisor can be satisfied with your own HA.

Risk

from the severity ( severity) and possibility (likehood) to consider. For example, financial institutions have very high requirements for this, and one of my clients cannot accept the huge losses caused by system downtime. Therefore, they require ZeroRTO and Zero RPO after risk assessment.

Second, consider the factors that affect the key architecture design ( Architecture Decisions)

Site:

local: some disaster recovery solutions can meet customer needs by Local implementation.

Dedicated DR Sites: whether a special DRSite is needed depends on the company's IT strategy and sustainable development. Of course, the cost has a great impact.

Shared DR Site: Shared DR Site can be used in addition to disaster recovery.

Cloud Based Recovery: the disaster Recovery solution of the Cloud service provider can be considered. For example, VMware hybrid cloud (vCHS) recently launched a specific disaster recovery solution.

StorageReplication

software: fully use Software to synchronize data and do not rely on SANReplication.

SAN based: most high-end storage devices support SANBased Replication. If you have special needs, you can also use software to implement advanced SANReplication. For example, EMC Recovery Point.

Networks between data centers

DR dedicated: completely dedicated to DR

MPLS: Public.

Determine whether the disaster recovery solution can meet the requirements based on the bandwidth and the amount of data synchronized. RTO and RPO

3. Evaluate suitable products ( Product Mapping)

there are many disaster recovery products and solutions on the market. We need to ask ourselves a series of questions to list what needs to be met feature, and then evaluate the indicators for each product.

Method 1: roughly evaluate several major aspects

such RTO RPO,Cost,F lexibility,Managability, etc.

Method 2: Detailed evaluation

reference:

Disaster Prevention and Recovery Architecture from RMI

DRBC Design- Disaster Recovery and Business Continuity Fundamentals

this article is forwarded from frankfan751CTO blog, original link: http://blog.51cto.com/frankfan/1288884 , if you need to reprint, please contact the original author.