Data Availability vs. 临界

Data Availability vs. 临界

In the modern world, information technologies and data are the foundation of running business. Data protection is crucial to ensuring normal business operations and reduce the likelihood of unplanned disruption. There are four objectives of data protection: data availability, data preservation, data responsiveness, and data confidentiality.

Data Protection Objective 观察
Data preservation The consequences of not having all data complete and accurate have to be thought through very carefully.
Data availability There is no utility in having data that cannot be accessed.
Data responsiveness Slowness kills—if response is too slow, the usefulness of information can go to zero.
Data confidentiality Confidentially has to be applied at all times

To protect data and satisfy the four objectives, businesses invest heavily on data protection solutions. 一些解决方案, especially high availability solutions, are expensive because they involve additional data center, 硬件, 软件, and data mirroring or replication tools. Data availability is vital to the day-to-day business operations.

The fundamental assumption is that a business needs to identify business data value to determine data availability needs. Simply put, if the data is deemed as critical then it requires higher availability. Storage Networking Industry Association (SNIA) forum published a data value classification to help identify data availability priority and five steps implementation guide. If the cost of implementing a solution is too high, the data value class and specific components need to be changed. 换句话说, if you cannot afford the solution, then the data value needs to be lowered.

SNIA Data 皇冠搏彩中心下载 Forum (DMF) Data value Classification

Data Value Class Data Availability RPO (Data Loss Risk) RTO(最大. 恢复时间) DPW (Copy Data Time)
1. Not Important to operations 90% 1周 7天
2. Important for productivity 99% 1天 1天 12个小时
3. Business important information 99.9% 2小时 2小时 10分钟
4. Business vital information 99.99% 10分钟 15分钟 没有一个
5. Mission-critical information 99.999% 1分钟 1.5分钟 没有一个

Five-step Implementation Guide

  1. Identify data value class
  2. Determine best solution
  3. Select specific components
  4. Check system cost
  5. Confirm decision of change

Is this the best approach? According to David G. 山, the author of “Data Protection: Governance, 风险皇冠搏彩中心下载, 和遵从性”, the answer is: No. No statistical correlation has been proven between the value of data and the need for the availability of that data. 例如, Amazon’s customer service site may not be deemed as business critical, but the availability is extremely important to the business and its customer satisfaction. On the other hand, an application is mission-critical may not require high availability. 例如, a Business Intelligence (BI) system may deemed as mission-critical for the management depends on the reports to plan pricing strategy. However, the BI system does not require that downtime to limited to seconds or minutes per year.

The assumption that mission-criticality leads to high availability may direct IT to provide higher availability that it is really necessary.

Equating mission-criticality and high availability can lead to “straight-jacketed” IT, 这意味着 IT may think it has to provide higher availability than is really necessary for key information, because it follows a formalistic approach rather than thinking through what availability each set of application data really requires. The revenue-generation, 操作, and decision-making processes of a business may run only during normal or extended business hours or at specified times that can be known in advance.

Even if the applications that support those processes are scheduled to run 24 hours a day, 每周7天, some reasonable amount of downtime, say hours in a year if absolutely necessary, may be acceptable if it can be spread out over time (say, once a quarter) and if it is planned.