Today, we will analyze a profile topic based on a Google Tech Talks case , shared by one of the company's engineers, and High Scalability magazine analyzed the story of an IT specialist from Google.
Based on the video of the speech and the post of expert, we have compiled a checklist: ')
The speed of data recovery depends on the simplicity of the backup organization.
Data storage cannot be scaled linearly - more disks and employees are not a solution - automation is needed
To achieve the required level of data availability, you need to diversify points of failure.
Only continuous testing can ensure the stable operation of data recovery systems.
The speed of data recovery depends on the simplicity of the backup organization.
Data storage cannot be scaled linearly - more disks and employees are not a solution - automation is needed
To achieve the required level of data availability, it is necessary to diversify points of failure, but redundancy alone does not guarantee work without any data loss and failures.
The main cause of the problems is service level errors, and not any malfunctions caused by weather conditions. Only data testing can ensure the stable operation of data recovery systems.
One way or another, there is no need to forget about the impact on the physical layer - here it is worth considering RAID technologies (for example, Google uses GFS, the analog of RAID), which allow you to work with geographically distributed data storage.
Tape copying is a good alternative for backups and a valid way to increase diversification (this tool is also used by Google)
Monitoring the situation helps to determine the level of a potential problem and respond adequately. Sometimes the human factor should be excluded from the recovery process.
The efficiency of the data storage system depends on your willingness to compromise and take into account a number of factors, including economic
When recovering data, it is important to pay attention to how they are used - it may be worthwhile to start with data recovery of current customers who are most actively working with the system.
With the growth of your company and as you add new products, not only does the company's capitalization grow, but the level of responsibility to customers grows as well.