Recommendation : 1. Reduce the impact of data from its storage and access
Are data replications between multiple Database Engine (Cluster) instances appropriate for sensitivity and availability requirement ?
Data
C People
A Planet
B Prosperity
Difficulty
***
Priority
High
Récurrence
OnUpdate
Tests
Is the duplicated data tracked and linked to its sensitivity?
Precisions
The same data can be duplicated to ensure availability and security. Redundancy is effective in securing data, but not all data justifies this precaution, which increases storage volumes and generates additional synchronisation processes. Only sensitive data should be secured with the most efficient redundancy mechanisms. Clustering of DBMS allows to manage the failure of an engine. However, if in addition each engine uses some redundant storage (Mirroring, RAID 5, ...) the same data ends up with a footprint that can be multiplied by 6 (most often a cluster contains 3 instances to manage the quorums in case of failure of a node).
Use Case
Technical documentation reports data expected reliability and level of redondancy required
Additional elements
Operational issues related to the project
Posts
Rule for assessing the level of compliance of the criterion
Formalized = 100 ; Planned = 75 ; Identified = 50 ; Ignored = 0 / 100
Life cycle
Déploiement
19 other criteria related to the recommendation: Reduce the impact of data from its storage and access
Data
Is the number of requests kept to a minimum (no looping) ?
Data
Are the slow query detection thresholds set effectively ?
Data processing
Does regulated data (personal, health, financial) comply with the recommendations for structuring these categories of data ?
security
Is sensitive user data secure ?
Data processing
Does the API provide limits, filters and the list of fields to return ?
Data processing
Is the data collected really useful ?
Data processing
Is sensitive data collected ?
Data processing
Does the data have an expiration date when it is deleted ?
Data
Is frequently accessed data available in RAM ?
Data
Are "live" and "dead" data handled differently (eg: Slow storage for "dead" data) ?
Data
Are EXPLAIN clauses used on "Slow queries" to optimize indexes ?
Data
Have the different data access solutions (queries, triggers, stored procedures) been tested ?
Data
Is a NoSql solution more efficient than its relational equivalent ?
Data
Is an alternative to the relational model being considered ?
Data
Are database indexes consistent with operations ?
Data
Is the removal of obsolete data being managed ?
Data
Can data be backed up incrementally ?
Data
Do implemented queries use joins rather than multiple queries ?
Data
Is an alternative to SQL queries used when possible (local storage or similar) ?