Page 338 - Cloud computing: From paradigm to operation

1 Framework and requirements for cloud computing

8.3 Data storage requirements

The data storage requirements include:
1) It is required for the CSP:BDIP to support different data types with sufficient storage space, elastic
storage capacity and efficient control methods;
2) It is required for the CSP:BDIP to support storage for different data formats and data models;
NOTE – Data formats include text, spreadsheet, video, audio, image, map, etc. Data models include relational models,
document models, key-value models, graph models, etc. (as described in clause 6.1).
3) It is required that the CSP:BDIP provides a flexible licensing policy for the databases;
NOTE – As database systems may be covered by vendor licenses, the CSP:BDIP that offers a database as part
of the big data service needs the ability to adapt the licensing conditions to the particular service and the
CSC:BDSU requirements.
4) It is recommended that the CSP:BDIP provides different types of databases;
NOTE – Examples of database types include relational databases (RDB), object relational databases (ORDB),
object oriented databases (OODB), NoSql (not only SQL) databases, etc.
5) It is recommended for the CSN:DP to expose application programming interfaces (APIs) for data
delivery;
6) It is recommended that the CSP:BDIP fulfils storage and database performance demands.

7) It is recommended that the CSP:BDIP supports a data retention policy covering a data retention
period before its destruction after termination of a contract. This is to protect the big data service
customer from losing private data through an accidental lapse of the contract.

8.4 Data analysis requirements

The data analysis requirements include:
1) It is required for the CSP:BDAP to support analysis of various data types and formats;
2) It is required for the CSP:BDAP to support batch processing;
3) It is required for the CSP:BDAP to support association analysis;
NOTE – Association analysis is the task of uncovering relationships among data.
4) It is required for the CSP:BDAP to support different data analysis algorithms;
NOTE – Data analysis algorithms include classification, clustering, regression, association, ranking, etc.
5) It is required that the CSP:BDAP provides flexible licensing policy for the analytical applications;

6) It is recommended for the CSP:BDAP to support user defined algorithms;
7) It is recommended for the CSP:BDAP to support data processing in distributed computing
environments;

8) It is recommended for the CSP:BDAP to support data indexing;
9) It is recommended that the CSP:BDAP supports data classification in parallel;
10) It is recommended that the CSP:BDAP provides different analytical applications;
11) It is recommended that the CSP:BDAP supports customization of analytical applications;
12) It is recommended for the CSP:BDAP to support real-time analysis of streaming data;

13) It is recommended for the CSP:BDAP to support user behaviour analysis;
NOTE – User behaviour includes user-related information, collected users' behaviour in real-time,
environmental information and the analysed information from the cumulative users' information on a
CSP:BDIP's storage. Scope of behaviour analysis is based on the user's agreement in advance.
14) The CSP:BDAP can optionally perform analysis of different data types and formats in real-time.

330