Advances, Systems and Applications
Table 1 Summary of evaluation findings
From: Cloud repositories for research data – addressing the needs of researchers
Criteria | Evaluation |
|---|---|
Trustworthiness (by end users) | Users were particularly keen that there data should be automatically replicated. This was seen as a major advantage over storing data on portable media and the perception of users was that their data would be safe. |
Quality of service | The facility to have multiple copies of datasets dispersed across multiple physical locations and multiple commercial storage providers was positively received. Users were concerned that data stored in individual commercial cloud providers may be lost or become unavailable. Users were more concerned about knowing the hosting environment of their data rather than the exact physical locations. |
Cost-effectiveness | The cost optimisation was seen to be useful, particularly if it provide to be more cost-effective than a rapid estimation of the storage costs by hand. Itemisation of the storage |
Facilitate data reuse | The structured approach to entering metadata was seen as very useful to enable reuse, as well as the ability to share data with other users. |
User workload | The classification of the data required by the system is already carried out by the users interviewed. The work required to enter metadata was not seen as a major burden, depending on the frequency and scale of data that is deposited. |
Security | The security and access management of the system needs to be more transparent. Access logs were requested to monitor who had been downloading datasets. Sharing was seen as a key enabler of the system and users wanted to restrict access to certain subsets of their data. |
Performance | In order to migrate and download large datasets, high bandwidth networks would be required. Hence the system was viewed as more suitable for smaller datasets, with potentially a large number of users. |
Operational issues | System administrators were concerned that the Kindura system would represent an additional layer of mission-critical middleware between users and cloud storage. Thus the value-add features such as cost optimisation and management of replication would need to offset this cost. |