Subproject 6 - Distributed backup system

General Informations

Status

Version

Leader

Assignees

Assigned0.2Guido Angelo IngenitoNo assigners list is defined for this project

Job Opportunities

Take a look at JDBS documentation. If you think you can help the subproject's leader send an e-mail.

Abstract Requirements

The subproject prime aim is to realize a distributed component for the data storage, that is integrate in InernetCafe and that anyway can be used like an external component to the applicative. For the law the data treated by InternetCafe must be persisting and available for many time but unfortunately at this moment we don't dispose of a detailed specification of the time lapse in which those data must be persistent.

The main idea is to have a 0..N replicated server which attend to the data persistence ( applicative backup ). The persistence could be achieve also by only one server configured in RAID. A possible net configuration could be that presented follow.

It'll be a designer team job, to locate the protocols and also the basic components for the realization of this applicative requirement and the individuation at an international level of the specifics concerning to the storage. Of great importance is the data security which must be available only for responsible staff provided of valid access requirements.

A not exhaustive estimate of the data treated by the subsystem is list below:

  • DB applicative, imagines and "core" data.
  • Sessions ( including associated connections' logs).
  • Data relative to the applicative configuration (Network architecture, Services, etc...).

Concerning to the data relative to the sessions log, we know that for a entire network with internet access only and configured for 200 machine, with a strong use we'll have logs in order of 100 Mb daily (DEIS).

The data filing must be programmed in precise time periods, we are considering storage strategies during the time in which the net is unloading ( night activity ) The backup strategy must be so flexible so it can be delineate in the place by the application user. The network definition of machine referred to carry out a storage job will be also entirely configurable. Is to be considered anyway that in the data distributed salvage and in the finding of the same, might verify the typical problems due to the presence of a net activity ( we think to a connection drop, discrepancies among versions, file corruption, etc...).

To do

Notes

From the specifications listed in the above section there is a clear intention to create a subsystem that can be integrated in the InternetCafe's context but probably should be used also as an external stand-alone application. According to this fact the application that will satisfy this role will take the name of JDBS (Java Distributed Backup System), a complete reference to its documentation can be found here.