To be translated and elaborated
The correct can either be in connection with
- lost data
- damaged data (action: replace damaged data)
For lost data the process must put the missing data to pillars that are missing the data. For damaged data the process must replace damaged data with correct data. After ended repair the integrity information must be updated, and possibly integrity checked. This is illustrated below.
Drawing to be made and inserted
For loss of small data this is handled by put as illustrated in the figure. However for mass loss, we may have to open possibility for two pillars to communicate directly - though coordinated by client as for replacement.
There may also need to be cases where we need to physically move e.g. a disk from one pillar to another in order to make the data transfer of large chunks of data. It can be treated as the scenario for pillar to pillar communication, where moved data can be identified as the pillar it was transported from. however the full solution has to be both organisational and technical.
To be further elaborated and analysed - are there specific requirements to message formats here?
The actual replace operations is described in more detail under Replace File.