The archivation is very important in distributed systems. The Massiv uses a global archivation where the global means that it runs simultaneously on all of the servers and the whole simulation state is being archived - every server saves its own part of the simulation (which it is currently managing). The archivation process starts periodically on every involved server and runs transparently in the background and thus does not block server functions at all. Once the achivation is started, all servers are notified about this action and the global archivation starts. Archives are stored in the server's ./archives/regular/ directory as filenames massiv_XXXXXXXXXX.archive where XXXXXXXXX represents the number of the created archive ( e.g.0000000001 ). Because of high amount of the created files, old files are consecutively deleted and only newer ones are kept. After the archive creation, an acknowledgement file massiv_XXXXXXXXXX.archive.__ack__ is created, which serves for the archive validation. Of course, the archives have its own validation system, but this created file means that no error occured during the archivation and data were successfully saved. Created archive files are descendant of the Massiv volumes and can be unpacked or listed by auxiliary utilities (see Chapter 25, Auxiliary Utilities). Note that because of archive's extended properties, it cannot be repacked as a new valid archive by the utilities.

For description of archivation configuration in brief implementation notes, please refer to Section 27.2, “Archive Database” and Section 27.3, “Archive Manager”.