Data archive in the cloud
To create a data archive and make it accessible from client machines, deploy a single-node file server or a distributed file system with multiple servers.
The first solution is simpler, but does not imply horizontal scaling: bandwidth and disk space will be limited. A single-node file server is suitable for information systems that do not have strict fault tolerance requirements. There may be some interruptions of operation during the restart or recovery of the virtual machine after a failure.
The second solution has several advantages over a single server configuration. You can distribute requests from users among different servers and increase or decrease the number of servers as the load changes. By placing virtual machines in different availability zones, you will be protected from failures in any one of them. To further increase the fault tolerance of a distributed archive, you can choose a file system that is designed specifically for network storage. Examples are MooseFS, LizardFS and GlusterFS.
In both cases, it is better to use virtual machines with network block storage and store the archive contents there. This configuration ensures data security even if the virtual machine fails. The server can be accessed over SMB, CIFS, NFSv3, or NFSv4.