One of the standard network class loading mechanisms in Java is for the client to request files individually relative to a directory URL, typically using the HTTP protocol.
Downloading individual files has the advantage that only those files explicitly requested by the client are transferred over the network. This ensures that no unneeded files are transferred, and that the files arrive in the order in which they are needed. HTTP connections are usually persistent, meaning that a single connection is used to transfer many files, and that the connection persists between client requests. In principle, it is also possible for the HTTP client and server to negotiate the use of compression; however, such compression is per-file and does not share redundant information between files. (In any case, the JDK 1.2.2 implementation of HTTP does not support such compression.)
Downloading individual files has the disadvantage that files are not pre-sent or prefetched, meaning that the network latency cost is paid for each file requested by the client. Given that network latency can be hundreds or even thousands of milliseconds, this disadvantage can be considerable.