Since MPE/iX 5.0, IMAGE/SQL has provided "jumbo datasets" to allow you to create datasets larger than the MPE 4 gigabyte file limit. It does so by splitting the (logical) dataset into several physical files, stored in the POSIX namespace. So SET05.DATA.SALES would be split into /SALES/DATA/SET05.001, /SALES/DATA/SET05.002, etc.
Suprtool attains its impressive serial read speed by reading datasets directly, with MR;NOBUF access. It establishes the physical filename(s) of the dataset, performs the serial MR;NOBUF reads, and deblocks the records to write them to the output file. Suprtool has supported jumbo datasets since their first release, and will read the "chunks" of the datasets automatically and transparently.
Netbase, from Quest Software, shadows files across CPUs. It allows programs to access remote files and databases as though they were on the local CPU. For IMAGE/SQL, it traps the intrinsic calls to the databases, and satisfies the requests with data from the remote database, and similarly processes regular file system calls.
Netbase supports the shadowing of these Jumbo datasets, but does not yet support NFA(Network File Access) directly to the POSIX namespace portion of these datasets. This functionality is currently under development at Quest. Accessing these Jumbo datasets with image intrinsics is supported via NFA.
What does this mean to you? If you're using NFA to access a remote database, and you convert a dataset in that database to jumbo format, Suprtool will be unable to perform MR;NOBUF reads to the "chunks" of the dataset as Netbase cannot trap the calls to the POSIX files. Suprtool will return like:
>GET MANIFEST-DETAILS NONEXISTENT PERMANENT FILE (FSERR 52) Error: Unable to open MPE file /ACCOUNT/DB/STORE36.001 Dev=DISC(Note the Posix file notation and extension, denoting the first "chunk" of a Jumbo Dataset.)
Workarounds?