Re: Data Transfer

On 06 Sep 2013 12:15, Jim Oberholtzer wrote:

<<SNIP>> A Save File has several limitations one of which is only
one object at a time ( a library being an object that contains other
objects ).

One *library* at a time, irrespective the number of objects. IIRC the actual limitation is more legitimately one LABEL per save request, wherein the label is only and implicitly defined by the save request; e.g. SAVCFG. The limitation of a Save File for SAVLIB and SAVOBJ [or the API] might better be described as: such a save request with the DEV(*SAVF) must be either a *save _of_ one library* or the save of any number of objects as a *save _from_ one library*

I have not done any real speed tests but I suspect there is not
going to be much difference between the two since the real wild cards
are numbers of disk units, type and amount of cache and which RAID
controllers your using etc. <<SNIP>>

Because each WRITE operation to a save file is validated synchronously, the FTP PUT to a SAVF is unlikely to stress the SM physical I/O as compared to what the FTP PUT to a stream file could achieve. That is, the former likely incurs significant /wait/ time due to work performed by the CPU vs disk, while the latter is unlikely to wait much on CPU-based work and instead be constrained mostly only by disk.