Ftp/Pftp; Hpss Installation Guide July - IBM High Performance Storage System HPSS Installation Manual

directly to tape, the buffer size used by the FTP server and the virtual volume block size defined for
the Storage Class being written to will have a significant impact. If the buffer size used by the FTP
server is not a multiple of the virtual volume block size, each buffer written will result in a distinct
storage segment on the tape. This will cause additional metadata to be stored in the system and extra
synchronization processing of the tape. However, if the buffer size is a multiple of the virtual volume
block size, each write will continue to append to the same storage segment as the previous write. This
will continue until the final write for the file, which will usually end the segment, thus reducing the
amount of metadata generated and media processing.

3.11.4. FTP/PFTP

Data transfers performed using the standard FTP interface are primarily affected by the buffer size
used by the FTP Daemon. The buffer size can be configured as described in Section 13.2: FTP
Daemon Configuration of the HPSS Management Guide. It should be a multiple of the storage
segment size, if possible. Otherwise, it should be at least a multiple of the virtual volume block size.
If the buffer size is too small, the FTP Daemon will need to issue a large number of individual read or
write requests; however, if the buffer size is too large, the FTP Daemon will require a large amount
of memory, which may cause additional paging activity on the system.
The size of the FTP Daemon buffer is extremely important if the FTP clients write files directly to a
tape Storage Class, as described in Section 3.11.3: Configuration on page 113.
Parallel FTP (PFTP) uses TCP/IP to move data.
Note that the PFTP data transfer commands (e.g., pput and pget) are not influenced by the FTP
Daemon buffer size because the data flows directly between the client and Movers.
Note that PFTP clients that use the standard FTP data transfer commands (e.g., put and get) have the
same performance considerations as standard FTP clients.
Parallel transfers move the data between the Mover and the end-client processes bypassing the HPSS
FTPD. Users should be educated to use the parallel functions rather than the non-parallel functions.
NOTE: ASCII transfers are not supported by the parallel functions and the non-parallel functions will
need to be specified for ASCII transfers. ASCII transfers are NOT typically required, but the end-
customer should familiarize themselves with the particulars.
Parallel transfers should be optimized so that the Class of Service (COS), Media Stripe Widths,
Network Stripe Widths, and Parallel Block Sizes are consistent with each other. For example, using a
Network Stripe Width of 4 with a Media Width of 2 may result in poorer performance than if both
specifications are equal. Specifying a Network Stripe Width of 4 where there is only one network
interface may not provide any improvement over a lower Network Stripe Width (2) if the bandwidth
of the network is (over-)filled by a 4-way stripe.
Non-parallel transfers occur via a "stage and forward" approach (Device <==> Mover <==> HPSS
FTP Daemon <==> FTP Client.) It is recommended that the "–h" option be specified on the
hpss_pftpd and that the "hpss_option HOST hostname" be specified in the ftpaccess file if the FTP
Daemon system has multiple interfaces. The hostname should refer to the highest speed interface
available for transmission of data between the FTP Daemon and HPSS Movers. NOTE: this interface
MUST be available to all client systems that contact the FTP Daemon.
Where reasonable, the standard FTP ports should be modified to something other than 20/21 on the
system acting as the HPSS FTP Daemon. The HPSS FTP Daemon should be set to use the 20/21
ports by default. This reduces the problem of requiring the end-customer to know which port to use
for transfers to HPSS. In conjunction with this, it is highly recommended that the {ftpbanner} file be
used with an appropriate message to provide information to the end-customer that they are accessing
HPSS as opposed to a standard system.
HPSS Installation Guide July 2008
Release 6.2 (Revision 2.0) 114