Optimize the dd command to improve performance

Use if=/dev/zero: Instead of reading from /dev/urandom,
which is slow for generating random data, use /dev/zero
to write zeros to the disk. Zeros are generated much faster.
Omit conv=fsync: The conv=fsync option forces dd to
synchronize the data after each write, which can be slow for
large writes.

Change-Id: Id845c60c39072a03171fa5906461eb254c2736a7
diff --git a/manila_tempest_tests/config.py b/manila_tempest_tests/config.py
index 0cd07d3..4d36944 100644
--- a/manila_tempest_tests/config.py
+++ b/manila_tempest_tests/config.py
@@ -336,4 +336,17 @@
                      "attempt to create an IPv6 subnet on the project network "
                      "they create for ping and SSH to the client test VM "
                      "where data path testing is performed."),
+    cfg.StrOpt("dd_input_file",
+               default="/dev/zero",
+               help="The input file (if) in the dd command specifies the "
+                    "source of data that dd will read and process, which can "
+                    "be a device, a regular file, or even standard input "
+                    "(stdin). dd copies, transforms, or performs actions on "
+                    "this data based on provided options and then writes it "
+                    "to an output file or device (of). When using /dev/zero "
+                    "in storage systems with default compression, although "
+                    "it generates highly compressible null bytes (zeros), "
+                    "writing data from /dev/zero might not yield significant "
+                    "space savings as these systems are already optimized for "
+                    "efficient compression."),
 ]