On the NFS side, we performed these tests: 64-Kbps linear read test; 64-Kbps linear write test; 1-MB linear read test; 1-MB linear write test; 2-MB linear read test; 2-MB linear write test; NFS IOps and NWC NFS Generalized Custom Test. We performed the NFS linear read and write tests with the same parameters as those of the comparable CIFS tests (100 percent read or 100 percent write, and 100 percent sequential). However, on the NFS IOPS test, we used an 8-KB transfer request size, rather than the 512-byte size we used for the CIFS test.
Furthermore, our NWC NFS Generalized Custom Test featured transfer request sizes of 8 KB with 33 percent access distribution, 8 KB with 34 percent access distribution and 64 KB with 33 percent access distribution. On the first 8-KB segment, we set the percent read/write distribution to 100 percent read and the percent random/ sequential distribution to 100 percent sequential. On the second 8-KB segment, we set the percent read/write distribution to 67 percent read/33 percent write, and the percent random/sequential to 100 percent random. On the 64-KB segment, the percent read/write distribution was 100 percent read and the percent random/sequential distribution was 100 percent read.
Sometimes, our tests don't mix with the operating systems presented.
In this review, two vendors, Coastline Micro and Dell, sent us NAS devices that
ran the new Windows for Storage 2003. When we ran our testing tool,
SourceForge's open-source Iometer, both devices returned CIFS and NFS results that
were far worse than expected. In addition, Iometer occasionally became unstable
while testing these two boxes.
With Dell's hardware configuration, for instance, there was no reason
the test results should be so poor or Iometer so flaky. We conducted some basic
tests in which we ran file copying and FTP to Windows for Storage 2003 servers,
which produced results more like we anticipated. We contacted Dell, which, in
turn, contacted Microsoft. Iometer creates a single file to test against, and
that file can be as big as the remaining space on the target hard disk. We
suspect a file-locking issue specific to Iometer and Windows for Storage 2003.
Dell and Microsoft are still investigating.
We're certain the problems we experienced are solely with Iometer,
and we don't expect Dell, Coastline Micro or any other vendor using Windows for
Storage 2003 to experience problems. This is simply one of those test-tool
compatibility issues that plagues every test shop from time to time. To be fair, we
pulled Dell and Coastline Micro from the review. When we have a reliable
method of testing them, we plan on revisiting both products, but we have no
specific time frame at this time.
Bob Friday, Chief AI Officer for Juniper Networks, explains how the advanced technology is transforming operations.
Contact center leaders from 8x8, Awaken Intelligence, and 360insight discuss the importance of agent experience.
AI may force enterprises to rewire parts of their data centers so they are fully optimized to run such workloads. The question is do you use Ethernet or InfiniBand?
