Re: [Iscsitarget-devel] Read Performance Tests

[Ming Z. <mi...@el...>]

HI, how can it achieve 300MB/s? cache effect at initiator side?

ming

On Thu, 2004-06-24 at 10:18, FUJITA Tomonoro wrote:
> Hello again,
> 
> Here is a read intensive benchmark, Network Server benchmark, designed
> to produce the workloads seen by HTTP or FTP servers. The benchmark
> can read many files in random order. I ran the benchmark with 512, 2
> MB files. All the files were placed in one directory.
> 
> The goal of the benchmark was to evaluate how much the page cache
> effected the performance. Therefore, I started the benchmark with both
> the initiator and target in a hot cache.
> 
> o Target
> CPU : Xeon 2.8GHz
> Mem : 2GB
> Disk : Maxtor Atlas 10K SCSI (10000 RPM) / LSI Logic 53C103
> NIC : Intel Pro/1000 MT
> Kernel : 2.4.25
> 
> o Initiator
> CPU : Xeon 2.0GHz
> Mem : 1GB
> NIC : Intel Pro/1000 MT
> Kernel : 2.6.4
> 
> - The initiator box runs the Cisco iSCSI driver v4.0.1.1.
> 
> - The initiator box uses an ext2 file system with a 4 KB block size on
> a logical unit that targets provide.
> 
> I did this test a few months ago by using a modified Ardis target,
> which became our target version 0.0.1. So you'll see slight difference
> results if you do the test with the current version.
> 
> Bye,
-- 
 --------------------------------------------------
| Ming Zhang, PhD. Student
| Dept. of Electrical & Computer Engineering
| College of Engineering
| University of Rhode Island
| Kingston RI. 02881
| e-mail: mingz at ele.uri.edu
| Tel. (401) 874-2293 
| Fax. (401) 782-6422
| http://www.ele.uri.edu/~mingz/
 --------------------------------------------------