Network Bandwidth Bottlenecks via Full Gigabit Network

christophernewman

Reputable
Mar 27, 2015
7
0
4,520
I have been trying to track down network bottlenecks, can could use some ideas to check for based on the detailed specifications I have made to attempt to remedy these issues.

Network/System Test Specifications:

Network Type: LAN Only
Router: Linksys EA6500
Systems:

Dell CS-24SC 1U rackmount wth VMWare ESXi 5.1 Enterprise 24G Ram Dual 2TB Seagate ST2000NM0053, Gigabit Ethernet

Custom AMD Vishera 6300 16G Ram 250G Samsung SSD EVO, ST2000NM0053 2TB Seagate Secondary drive, On-board Gigabit Ethernet.


Cabling: All 550MHZ Cat6

Configuration:

VM: Windows Server 2012 R2 8gig Ram with Updates running only file services/FTP and BITS
Ethernet Config: FULL Duplex Gigabit Intel Adapter

Workstation: Windows 7 Ultimate / Gigabyte GA-970A-D3P
Ethernet Config: Full Duplex Gigbit

Average transfer rate for 6.6Gig file: 36Mb/s to 44Mb/s

Software:

VM: Symantec Endpoint 2010
Workstation: AVG Free 2015 / Comodo Firewall (Game Mode for testing)

Note: connections to router, and modem are also using the same Cat6 Cabling

Please let me know if there is anything that I could be missing, or wouled require to help with this issue.
 

christophernewman

Reputable
Mar 27, 2015
7
0
4,520


Apologies. I will reiterate.

ALL Drives within the Server are .3.5 2TB Seagate ST2000NM0053 drives. The secondary on the other system is a 2TB Seagate ST2000NM0053 (Same Drive same specs) The primary drive on the other system (Not the server) is running a Samsung 840 EVO 250G SSD.

 

christophernewman

Reputable
Mar 27, 2015
7
0
4,520


Thank you for the info on the tools. For the life of me I couldn't remember which ones I had used in the past.
 
Looking from your setup, actual (physical) Ethernet does not play at all. You are transferring files between virtual servers hosted on same physical server, or from a workstation (if this "AMD Vishera" is your workstatin) to hosted VM server.

So - find out what problem you have, file server bottleneck, or network.
 

christophernewman

Reputable
Mar 27, 2015
7
0
4,520


Good point. I will go that route to see where the culprit is.
 

popatim

Titan
Moderator




Excellent catch! I completely missed where he said one was a VM. Thanks.
 

Kewlx25

Distinguished


Not that they can't be, but they shouldn't be, when trying to isolate issues.

I get about 960Mb on file copies and about 980mb with iperf. Negligible difference.
 

kanewolf

Titan
Moderator


You may have better disk I/O hardware than others. I will stand by my statement that file copies aren't a reliable way to test network I/O.