WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

[Xen-devel] Xen network performance issue

To: xen-devel@xxxxxxxxxxxxxxxxxxx
Subject: [Xen-devel] Xen network performance issue
From: tangming <tangming__06@xxxxxxx>
Date: Sun, 30 May 2010 02:51:02 +0800 (CST)
Delivery-date: Wed, 02 Jun 2010 11:07:42 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
List-help: <mailto:xen-devel-request@lists.xensource.com?subject=help>
List-id: Xen developer discussion <xen-devel.lists.xensource.com>
List-post: <mailto:xen-devel@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-devel>, <mailto:xen-devel-request@lists.xensource.com?subject=unsubscribe>
Sender: xen-devel-bounces@xxxxxxxxxxxxxxxxxxx
Hi, Guys
I was obsessed by the following issue for weeks :-), and could you please help me get rid of it? Many thanks!!! 
Recentlly I used the standard Suse10sp2 as HVM guest(domU), and installed the para-virtualization network card driver(xen-vnif.ko) within it. But when I test the performance on domU by continuously sending/receiving large amount of packages I found that the writing(sending) speed can up to 100MB/s, while the reading(receiving) speed only 50MB/s, not sure why.
I know that the para-virtualization network driver used two I/O ring buffers, one for transmitting and the other for receiving, and I think the rx_buffer may be the issue, but not sure.
Two more questions:
1. Does the front_end driver use polling or event notification for discovering the data arrival?
2. Does the back_end driver use transferring or copying for putting the data to the grant pages of front_end?
Do you guys think the above methods used by the network driver is the root cause? And how can we decide to use  which method (polling vs. notification,  transferring vs. copying) in the specified situation for high throughput?
Could you please give me some suggestions for locating this issue and resolving it? Thanks in advance!!!



网易为中小企业免费提供企业邮箱(自主域名)
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel
<Prev in Thread] Current Thread [Next in Thread>
  • [Xen-devel] Xen network performance issue, tangming <=