[PATCH net-next] tcp: double default TSQ output bytes limit

[PATCH net-next] tcp: double default TSQ output bytes limit

[Wei Liu]

Xen virtual network driver has higher latency than a physical NIC.
Having only 128K as limit for TSQ introduced 30% regression in guest
throughput.

This patch raises the limit to 256K. This reduces the regression to 8%.
This buys us more time to work out a proper solution in the long run.

Signed-off-by: Wei Liu <wei.liu2@citrix.com>
Cc: David Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
---
 net/ipv4/tcp_output.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 190538a..eeb59be 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -50,8 +50,8 @@ int sysctl_tcp_retrans_collapse __read_mostly = 1;
  */
 int sysctl_tcp_workaround_signed_windows __read_mostly = 0;
 
-/* Default TSQ limit of two TSO segments */
-int sysctl_tcp_limit_output_bytes __read_mostly = 131072;
+/* Default TSQ limit of four TSO segments */
+int sysctl_tcp_limit_output_bytes __read_mostly = 262144;
 
 /* This limits the percentage of the congestion window which we
  * will allow a single TSO frame to consume.  Building TSO frames
-- 
1.9.1

Re: [PATCH net-next] tcp: double default TSQ output bytes limit

[Eric Dumazet]

On Wed, 2015-06-03 at 11:10 +0100, Wei Liu wrote:
> Xen virtual network driver has higher latency than a physical NIC.
> Having only 128K as limit for TSQ introduced 30% regression in guest
> throughput.
> 
> This patch raises the limit to 256K. This reduces the regression to 8%.
> This buys us more time to work out a proper solution in the long run.
> 
> Signed-off-by: Wei Liu <wei.liu2@citrix.com>
> Cc: David Miller <davem@davemloft.net>
> Cc: Eric Dumazet <eric.dumazet@gmail.com>
> ---
>  net/ipv4/tcp_output.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
> index 190538a..eeb59be 100644
> --- a/net/ipv4/tcp_output.c
> +++ b/net/ipv4/tcp_output.c
> @@ -50,8 +50,8 @@ int sysctl_tcp_retrans_collapse __read_mostly = 1;
>   */
>  int sysctl_tcp_workaround_signed_windows __read_mostly = 0;
>  
> -/* Default TSQ limit of two TSO segments */
> -int sysctl_tcp_limit_output_bytes __read_mostly = 131072;
> +/* Default TSQ limit of four TSO segments */
> +int sysctl_tcp_limit_output_bytes __read_mostly = 262144;
>  
>  /* This limits the percentage of the congestion window which we
>   * will allow a single TSO frame to consume.  Building TSO frames

Acked-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH net-next] tcp: double default TSQ output bytes limit

[David Miller]

From: Wei Liu <wei.liu2@citrix.com>
Date: Wed, 3 Jun 2015 11:10:42 +0100

> Xen virtual network driver has higher latency than a physical NIC.
> Having only 128K as limit for TSQ introduced 30% regression in guest
> throughput.
> 
> This patch raises the limit to 256K. This reduces the regression to 8%.
> This buys us more time to work out a proper solution in the long run.
> 
> Signed-off-by: Wei Liu <wei.liu2@citrix.com>

Applied, thanks.