LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Linux Director Reliability

To: "LVS Users" <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Linux Director Reliability
From: "Nicolas Chiappero" <Nicolas.Chiappero@xxxxxxxxx>
Date: Thu, 25 Sep 2003 17:19:14 +0200
> 
> If anyone is able to volunteer this information either
> on the list or to me privately then I would be most grateful.
> Please include some rough information on your system:
> What architecture, which kernel, which version of LVS,
> approximately how many nodes, and how busy it is.
> 

Hello Simon,

We're currently running two sites using LVS.

The first one is using LVS for 3 years, handling only HTTP traffic :
    - 1 Director, LVS-DR setup.
    PIII 550 Mhz, 256 Mo RAM, 1 Compaq fast ethernet NIC.
    It runs RH 7.3/linux 2.4.19 and ipvs 1.0.6.
    3 HTTP VIP and 1 HTTPS VIP.
    - 8 linux realservers.
    It currently handles 700 connexions/s and 4000 packet/s
    (ipvsadm -L -n --rate output).

The second one is using LVS for 1 year, handling only HTTP traffic :
    - 2 Directors (keepalived failover), LVS-NAT setup + firewall:
    PIII 1Ghz, 512 Mo RAM, 2 Intel fast ethernet NICS.
    They run RH 7.3/linux 2.4.18 and ipvs 1.0.4.
    4 HTTP VIP and 2 HTTPS VIP.
    - 10 linux realservers.
    It currently handles 1400 connexions/s and 8000 packets/s
    (ipvsadm -L -n --rate output).
    CPU usage is very low and is mainly used by keepalived to
    monitor realservers.

These setup are a little bit specific as they handle mostly
short length packets :
      0 < 28 % <=   64 bytes.
     65 <  6 % <=  127 bytes.
    128 <  6 % <=  255 bytes.
    256 < 32 % <=  511 bytes.
    512 < 18 % <= 1023 bytes.
   1024 < 10 % <= 1518 bytes.
    
Nothing related to LVS has ever caused a failure on both sites:
It rocks !

Regards,
Nicolas.

<Prev in Thread] Current Thread [Next in Thread>