LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: [LVS - NAT] alternatives

To: <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: [LVS - NAT] alternatives
From: "Nicolas Chiappero" <Nicolas.Chiappero@xxxxxxxxx>
Date: Tue, 11 Sep 2001 17:01:25 +0200
Thanks for all replies which have at least make
me read a lot of papers and HOWTO's to go deeper
than my first look.

I come back with 2 question to help me chose/tune
VS-DR or VS-NAT setup:

1. From what I've read and understood, I still
can't see how using LV-DR I can setup what
I want to do. Let's rember it quickly :

Suppose I have 1 Director and 1 Realserver.
My Realserver has 2 IP aliases RIP1 and RIP2.
What I want to do is to load balance my VIP
service between RIP1-1 and RIP1-2 giving an
output of ipvsadm looking like that:

IP Virtual Server version 1.0.8 (size=65536)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port             Forward Weight ActiveConn InActConn
TCP VIP:www            rr
  -> RIP1-1:www                     XXXX    1      0          0
  -> RIP1-2:www                     XXXX    1      0          0

So logically there are two Realservers, but physically
only one with 2 IP aliases.

I concluded that VS-NAT was the only suitable setup.
Am I wrong ?

2. Playing with VS-NAT, I ran some tests using testlvs
and I was wondering if I could find others results to
make some performance comparisons...

* ipvsadm setup *

IP Virtual Server version 1.0.8 (size=65536)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port             Forward Weight ActiveConn InActConn
TCP VIP:www            rr
  -> RIP1-1:www                     Masq    1      0          0
  -> RIP1-2:www                     Masq    1      0          0
  -> RIP2-1:www                     Masq    1      0          0
  -> RIP2-2:www                     Masq    1      0          0
  -> RIP3-1:www                     Masq    1      0          0
  -> RIP3-2:www                     Masq    1      0          0

* testlvs command line *

./testlvs VIP:80 -tcp -srcnum 100000 -packets 5000000

Here is what I observed :

* Director *
'vmstat 1' before running test :
r  b  w   swpd   free   buff  cache  si  so    bi    bo    in    cs  us  sy
id
0  0  0      0 159936  18656  16260   0   0     0     0   110    13   0   1
99

'vmstat 1' while running test :
r  b  w   swpd   free   buff  cache  si  so    bi    bo    in    cs  us  sy
id
0  0  0      0 159936  18656  16260   0   0     0     0 19204    13   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19167    10   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19187    12   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19145    12   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19240    10   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19104    34   0   3
97
0  0  0      0 159936  18656  16260   0   0     0     0 19346    19   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19194    10   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19182    12   0   1
99
0  0  0      0 159936  18656  16260   0   0     0     0 19099    10   0   1
99

'ipvsadm' while running test :

IP Virtual Server version 1.0.8 (size=65536)
Prot LocalAddress:Port Scheduler Flags
  -> RemoteAddress:Port             Forward Weight ActiveConn InActConn
TCP VIP:www            rr
  -> RIP1-1:www                     Masq    1      0          16635
  -> RIP1-2:www                     Masq    1      0          16637
  -> RIP2-1:www                     Masq    1      0          16632
  -> RIP2-2:www                     Masq    1      0          16629
  -> RIP3-1:www                     Masq    1      0          16637
  -> RIP3-2:www                     Masq    1      0          16635

* RealServer *
./show_traffic.sh eth0 : 3150 packets/s

All numbers mentionned above have been catched just after they were quite
stable.

Regards,
Nico.




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