Fwd: Netmap for routers (em0 device)

From: Archy Cho <archycho_at_gmail.com>
Date: Sat, 5 Nov 2011 03:50:21 +0800
---------- Forwarded message ----------
From: Archy Cho <archycho_at_gmail.com>
Date: 2011/11/5
Subject: Re: Netmap for routers (em0 device)
To: Darren Pilgrim <darren_at_bluerosetech.com>




2011/11/5 Darren Pilgrim <darren_at_bluerosetech.com>

> On 2011-11-04 09:06, Archy Cho wrote:
>
>> Hello
>>
>> I am very happy to see freebsd could have such network performance
>> with netmap , since I am currently using freebsd as core routers
>> instead of cisco.
>>
>
> I'm looking at doing the same at a site, but we're not sure about the
> performance as it will sit on 10 GE.  Would you tell me about:
>
> - the hardware you use
> - previous hardware that didn't work
> - functions performed (firewall, VPN, BGP feed, etc.)
> - the wire speeds involved
> - the bit and packet rates achieved
> - the CPU load
>
> Thanks in advance
>
I am currently using FreeBSD 7.4-AMD64
seldom under large pps of DDoS the router box ( freebsd ) will drop packets
,
one of the CPU core getting 100% of swi1:net with command `top -HSP`

The following configuration , could only forward around 600Kpps

Intel 3420GVP mainboard
Intel Xeon X3480 CPU with HT disabled
8GB DDR3-1333 Ram
450GB 10000rpm WDHLHX
Intel 82578DM on board as EM0
Intel 82574L on board as EM1

>100 ipfw rules
1Gbps uplink with upstream
Quagga as BGP router with 2 upstream providers full routes

kern.clockrate: { hz = 1000, tick = 1000, profhz = 2000, stathz = 133 }
kern.dcons.poll_hz: 100
kern.hz: 1000
debug.psm.hz: 20
kern.polling.idlepoll_sleeping: 1
kern.polling.stalled: 114
kern.polling.suspect: 5775
kern.polling.phase: 0
kern.polling.enable: 0
kern.polling.handlers: 2
kern.polling.residual_burst: 0
kern.polling.pending_polls: 1
kern.polling.lost_polls: 871230
kern.polling.short_ticks: 1636
kern.polling.reg_frac: 20
kern.polling.user_frac: 10
kern.polling.idle_poll: 0
kern.polling.each_burst: 1000
kern.polling.burst_max: 1000
kern.polling.burst: 1000
dev.em.0.rx_processing_limit=10000000
dev.em.1.rx_processing_limit=10000000
net.inet.ip.forwarding=1
net.inet.ip.fastforwarding=1
net.inet.icmp.icmplim=10485760
net.inet.ip.dummynet.pipe_byte_limit=104857600
net.inet.ip.dummynet.pipe_slot_limit=104857600
net.inet.ip.dummynet.hash_size=65535
net.inet.ip.fw.tables_max=65535
net.inet.ip.fw.dyn_keepalive=1
net.inet.ip.fw.dyn_max=65535
net.inet.ip.fw.dyn_short_lifetime=5
net.inet.ip.fw.dyn_buckets=4096
net.local.stream.recvspace=524288
net.local.stream.sendspace=524288
net.local.dgram.recvspace=614400
net.inet.tcp.sendspace=614400
net.inet.tcp.recvspace=614400
net.inet.udp.recvspace=420800
net.inet.sctp.recvspace=2330160
net.inet.sctp.sendspace=2330160
net.inet.raw.recvspace=614400
net.raw.recvspace=524288
net.raw.sendspace=524288
Received on Fri Nov 04 2011 - 18:50:22 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:20 UTC