Here's a suggestion somewhat out of left field.  What about MTU?  Any
chance the provider changed it on you?  A machine right on the edge
would handle fragmentation somewhat more gracefully than a firewall
that might decide to drop certain inappropriately fragmented frames.
This would also cause potential slowdown in general.

One thing I didn't see (although I'm likely just missing it), is what
your transfer speeds between DMZ and LAN are.  Also, any chance for a
test, you can remove the router?  And again test LAN to DMZ and LAN to
Internet.  Based on your equipment specs I'm highly skeptical of this
being a hardware capacity issue (a number of us have outperformed your
numbers on _much_ lower end hardware - consider that a Soekris 4801
@266Mhz can easily hit 16Mbit of "normal" traffic, and iperf tests can
get it upwards of 35Mbit).  It might however be a hardware issue.
Also, there are some sysctl's available for troubleshooting the Intel
driver.

Substitute '0' for whichever interface you are trying to debug
sysctl -w dev.em.0.debug=1
sysctl -w dev.em.0.stats=1
The Intel driver will reset these sysctl to their default value on
it's own, it's a one time use type thing.  The results will be
available in dmesg and look like:
em0: Adapter hardware address = 0xc21e9a24
em0: CTRL = 0x40c00249 RCTL = 0x801a
em0: Packet buffer = Tx=16k Rx=48k
em0: Flow control watermarks high = 47104 low = 45604
em0: tx_int_delay = 66, tx_abs_int_delay = 66
em0: rx_int_delay = 0, rx_abs_int_delay = 66
em0: fifo workaround = 0, fifo_reset_count = 0
em0: hw tdh = 41, hw tdt = 41
em0: hw rdh = 102, hw rdt = 101
em0: Num Tx descriptors avail = 256
em0: Tx Descriptors not avail1 = 0
em0: Tx Descriptors not avail2 = 0
em0: Std mbuf failed = 0
em0: Std mbuf cluster failed = 0
em0: Driver dropped packets = 0
em0: Driver tx dma failure in encap = 0
em0: Excessive collisions = 0
em0: Sequence errors = 0
em0: Defer count = 0
em0: Missed Packets = 0
em0: Receive No Buffers = 0
em0: Receive Length Errors = 0
em0: Receive errors = 0
em0: Crc errors = 0
em0: Alignment errors = 0
em0: Collision/Carrier extension errors = 0
em0: RX overruns = 251
em0: watchdog timeouts = 0
em0: XON Rcvd = 0
em0: XON Xmtd = 0
em0: XOFF Rcvd = 0
em0: XOFF Xmtd = 0
em0: Good Packets Rcvd = 3269510
em0: Good Packets Xmtd = 647392
em0: TSO Contexts Xmtd = 0
em0: TSO Contexts Failed = 0

Lastly...if in interrupt mode still (I recommend it vs polling mode, I
don't think we've done the appropriate tuning for polling to give a
benefit), check net.inet.ip.intr_queue_drops <--- that should be 0, if
it's not, something really wierd is happening on your box.

--Bill

On Thu, Jul 31, 2008 at 8:06 AM, Ted Crow <[EMAIL PROTECTED]> wrote:
>
> I don't consider myself a Cisco expert either, I've just been using
> their hardware for the better part of 15 years.  I have access to a fair
> number of good Cisco resources to aid me in selecting and configuring
> the hardware.  I've never liked Cisco firewalls though, go figure.
>
> I actually sized the router based on an estimated max traffic flow of
> 25Mbps.  It does have a very small ACL set running on it, mainly to keep
> weird stuff from molesting my DMZ hosts (spoofing, etc.)  From the DMZ,
> the speeds are pretty respectable considering the router was only
> designed to handle a max of 46Mbps.  This one is the baby of the 2800
> series and will probably be fine when the speed is dropped back down
> below 25Mbps.
>
> Ted Crow
> Information Technology Manager
> Tuttle Services, Inc.
>
> -----Original Message-----
> From: Paul Mansfield [mailto:[EMAIL PROTECTED]
> Sent: Thursday, July 31, 2008 5:56 AM
> To: [email protected]
> Subject: Re: [pfSense Support] pfSense 1.2-RELEASE: Performance Issue?
>
> It's not clear exactly what the cisco 2801 is doing... does it have
> access control lists which can make a big difference in speed... AIUI
> access lists can have two different execution paths and if you write
> them wrongly they're much more CPU intensive. Sorry, I am not a cisco
> expert in this instance.
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to