[pfSense Support] throughput tuning in 2.0
2.0-RC1 (amd64) built on Tue Mar 1 15:52:28 EST 2011 Core i3 550 3.2 GHz 4GB RAM Intel GBE I've just set this system up doing some crude throughput testing with iperf. The most I can push through this box from LAN to WAN is a steady 503-520 mbps, using the default mtu (higher mtu values produce no throughput on iperf for reasons I haven't looked into. I'm suspecting no support in the switch). top -SH is showing ~25% interrupt usage and 30%+ idle on both cores. Hyperthreading is disabled. I'm using a single NIC with vlans, but testing in only one direction, so the NIC is sending and receiving a total of about 530 mbit x2 during the test. iperf test machines show minimal CPU usage during the test, and have no other significant network activity happening concurrently. The switch is a Netgear ProSafe GS108E, which is ostensibly non-blocking. I expected better throughput than that. Any ideas what is holding this thing back, or where I could look to find out? Thanks, db - To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org
Re: [pfSense Support] throughput tuning in 2.0
I am not sure how/where you would check this but maybe the card is operating in simplex mode in which case I believe it makes sense you are getting approximately half of gigabit. Someone please correct me if I am wrong. Moshe On Tuesday, March 1, 2011, David Burgess apt@gmail.com wrote: 2.0-RC1 (amd64) built on Tue Mar 1 15:52:28 EST 2011 Core i3 550 3.2 GHz 4GB RAM Intel GBE I've just set this system up doing some crude throughput testing with iperf. The most I can push through this box from LAN to WAN is a steady 503-520 mbps, using the default mtu (higher mtu values produce no throughput on iperf for reasons I haven't looked into. I'm suspecting no support in the switch). top -SH is showing ~25% interrupt usage and 30%+ idle on both cores. Hyperthreading is disabled. I'm using a single NIC with vlans, but testing in only one direction, so the NIC is sending and receiving a total of about 530 mbit x2 during the test. iperf test machines show minimal CPU usage during the test, and have no other significant network activity happening concurrently. The switch is a Netgear ProSafe GS108E, which is ostensibly non-blocking. I expected better throughput than that. Any ideas what is holding this thing back, or where I could look to find out? Thanks, db - To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org -- -- Moshe Katz KatzNet Computers -- mo...@ymkatz.net -- kohenk...@gmail.com -- mk...@zment.com -- mmk...@umd.edu -- kohenk...@aim.com -- moshek...@verizon.net -- kohenk...@inbox.com -- kohenk...@protonic.com -- +1(301)867-3732 - To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org
Re: [pfSense Support] throughput tuning in 2.0
Op 2-3-2011 3:44, David Burgess schreef: 2.0-RC1 (amd64) built on Tue Mar 1 15:52:28 EST 2011 Core i3 550 3.2 GHz 4GB RAM Intel GBE I'm seeing atleast 600mbit of iscsi throughput through a Dell R310 with this processor, 4 port igb card and 2 bce onboard. I'm routing it from one interface to another although it's destination is also a VLAN on that other interface. Maybe that's where the issue lies. I have not performed testing from one interface to another without vlans. I am seeing roughly 200mbit sustained during the backups at night. Regards, Seth - To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org
Re: [pfSense Support] throughput tuning in 2.0
On Wed, Mar 2, 2011 at 12:38 AM, Seth Mos seth@dds.nl wrote: I'm routing it from one interface to another although it's destination is also a VLAN on that other interface. Maybe that's where the issue lies. It would be unfortunate if vlan-vlan traffic on a given interface has its maximum throughput reduced by almost half. I would be interested to see how your throughput would differ using two distinct physical interfaces, all else being equal. db - To unsubscribe, e-mail: support-unsubscr...@pfsense.com For additional commands, e-mail: support-h...@pfsense.com Commercial support available - https://portal.pfsense.org