BSD Router Project

User Tools

Site Tools

Trace:
documentation:examples:forwarding_performance_lab_of_a_netgate_rcc-ve_4860
no way to compare when less than two revisions

Differences

This shows you the differences between two versions of the page.


Next revision
—documentation:examples:forwarding_performance_lab_of_a_netgate_rcc-ve_4860 [2019/03/25 12:38] – external edit 127.0.0.1
Line 1: Line 1:
 +====== Forwarding performance lab of Netgate RCC-VE 4860 ======
 +{{description>Forwarding performance lab of a quad core Intel Atom C2558E (2.40GHz) with 2+4 Gigabit Intel NIC}}
 +===== Hardware detail =====
  
 +This lab will test a [[http://store.netgate.com/ADI/RCC-VE-4860.aspx|Netgate RCC-VE 4860]] ([[Netgate RCC-VE 4860|dmesg]]):
 +   * Quad cores Intel Atom C2558 (2.40GHz)
 +   * 2 Gigabit Intel i211
 +   * 4 Gigabit Intel i350
 +   * 8Gb of RAM
 +===== Lab set-up =====
 +
 +For more information about full setup of this lab: [[documentation:examples:Setting up a forwarding performance benchmark lab]] (switch configuration, etc.).
 +
 +BSDRP 1.92 is used (based on FreeBSD 12-stable r345325) is used.
 +
 +==== Diagram ====
 +
 +<code>
 + +------------------------------------------+    +-----------------------+
 + |            Device under Test             |    |     Packet gen        |
 + |                  igb2:   198.18.0.209/24 |<===| igb2: 198.18.0.203    |
 + |                           2001:2::209/64 |    |     2001:2::203/64    |
 + |                     (00:08:a2:09:33:da)  |    | (00:1b:21:c4:95:7a)   |
 + |                                          |    |                       |
 + |                  igb3:   198.19.0.209/24 |    | igb3: 198.19.0.203    |
 + |                    2001:2:0:8000::209/64 |    | 2001:2:0:8000::203/64 |
 + |                     (00:08:a2:09:33:db)  |===>| (00:1b:21:c4:95:7b)   |
 + |                                          |    |                       |
 + |               static routes              |    |                       |
 + |      198.19.0.0/16 => 198.19.0.203       |    +-----------------------+
 + |      198.18.0.0/16 => 198.18.0.203       |
 + | 2001:2::/49        => 2001:2::203        |
 + | 2001:2:0:8000::/49 => 2001:2:0:8000::203 |
 + |                                          |
 + |            static arp and ndp            |
 + | 198.18.0.203        => 00:1b:21:c4:95:7a |
 + | 2001:2::203                              |
 + |                                          |
 + | 198.19.0.203        => 00:1b:21:c4:95:7b |
 + | 2001:2:0:8000::203                       |
 + |                                          |
 + |                                          |
 + +------------------------------------------+
 +
 +</code>
 +
 +This device use 2 kinds of Intel NIC:
 +  * igb0 and igb1: Intel i211 with 2 queues, should be used for admin purpose
 +  * igb2 to igb5: Intel i350 with 4 queues (and iPXE support) should be used for forwarding/firewalling purpose
 +
 +
 +The generator **MUST** generate lot's of IP flows (multiple source/destination IP addresses and/or UDP src/dst port) and minimum packet size (for generating maximum packet rate) with one of these commands:
 +
 +Multiple source/destination IP addresses (don't forget to precise UDP port to use for avoiding using number 0 filtered by pf):
 +<code>
 +pkt-gen -i igb2 -f tx -n 80000000 -l 60 -d 198.19.10.1:2000-198.19.10.20 -D 00:08:a2:09:33:da -s 198.18.10.1:2000-198.18.10.100 -S 00:1b:21:c4:95:7a -w 4 -U
 +</code>
 +
 +And the same with IPv6 flows (minimum frame size of 62 for having a correct empty UDP packet):
 +<code>
 +pkt-gen -f tx -i igb2 -n 1000000000 -l 62 -6 -d "[2001:2:0:8001::1]-[2001:2:0:8001::64]" -D 00:08:a2:09:33:da -s "[2001:2:0:1::1]-[2001:2:0:1::14]" -S 00:1b:21:c4:95:7a -w 4 -U
 +</code>
 +
 +Receiver will use these commands:
 +<code>
 +pkt-gen -i igb3 -f rx -w 4
 +</code>
 +
 +===== Basic configuration and tuning =====
 +
 +==== Disabling Ethernet flow-control ===
 +
 +First, disable Ethernet flow-control:
 +<code>
 +echo "dev.igb.2.fc=0" >> /etc/sysctl.conf
 +echo "dev.igb.3.fc=0" >> /etc/sysctl.conf
 +service sysctl restart
 +</code>
 +
 +====Enabling Tx abdicate (iflib drivers) ===
 +
 +Second, if NIC drivers is iflib based you should enable TX abdicate (automatically done with BSDRP but not on a generic FreeBSD):
 +<code>
 +cat <<EOF >> /etc/sysctl.conf
 +# Enabling Tx abdicate
 +dev.igb.0.iflib.tx_abdicate=1
 +dev.igb.1.iflib.tx_abdicate=1
 +dev.igb.2.iflib.tx_abdicate=1
 +dev.igb.3.iflib.tx_abdicate=1
 +dev.igb.4.iflib.tx_abdicate=1
 +dev.igb.5.iflib.tx_abdicate=1
 +EOF
 +service sysctl restart
 +</code>
 +
 +==== Disabling ICMP redirect ===
 +
 +ICMP redirect is enabled by default on a generic FreeBSD (it's disabled by default on BSDRP), and this feature disable the fast tryforward code path.
 +
 +<code>
 +cat <<EOF >> /etc/sysctl.conf
 +# Enabling fastforwarding by disabling ICMP redirect
 +net.inet.ip.redirect=0
 +net.inet6.ip6.redirect=0
 +EOF
 +service sysctl restart
 +</code>
 +==== IP Configuration ==== 
 +
 +Configure IP addresses, static ARP entries and static routes.
 +
 +A router [[Documentation:Technical docs:Performance|should not use LRO and TSO]]. BSDRP disable by default using a RC script (disablelrotso_enable="YES" in /etc/rc.conf.misc).
 +
 +<code>
 +# IPv4 router
 +gateway_enable="YES"
 +ifconfig_igb2="inet 198.18.0.209/25 -tso4 -tso6 -lro"
 +ifconfig_igb3="inet 198.19.0.209/25 -tso4 -tso6 -lro"
 +static_routes="generator receiver"
 +route_generator="-net 198.18.0.0/16 198.18.0.203"
 +route_receiver="-net 198.19.0.0/16 198.19.0.203"
 +static_arp_pairs="receiver generator"
 +static_arp_generator="198.18.0.203 00:1b:21:c4:95:7a"
 +static_arp_receiver="198.19.0.203 00:1b:21:c4:95:7b"
 +
 +# IPv6 router
 +ipv6_gateway_enable="YES"
 +ipv6_activate_all_interfaces="YES"
 +ifconfig_igb2_ipv6="inet6 2001:2::209 prefixlen 64"
 +ifconfig_igb3_ipv6="inet6 2001:2:0:8000::209 prefixlen 64"
 +ipv6_static_routes="generator receiver"
 +ipv6_route_generator="2001:2:: -prefixlen 49 2001:2::203"
 +ipv6_route_receiver="2001:2:0:8000:: -prefixlen 49 2001:2:0:8000::203"
 +static_ndp_pairs="generator receiver"
 +static_ndp_generator="2001:2::203 00:1b:21:c4:95:7a"
 +static_ndp_receiver="2001:2:0:8000::203 00:1b:21:c4:95:7b"
 +</code>
 +
 +===== forwarding rate =====
 +
 +We start the first test by starting one packet generator at gigabit line-rate (1.488Mpps) the pkt-gen receiver output:
 +
 +<code>
 +378.931863 main_thread [2514] 1.126 Mpps (1.127 Mpkts 541.181 Mbps in 1001661 usec) 10.97 avg_batch 1014 min_space
 +379.933862 main_thread [2514] 1.124 Mpps (1.126 Mpkts 540.628 Mbps in 1002000 usec) 10.79 avg_batch 1015 min_space
 +380.934863 main_thread [2514] 1.122 Mpps (1.123 Mpkts 539.050 Mbps in 1001000 usec) 10.87 avg_batch 1015 min_space
 +381.936861 main_thread [2514] 1.121 Mpps (1.123 Mpkts 539.214 Mbps in 1001999 usec) 10.94 avg_batch 1015 min_space
 +382.938075 main_thread [2514] 1.122 Mpps (1.123 Mpkts 539.200 Mbps in 1001213 usec) 11.11 avg_batch 1015 min_space
 +383.938862 main_thread [2514] 1.121 Mpps (1.122 Mpkts 538.636 Mbps in 1000788 usec) 11.26 avg_batch 1014 min_space
 +384.940861 main_thread [2514] 1.121 Mpps (1.123 Mpkts 539.078 Mbps in 1001999 usec) 11.40 avg_batch 1015 min_space
 +385.979862 main_thread [2514] 1.121 Mpps (1.165 Mpkts 558.969 Mbps in 1039001 usec) 11.38 avg_batch 1014 min_space
 +386.980862 main_thread [2514] 1.121 Mpps (1.122 Mpkts 538.379 Mbps in 1000999 usec) 11.69 avg_batch 1015 min_space
 +387.982862 main_thread [2514] 1.121 Mpps (1.123 Mpkts 539.269 Mbps in 1002001 usec) 12.01 avg_batch 1015 min_space
 +388.983863 main_thread [2514] 1.120 Mpps (1.121 Mpkts 538.263 Mbps in 1001000 usec) 12.33 avg_batch 1016 min_space
 +389.998864 main_thread [2514] 1.121 Mpps (1.138 Mpkts 546.116 Mbps in 1015001 usec) 11.92 avg_batch 1014 min_space
 +</code>
 +
 +It's able to forward at about about 1.12 Mpps.
 +
 +===== Firewalls impact =====
 +
 +This test will generate 2000 different flows by using a mix of 20 different sources IP and 100 different destination IP.
 +
 +[[https://github.com/ocochard/netbenches/tree/master/Atom_C2558_4Cores-Intel_i350/forwarding-pf-ipfw/configs|Firewalls configurations used on github]].
 +
 +==== Graph ====
 +
 +scale information about Gigabit Ethernet:
 +  * 1.488Mpps is the maximum paquet-per-second (pps) rate with smallest 46 bytes packets.
 +  * 81Kpps is the minimum pps rate with biggest 1500 bytes packets.
 +
 +{{bench.ipfw.pf.netgate.png|forwarding and firewalling rate on a Netgate RCC-VE 4860 running FreeBSD 10.1}}
 +
 +
 +
 +===== Netmap performance =====
 +
 +==== pkt-gen  ====
 +
 +Intel NIC drivers are netmap ready… what's about the rate with the netmap's packet generator/receiver ?
 +
 +As a receiver (the sender is emitting at 1.48 Mpps):
 +<code>
 +[root@BSDRP]~# pkt-gen -i igb2 -f rx -w 4
 +186.632724 main [1705] interface is igb2
 +186.632937 extract_ip_range [289] range is 10.0.0.1:0 to 10.0.0.1:0
 +186.632963 extract_ip_range [289] range is 10.1.0.1:0 to 10.1.0.1:0
 +186.957406 main [1896] mapped 334980KB at 0x801dff000
 +Receiving from netmap:igb2: 4 queues, 1 threads and 1 cpus.
 +186.957504 main [1982] Wait 4 secs for phy reset
 +190.958543 main [1984] Ready...
 +190.958678 nm_open [456] overriding ifname igb2 ringid 0x0 flags 0x1
 +190.958863 receiver_body [1238] reading from netmap:igb2 fd 4 main_fd 3
 +191.960540 main_thread [1502] 0 pps (0 pkts in 1001704 usec)
 +192.022046 receiver_body [1245] waiting for initial packets, poll returns 0 0
 +192.962539 main_thread [1502] 0 pps (0 pkts in 1001999 usec)
 +193.040557 receiver_body [1245] waiting for initial packets, poll returns 0 0
 +193.964552 main_thread [1502] 329506 pps (330169 pkts in 1002013 usec)
 +194.966546 main_thread [1502] 1488135 pps (1491104 pkts in 1001995 usec)
 +195.968552 main_thread [1502] 1488144 pps (1491128 pkts in 1002005 usec)
 +196.970547 main_thread [1502] 1488091 pps (1491060 pkts in 1001995 usec)
 +197.972545 main_thread [1502] 1488169 pps (1491142 pkts in 1001998 usec)
 +198.974550 main_thread [1502] 1488141 pps (1491125 pkts in 1002005 usec)
 +199.976507 main_thread [1502] 1488138 pps (1491052 pkts in 1001958 usec)
 +200.978546 main_thread [1502] 1488118 pps (1491151 pkts in 1002038 usec)
 +201.980545 main_thread [1502] 1488143 pps (1491118 pkts in 1001999 usec)
 +202.982551 main_thread [1502] 1488145 pps (1491130 pkts in 1002006 usec)
 +203.984386 main_thread [1502] 1488136 pps (1490867 pkts in 1001835 usec)
 +204.985545 main_thread [1502] 1488137 pps (1489862 pkts in 1001159 usec)
 +205.987544 main_thread [1502] 1488135 pps (1491110 pkts in 1001999 usec)
 +206.989545 main_thread [1502] 1488113 pps (1491091 pkts in 1002001 usec)
 +207.991543 main_thread [1502] 1488143 pps (1491116 pkts in 1001998 usec)
 +208.993545 main_thread [1502] 1488147 pps (1491126 pkts in 1002002 usec)
 +209.994551 main_thread [1502] 1488135 pps (1489632 pkts in 1001006 usec)
 +210.995544 main_thread [1502] 1488116 pps (1489595 pkts in 1000994 usec)
 +211.997543 main_thread [1502] 1488145 pps (1491120 pkts in 1001999 usec)
 +212.998542 main_thread [1502] 1488153 pps (1489638 pkts in 1000998 usec)
 +</code>
 +
 +It support the maximum Ethernet frame rate (1.48Mpps) with netmap as a receiver.
 +
 +As a packet generator:
 +<code>
 +[root@BSDRP]~# pkt-gen -i igb2 -f tx -w 4 -c 2 -n 80000000 -l 60 -d 2.1.3.1-2.1.3.20 -D 00:1b:21:d4:3f:2a -s 1.1.3.3-1.1.3.100
 +550.392443 main [1705] interface is igb2
 +550.392739 extract_ip_range [289] range is 1.1.3.3:0 to 1.1.3.100:0
 +550.392771 extract_ip_range [289] range is 2.1.3.1:0 to 2.1.3.20:0
 +550.717236 main [1896] mapped 334980KB at 0x801dff000
 +Sending on netmap:igb2: 4 queues, 1 threads and 2 cpus.
 +1.1.3.3 -> 2.1.3.1 (00:00:00:00:00:00 -> 00:1b:21:d4:3f:2a)
 +550.717342 main [1958] --- SPECIAL OPTIONS: copy
 +
 +550.717352 main [1980] Sending 512 packets every  0.000000000 s
 +550.717361 main [1982] Wait 4 secs for phy reset
 +554.743524 main [1984] Ready...
 +554.743659 nm_open [456] overriding ifname igb2 ringid 0x0 flags 0x1
 +554.743854 sender_body [1064] start, fd 4 main_fd 3
 +555.753521 main_thread [1502] 1466573 pps (1480787 pkts in 1009692 usec)
 +556.754522 main_thread [1502] 1488145 pps (1489635 pkts in 1001001 usec)
 +557.756031 main_thread [1502] 1488145 pps (1490391 pkts in 1001509 usec)
 +558.756528 main_thread [1502] 1488112 pps (1488852 pkts in 1000497 usec)
 +559.757526 main_thread [1502] 1488130 pps (1489615 pkts in 1000998 usec)
 +560.759520 main_thread [1502] 1488142 pps (1491109 pkts in 1001994 usec)
 +561.761527 main_thread [1502] 1488130 pps (1491105 pkts in 1001999 usec)
 +562.763520 main_thread [1502] 1488135 pps (1491113 pkts in 1002001 usec)
 +563.764527 main_thread [1502] 1488119 pps (1489619 pkts in 1001008 usec)
 +564.766519 main_thread [1502] 1488164 pps (1491127 pkts in 1001991 usec)
 +565.768520 main_thread [1502] 1488116 pps (1491094 pkts in 1002001 usec)
 +566.770520 main_thread [1502] 1488151 pps (1491127 pkts in 1002000 usec)
 +567.771520 main_thread [1502] 1488112 pps (1489600 pkts in 1001000 usec)
 +568.773518 main_thread [1502] 1488140 pps (1491113 pkts in 1001998 usec)
 +^C568.885112 sigint_h [326] received control-C on thread 0x801806800
 +568.885184 sender_body [1162] flush tail 350 head 350 on thread 0x801806800
 +568.885222 sender_body [1170] pending tx tail 628 head 582 on ring 0
 +568.885280 sender_body [1170] pending tx tail 637 head 582 on ring 0
 +568.885315 sender_body [1170] pending tx tail 659 head 582 on ring 0
 +568.885393 sender_body [1170] pending tx tail 685 head 582 on ring 0
 +568.885450 sender_body [1170] pending tx tail 709 head 582 on ring 0
 +569.775515 main_thread [1502] 165737 pps (166068 pkts in 1001997 usec)
 +Sent 21022355 packets, 60 bytes each, in 14.15 seconds.
 +Speed: 1.49 Mpps Bandwidth: 713.30 Mbps (raw 998.62 Mbps)
 +</code>
 +
 +No problem in this direction too: Gigabit line-rate is a piece of cake with netmap on this platform.
documentation/examples/forwarding_performance_lab_of_a_netgate_rcc-ve_4860.txt · Last modified: 2019/12/16 20:51 by olivier
Except where otherwise noted, content on this wiki is licensed under the following license: BSD 2-Clause
Donate Powered by PHP Valid HTML5 Valid CSS Driven by DokuWiki