Add aws experiment setting and commands

exp-setup/aws/bench/baseline.md → exp-setup/aws/bench/00baseline.md

@@ -1,5 +1,8 @@
 # Baseline speed
 
+
+## c5n.xlarge
+
 iperf3
 
 ```bash
@@ -133,4 +136,12 @@ Removing all streams from port(s) [0._]:                     [SUCCESS]
 Attaching 1 streams to port(s) [0._]:                        [SUCCESS]
 Starting traffic on port(s) [0._]:                           [FAILED]
 start - Port 0 : *** Expected L1 B/W: '16.8 Gbps' exceeds port line rate: '16.384 Gbps'
-```
+```
+
+
+
+======================================
+
+## c5n.18large
+
+

exp-setup/aws/bench/01drop.md

@@ -0,0 +1,98 @@
+# Packet Drop Performance
+
+
+**TRex command**
+
+```bash
+trex> start -f stl/udp_for_benchmarks.py --port 0 -d 60 -m XXmpps -t packet_len=64,stream_count=XX
+```
+
+## Linux
+
+### iptables raw
+
+Add rule
+```bash
+sudo iptables --table raw --insert PREROUTING --jump DROP --in-interface ens6
+```
+
+> Remember to specify the interface (-i) , else all interface will be matched
+> Since that also includes management interface , you might lose connection to machine
+
+List rule
+```bash
+sudo iptables --table raw --list
+```
+
+Get packet count 
+```bash
+sudo iptables --table raw --list -v
+```
+
+Reset counter 
+```bash
+sudo iptables --table raw --list -Z
+```
+
+Use mmwatch tool by Cloudflare
+
+```bash
+mmwatch 'sudo iptables -t raw -L -v -n -x | head'
+```
+
+To remove rule
+```bash
+sudo iptables -t raw -D PREROUTING -j DROP -i ens6
+```
+
+### iptables filter (with conntrack)
+
+Add rule 
+```bash
+sudo iptables -I FORWARD -d 48.0.0.1/32 -p udp --dport 12  -j DROP
+```
+
+Verify conntrack
+```
+vagrant@xdp-DUT:~$ sudo conntrack -L
+conntrack v1.4.6 (conntrack-tools): 0 flow entries have been shown.
+```
+
+Get packet count 
+```bash
+sudo iptables  --list -v
+```
+
+Reset counter 
+```bash
+sudo iptables --list -Z
+```
+
+Use mmwatch tool by Cloudflare
+
+```bash
+mmwatch 'sudo iptables -L -v -n -x | head'
+```
+
+To remove rule
+```bash
+sudo iptables -D FORWARD -d 48.0.0.1/32 -p udp --dport 12  -j DROP
+```
+
+## XDP
+
+
+```bash
+sudo ./xdp_rxq_info --dev ens6 --action XDP_DROP
+```
+
+
+
+## Repeat 
+
+**Redirect Traffic to N core**
+```bash
+sudo ethtool -L ens6 combined N
+```
+
+*N -> 1 to 6*

exp-setup/aws/bench/02cpu-drop.md

@@ -0,0 +1,40 @@
+# CPU Usage
+
+
+**TRex command**
+
+```bash
+trex> start -f stl/udp_for_benchmarks.py --port 0 -m XXkpps -t packet_len=64,stream_count=1
+```
+
+**Turn off CPU1**
+```bash
+echo 0 > /sys/devices/system/cpu/cpu1/online
+```
+> Never turn off cpu0!
+
+**Redirect Traffic to single core**
+```bash
+sudo ethtool -L ens6 combined 1
+```
+
+**mpstat command**
+```mpstat```
+
+```bash
+ mpstat -u -I SUM -P ALL 10 3
+```
+
+## Linux
+
+```bash
+sudo iptables --table raw --insert PREROUTING --jump DROP --in-interface ens6
+```
+
+## XDP
+
+
+```bash
+sudo ./xdp_rxq_info --dev ens6 --action XDP_DROP
+```
+

exp-setup/aws/bench/03forward.md

@@ -0,0 +1,22 @@
+
+# Packet Forwading Performance
+
+Similar to packet drop
+
+**TRex command**
+
+```bash
+trex> start -f stl/udp_for_benchmarks.py --port 0 -m 1mpps -t packet_len=64,stream_count=XX
+```
+
+## XDP
+
+Same NIC / and Different NIC
+
+// TODO validate the working of below command
+```bash
+sudo ./xdp_rxq_info --dev ens6 --action XDP_REDIRECT
+```
+
+You can use XDP_TX to forward packets using the same NIC by which it was received
+You can use XDP_REDIRECT to forward packets to a different network interface

exp-setup/aws/bench/04route.md

@@ -0,0 +1,46 @@
+# Software Routing
+
+**TRex command**
+
+```bash
+trex> start -f stl/udp_multi_ip.py -t packet_len=64,stream_count=20000,port_count=1 --port 0 -m 100%
+```
+
+## Linux
+
+Route everything through more specific route trick:
+
+ip r add 0.0.0.0/1   via 198.18.2.11
+ip r add 128.0.0.0/1 via 198.18.2.11
+
+
+## XDP
+
+
+```bash
+sudo ./xdp_fwd -D ens6 ens7 -S 
+```
+
+Verify
+
+Starting xdp_fwd program
+sudo ./xdp_fwd -D mlx5p1 mlx5p2 i40e1
+
+
+## Measuring 
+Measure throughput with ethtool_stats.pl
+
+Use ethtool_stats.pl
+https://github.com/netoptimizer/network-testing/blob/master/bin/ethtool_stats.pl
+
+
+## Repeat
+
+**Installing full route table**
+```
+wget https://kau.toke.dk/xdp/all-bgp-routes-20180419.txt.xz
+xzcat all-bgp-routes-20180419.txt.xz | sudo ./read_route_table.sh 198.18.100.66
+$ ip r | wc -l
+752147
+```
+

exp-setup/aws/bench/drop.md → exp-setup/aws/exp-data/01drop-data.md

@@ -1,5 +1,4 @@
 
-
 ## tRex
 
 start -f stl/udp_for_benchmarks.py --port 0 -m 24mpps -t packet_len=64,stream_count=1
@@ -46,3 +45,8 @@ rx_queue_index    0:0   582609      0
 rx_queue_index    0:sum 582609     
 ```
 
+
+
+
+===================================================
+

exp-setup/aws/exp-settings.md

@@ -0,0 +1,69 @@
+
+
+
+## tRex
+
+**Configure Interface**
+```bash
+sudo ./dpdk_setup_ports.py -i
+```
+
+```yaml
+- version: 2
+  interfaces: ['00:06.0', '00:07.0']
+  port_info:
+      - ip: 198.18.1.11
+        default_gw: 198.18.1.10
+      - ip: 198.18.2.11
+        default_gw: 198.18.2.10
+
+  platform:
+      master_thread_id: 0
+      latency_thread_id: 1
+      dual_if:
+        - socket: 0
+          threads: [2,3]
+```
+
+```
+cd /usr/lib/x86_64-linux-gnu/
+ln -s -f libc.a liblibc.a
+```
+
+Launch the TRex server in Stateless mode:
+
+```bash
+cd v3.XX
+vagrant@tRex:/opt/trex/v3.00$ sudo ./t-rex-64 -i
+```
+
+Launch the TRex console
+
+```bash
+vagrant@tRex:/opt/trex/v3.00$ ./trex-console
+```
+
+Start traffic
+```
+trex> start -f stl/udp_for_benchmarks.py --port 0 -m 1mpps -t packet_len=64,stream_count=1
+```
+
+
+
+## DUT
+
+sudo echo 1 > /proc/sys/net/ipv4/ip_forward
+sudo route add -net 16.0.0.0 netmask 255.0.0.0 gw 198.18.1.11
+sudo route add -net 48.0.0.0 netmask 255.0.0.0 gw 198.18.2.11
+
+
+### XDP in ENI issue
+
+```bash
+# set MTU with in XDP prog limit
+ip link set dev <interface> mtu 3498
+# Number of RX/TX queues must be half the available channels, make just 1 for 1 CPU. 2 for 2 CPU
+# Must be half of number of CPU fpr this to work
+ethtool -L <interface> combined 1
+```
+