Version 17 (modified by, 9 years ago) (diff)


IG-MON-4: Infrastructure Device Performance Test

This page captures status for the test case IG-MON-4, which verifies that the rack head node performs well enough to run all the services it needs to run while OpenFlow and non-OpenFlow experiments are running. For overall status see the InstaGENI Acceptance Test Status page.

Last update: 2013/01/29

Test Status

This section captures the status for each step in the acceptance test plan.

StepState Ticket Notes
Step 1
Step 2
Step 3

State Legend Description
Color(green,Pass)? Test completed and met all criteria
Color(#98FB98,Pass: most criteria)? Test completed and met most criteria. Exceptions documented
Color(red,Fail)? Test completed and failed to meet criteria.
Color(yellow,Complete)? Test completed but will require re-execution due to expected changes
Color(orange,Blocked)? Blocked by ticketed issue(s).
Color(#63B8FF,In Progress)? Currently under test.

Test Plan Steps

This test cases sets up several experiments to generate resource usage for both compute and network resources in an InstaGENI Rack. The rack used is the GPO rack and the following experiments are set up before head node device performance is reviewed:

  • IG-MON-4-exp1: IG GPO non-OpenFlow experiment with 10 VM nodes, all nodes exchanging traffic.
  • IG-MON-4-exp2: IG GPO non-OpenFlow experiment with 1 VM and one dedicated raw-pc, both exchanging traffic.
  • IG-MON-4-exp3: IG GPO OpenFlow experiment with 2 nodes in rack exchanging traffic with 2 site GPO OpenFlow campus resources.
  • IG-MON-4-exp4: IG GPO OpenFlow experiment with 4 nodes exchanging traffic within the GPO rack.

The setup of the experiments above is not captured in this test case, but the RSpec are available [insert_link_here]. Also traffic levels and types will be captured when this test is run.

In addition to the 4 experiment listed above the following experiments were running on the GPO IG rack: [Insert_capture for]

While the 4 experiments involving OpenFlow and compute slivers are running, execute the steps below.

1. View OpenFlow control monitoring at GMOC

Before starting any experiments that request compute or network resources, collected information for baseline performance for the GPO rack for the boss, FlowVisor and FOAM nodes.

Checked existing experiments within the rack, which also shows node allocation by node type:

Top statistics for boss node:

last pid:  9133;  load averages:  0.02,  0.03,  0.00   up 51+16:06:27  09:09:23
140 processes: 1 running, 138 sleeping, 1 zombie
CPU:  0.0% user,  0.0% nice,  0.5% system,  0.0% interrupt, 99.5% idle
Mem: 369M Active, 1313M Inact, 187M Wired, 8632K Cache, 93M Buf, 123M Free
Swap: 2047M Total, 2628K Used, 2045M Free

The above shows the boss node CPU Load averages:

  • 1 minutes: 0.02
  • 5 minutes: 0.03
  • 15 minutes: 0.00

Top statistics for FOAM node:

top - 09:16:02 up 49 days, 21:19,  2 users,  load average: 0.17, 0.12, 0.08
Tasks:  71 total,   1 running,  68 sleeping,   0 stopped,   2 zombie
Cpu(s): 17.6%us,  6.0%sy,  0.0%ni, 76.4%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:    756268k total,   613788k used,   142480k free,   149172k buffers
Swap:   794620k total,     4444k used,   790176k free,   334824k cached

The above shows the boss node CPU Load averages:

  • 1 minutes: 0.17
  • 5 minutes: 0.12
  • 15 minutes: 0.08

Checked for existing FOAM sliver (also shown in the EID column of the "Active Nodes" above:

lnevers@foam:~$ foamctl geni:list-slivers --passwd-file=/etc/foam.passwd |grep sliver_urn
   "sliver_urn": "", 
   "sliver_urn": "", 
   "sliver_urn": "", 
   "sliver_urn": "", 

Top statistics for FlowVisor node:

top - 09:16:32 up 49 days, 16:24,  2 users,  load average: 0.00, 0.01, 0.05
Tasks:  68 total,   1 running,  67 sleeping,   0 stopped,   0 zombie
Cpu(s):  0.0%us,  0.2%sy,  0.0%ni, 99.8%id,  0.0%wa,  0.0%hi,  0.0%si,  0.0%st
Mem:   4031252k total,   964924k used,  3066328k free,    92380k buffers
Swap:  4202492k total,        0k used,  4202492k free,   364268k cached

The above shows the boss node CPU Load averages:

  • 1 minutes: 0.00
  • 5 minutes: 0.01
  • 15 minutes: 0.05

Checked for existing FlowSpace rules:

lnevers@flowvisor:~$ fvctl --passwd-file=/etc/flowvisor.passwd listFlowSpace
Got reply:
rule 0: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x800,nw_dst=,nw_src=]],actionsList=[Slice:8a0abd6f-0f5a-469f-91d2-c7f990b8494e=4],id=[13],priority=[2000],]
rule 1: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x806,nw_dst=,nw_src=]],actionsList=[Slice:8a0abd6f-0f5a-469f-91d2-c7f990b8494e=4],id=[14],priority=[2000],]
rule 2: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x800,nw_dst=,nw_src=]],actionsList=[Slice:a92990b6-1ede-4dd7-b6f6-7b4a4bd36fd7=4],id=[15],priority=[2000],]
rule 3: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x806,nw_dst=,nw_src=]],actionsList=[Slice:a92990b6-1ede-4dd7-b6f6-7b4a4bd36fd7=4],id=[16],priority=[2000],]
rule 4: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x800,nw_dst=,nw_src=]],actionsList=[Slice:b7850c93-110f-4e63-a121-26f3449dac44=4],id=[21],priority=[2000],]
rule 5: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x806,nw_dst=,nw_src=]],actionsList=[Slice:b7850c93-110f-4e63-a121-26f3449dac44=4],id=[22],priority=[2000],]
rule 6: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x800,nw_dst=,nw_src=]],actionsList=[Slice:fd82bb82-eef3-407d-b092-c1393773791c=4],id=[50],priority=[2000],]
rule 7: FlowEntry[dpid=[06:d6:84:34:97:c6:c9:00],ruleMatch=[OFMatch[dl_type=0x806,nw_dst=,nw_src=]],actionsList=[Slice:fd82bb82-eef3-407d-b092-c1393773791c=4],id=[51],priority=[2000],]

2. View VLAN 1750 data plane monitoring

Verify the VLAN 1750 data plane monitoring which pings the rack's interface on VLAN 1750, and verify that packets are not being dropped

3. Verify CPU idle percentage on the server host and the FOAM VM are both nonzero

Additional Scenarios

This section captures some scenarios that are not in the original test plan, but have been captured as a data point for performance.

Attachments (1)

Download all attachments as: .zip