[j-nsp] SRX650 Cluster throughput

Jeff Rooney jtrooney at nexdlevel.com
Tue Apr 24 13:32:25 EDT 2012


So running multiple flows seems to give me the same results....4 tcp
based flows gives me an aggregate of ~317Mbits, udp is the same...4
flows aggregate ~500Mbit.

The two reth interfaces in use here:
root at core-rtr> show interfaces reth9 statistics
Physical interface: reth9, Enabled, Physical link is Up
  Interface index: 137, SNMP ifIndex: 596
  Link-level type: Ethernet, MTU: 1514, Speed: 1Gbps, BPDU Error: None,
  MAC-REWRITE Error: None, Loopback: Disabled, Source filtering: Disabled,
  Flow control: Disabled, Minimum links needed: 1, Minimum bandwidth needed: 0
  Device flags   : Present Running
  Interface flags: SNMP-Traps Internal: 0x0
  Current address: 00:10:db:ff:10:09, Hardware address: 00:10:db:ff:10:09
  Last flapped   : 2012-04-24 00:21:10 UTC (1d 01:09 ago)
  Statistics last cleared: Never
  Input rate     : 0 bps (0 pps)
  Output rate    : 0 bps (0 pps)
  Input errors: 0, Output errors: 0

  Logical interface reth9.0 (Index 100) (SNMP ifIndex 583)
    Description: Backup VLAN
    Flags: SNMP-Traps 0x0 Encapsulation: ENET2
    Statistics        Packets        pps         Bytes          bps
    Bundle:
        Input :      40675689          0   60533484597            0
        Output:       8384721          0     438616490            0
    Security: Zone: backup
    Allowed host-inbound traffic : ping
    Protocol inet, MTU: 1500
      Flags: Sendbcast-pkt-to-re
      Addresses, Flags: Is-Preferred Is-Primary
        Destination: 10.2.100/24, Local: 10.2.100.1, Broadcast: 10.2.100.255

{primary:node0}
root at core-rtr> show interfaces ge-2/0/7 statistics
Physical interface: ge-2/0/7, Enabled, Physical link is Up
  Interface index: 283, SNMP ifIndex: 531
  Link-level type: Ethernet, MTU: 1514, Link-mode: Full-duplex, Speed: 1000mbps,
  BPDU Error: None, MAC-REWRITE Error: None, Loopback: Disabled,
  Source filtering: Disabled, Flow control: Enabled, Auto-negotiation: Enabled,
  Remote fault: Online
  Device flags   : Present Running
  Interface flags: SNMP-Traps Internal: 0x0
  Link flags     : None
  CoS queues     : 8 supported, 8 maximum usable queues
  Current address: 00:10:db:ff:10:07, Hardware address: 64:87:88:1c:f0:3f
  Last flapped   : 2012-04-24 03:17:27 UTC (22:15:42 ago)
  Statistics last cleared: Never
  Input rate     : 0 bps (0 pps)
  Output rate    : 0 bps (0 pps)
  Input errors: 0, Output errors: 0
  Active alarms  : None
  Active defects : None
  Interface transmit statistics: Disabled

  Logical interface ge-2/0/7.0 (Index 113) (SNMP ifIndex 638)
    Flags: SNMP-Traps Encapsulation: ENET2
    Input packets : 7167489
    Output packets: 41515426
    Security: Zone: Null
    Protocol aenet, AE bundle: reth7.0   Link Index: 0



On Tue, Apr 24, 2012 at 12:04 PM, Stefan Fouant
<sfouant at shortestpathfirst.net> wrote:
> On 4/24/2012 1:01 PM, Jeff Rooney wrote:
>>
>> Single flow using iperf for all of my tests, only variation is that
>> I'm using both udp and tcp tests, and via the cluster udp seems to
>> handle better, but still only seeing 500Mbit
>
>
> Just for giggles why don't you increase your test to utilize multiple flows
> and lets see if you get different performance (this is going to be the more
> likely scenario anyways since it's unlikely you'll have a single host
> sending 900+ Mbits of traffic).
>
> Based on your results we can probably start to narrow down performance
> limitations.
>
>
> --
> Stefan Fouant
> JNCIE-SEC, JNCIE-SP, JNCIE-ENT, JNCI
> Technical Trainer, Juniper Networks
>
> Follow us on Twitter @JuniperEducate


More information about the juniper-nsp mailing list