01-27-2014 01:56 PM - edited 03-07-2019 05:49 PM
Hi.
We have a Windows 2012 server connected via 4 ethernet to 4 ports on a 3750 CISCO switch. We have in that same switch a 1000fx SFP port connected to a iscsi target provided to us by our ISP. We are trying to utilise the target for server backups. The 4 ports are in a configuration of a LACP port channel, although there is only target and source, so this ight be a mute point. The issue we are experiencing happens if its one dedicated port, or if its just one port channel.
The issue is the backup speed, as reported by Windows is extremely slow. the switch Network Management System (PRTG) shows the port channel operating at 100M, however the results from Windows in terms of speed is all over the place. Attached below is the speeds we are getting from the windows copy service on the copy (in this example its stalled, and will come back again after a few minutes, and then sit at about 10MB/s - it starts at about 100MB/s). I thought it could be a buffer issue on the switch, but the port channel connecting to the server has no errors, and nor does the gig link to the iscsi target. Im a bit confused - I dont know if its the switch or the server, although I thought I would start with the switch. CEF is not enabled, but then its all on the same VLAN so didnt think that was necessary. Switch CPU is fine (doesnt budge).
Any help would be much appreciated.
Port-channel1 is up, line protocol is up (connected)
Hardware is EtherChannel, address is 0016.46c3.6085 (bia 0016.46c3.6085)
Description: Backup_Port_Channel_1
MTU 1500 bytes, BW 400000 Kbit, DLY 100 usec,
reliability 255/255, txload 1/255, rxload 38/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, link type is auto, media type is unknown
input flow-control is off, output flow-control is unsupported
Members in this channel: Fa1/0/1 Fa1/0/2 Fa1/0/3 Fa1/0/4
ARP type: ARPA, ARP Timeout 04:00:00
Last input never, output 00:00:00, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 975
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 59922000 bits/sec, 4945 packets/sec
5 minute output rate 1343000 bits/sec, 2518 packets/sec
15121218 packets input, 21555551584 bytes, 0 no buffer
Received 82075 broadcasts (79043 multicasts)
0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
0 watchdog, 79043 multicast, 0 pause input
0 input packets with dribble condition detected
12049544 packets output, 2902056803 bytes, 0 underruns
0 output errors, 0 collisions, 0 interface resets
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier, 0 PAUSE output
0 output buffer failures, 0 output buffers swapped out
GigabitEthernet1/0/1 is up, line protocol is up (connected)
Hardware is Gigabit Ethernet, address is 0016.46c3.6081 (bia 0016.46c3.6081)
Description:Backup_Fibre_Linknet
MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,
reliability 255/255, txload 21/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive not set
Full-duplex, 1000Mb/s, link type is auto, media type is 1000BaseSX SFP
input flow-control is off, output flow-control is unsupported
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:00, output 00:00:37, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 1904000 bits/sec, 3628 packets/sec
5 minute output rate 86268000 bits/sec, 7126 packets/sec
11047141 packets input, 2667451011 bytes, 0 no buffer
Received 660835 broadcasts (655049 multicasts)
0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
0 watchdog, 655049 multicast, 0 pause input
0 input packets with dribble condition detected
17927644 packets output, 25902052643 bytes, 0 underruns
0 output errors, 0 collisions, 0 interface resets
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier, 0 PAUSE output
0 output buffer failures, 0 output buffers swapped out
01-28-2014 04:02 AM
Hello, Brad.
It's clear why transfer over Etherchannel can't go over 100M - because LACP is doing load-sharing based on src-dst-MAC (or even in case of IP - they are the same), so the traffic is always using single link in your bundle.
Regarding gaps down from 100M - I guess you could try to capture performance couters for the iSCSI target.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide