Performance problem w/ SG300-20?

portmannth · ‎10-31-2013

Hello,

I am having performance problems with SG300-20 switches. I chose them to put in a few offices needing access to several different VLANs and since they are fanless this seemed the perfect choice (we have other Cisco devices, too). After putting one in service for tests, I see quite some performance problems. Let me describe the environment:

We have Oracle SunRay ultra thin clients. The clients (SunRays) connect to Catalyst 2960s switches.

1. When I connect a SunRay directly to an 2960s (100F), the SunRay server shows very little packet loss (around 0.02%).

2. When I have an SG300-20 between the SunRay (100F) and the 2960s (100F), then the packet loss is comparable to 1. that is around 0.02%.

3. When I again, like in 2., have the SG300-20 between the Catalyst and the SunRay, but the uplink (Catalyst to SG300) ist 1000F, then the packet loss rises to somewhere between 7 and 13%!!!

4. When I replace in 3. the SG300-20 and put a 3Com (OfficeConnect Managed Switch 9) between the Catalyst and the SunRay, the packet loss reported by the SunRay server is 0.000%.

In these tests I only had 2 ports used on the SG300 and the 3Com, the uplink and the client and only access ports where used, no tagging. The firmware installed is:

SW version 1.3.0.62 ( date 02-May-2013 time 14:55:01 )

Boot version 1.1.0.6 ( date 11-May-2011 time 18:31:00 )

HW version V02

Since the SG300-20 does wirespeed forwarding, I would expect no packet loss at all, at least no a very important one. Is there something in the configuration to be tweaked or is this a known firmware problem?

I should also mention that I did not see anything special on the port of the Catalyst, nor on the SG300-20. I also tested with another SG300-20 and had the same behavior.

Thanks for your input and best regards

Tom

Tom Watts · ‎10-31-2013

Hi Portmannth, could you try to downgrade the software to 1.1.2.0 then factory reset the switch and try again?

-Tom
Please mark answered for helpful posts

-Tom Please mark answered for helpful posts http://blogs.cisco.com/smallbusiness/

portmannth · ‎11-05-2013

Hello Tom,

Tried the downgrade to 1.1.2.0 + factory reset + minimal config (VLANs) for testing, but I see no difference in quality.

I may add that the SunRay traffic is quite bursty and UDP and this may be the source of the problem. Doing some tests to render to whole a bit more reproductible, I use iperf with 2 PCs:

A 192.168.254.21, connected 1000F in our network (iperf server: iperf -u -s)

B 192.168.254.11, connected 100F in my office (iperf client), connected to the various test switches

The interesting part is when A sends B the data too fast (150M), with both Cisco switches, the received data is just a small fraction from what it is with the 3com switch and I think this is related to my poor performance with the sunrays because the sunray servers are all connected 1000F.

B on sg300-20> iperf -u -c 192.168.254.21 -r -b 150M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 4] local 192.168.254.11 port 59038 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 4] 0.0-10.0 sec 114 MBytes 95.7 Mbits/sec

[ 4] Sent 81353 datagrams

[ 4] Server Report:

[ 4] 0.0-10.0 sec 114 MBytes 95.6 Mbits/sec 0.528 ms 0/81352 (0%)

[ 4] 0.0-10.0 sec 1 datagrams received out-of-order

[ 3] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 41197

[ 3] 0.0-10.3 sec 7.18 MBytes 5.88 Mbits/sec 15.335 ms 123049/128173 (96%)

[ 3] 0.0-10.3 sec 2 datagrams received out-of-order

B on sg300-20>

iperf -u -c 192.168.254.21 -r -b 90M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 3] local 192.168.254.11 port 43250 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 3] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec

[ 3] Sent 76924 datagrams

[ 3] Server Report:

[ 3] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.016 ms 0/76923 (0%)

[ 3] 0.0-10.0 sec 1 datagrams received out-of-order

[ 4] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 36854

[ 4] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.005 ms 1/76923 (0.0013%)

[ 4] 0.0-10.0 sec 2 datagrams received out-of-order

B on 3com> iperf -u -c 192.168.254.21 -r -b 150M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 4] local 192.168.254.11 port 52562 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 4] 0.0-10.0 sec 114 MBytes 95.7 Mbits/sec

[ 4] Sent 81372 datagrams

[ 4] Server Report:

[ 4] 0.0-10.0 sec 114 MBytes 95.6 Mbits/sec 0.070 ms 0/81371 (0%)

[ 4] 0.0-10.0 sec 1 datagrams received out-of-order

[ 3] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 55654

[ 3] 0.0-10.3 sec 107 MBytes 87.6 Mbits/sec 14.746 ms 51848/128194 (40%)

[ 3] 0.0-10.3 sec 2 datagrams received out-of-order

B on 3com> iperf -u -c 192.168.254.21 -r -b 90M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 4] local 192.168.254.11 port 45013 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 4] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec

[ 4] Sent 76924 datagrams

[ 4] Server Report:

[ 4] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.017 ms 0/76923 (0%)

[ 4] 0.0-10.0 sec 1 datagrams received out-of-order

[ 3] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 44381

[ 3] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.006 ms 5/76923 (0.0065%)

[ 3] 0.0-10.0 sec 6 datagrams received out-of-order

B on c2960s> iperf -u -c 192.168.254.21 -r -b 150M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 5] local 192.168.254.11 port 41923 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 5] 0.0-10.0 sec 114 MBytes 95.7 Mbits/sec

[ 5] Sent 81352 datagrams

[ 5] Server Report:

[ 5] 0.0-10.0 sec 114 MBytes 95.6 Mbits/sec 0.528 ms 0/81351 (0%)

[ 5] 0.0-10.0 sec 1 datagrams received out-of-order

[ 4] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 43634

[ 4] 0.0-10.3 sec 13.9 MBytes 11.4 Mbits/sec 14.125 ms 118246/128173 (92%)

[ 4] 0.0-10.3 sec 13 datagrams received out-of-order

B on c2960s> iperf -u -c 192.168.254.21 -r -b 90M

------------------------------------------------------------

Server listening on UDP port 5001

Receiving 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

Client connecting to 192.168.254.21, UDP port 5001

Sending 1470 byte datagrams

UDP buffer size: 208 KByte (default)

------------------------------------------------------------

[ 3] local 192.168.254.11 port 34602 connected with 192.168.254.21 port 5001

[ ID] Interval Transfer Bandwidth

[ 3] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec

[ 3] Sent 76924 datagrams

[ 3] Server Report:

[ 3] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.010 ms 0/76923 (0%)

[ 3] 0.0-10.0 sec 1 datagrams received out-of-order

[ 4] local 192.168.254.11 port 5001 connected with 192.168.254.21 port 41599

[ 4] 0.0-10.0 sec 108 MBytes 90.5 Mbits/sec 0.008 ms 3/76923 (0.0039%)

[ 4] 0.0-10.0 sec 4 datagrams received out-of-order

tom

Kremena Ivanova · ‎11-01-2013

Hi,

Test with Green Ethernet disabled on SG300. If SG300 is PoE try disabling the PoE on the uplink port with the catalyst

Regards,

Kremena

portmannth · ‎11-05-2013

Hi Kremena,

Also tried to disable eee and green-ethernet globaly and on all interfaces, but with no change in the results. This is not an PoE switch, so your second suggestion does not apply.

Thank you

tom

Tom Watts · ‎11-05-2013

Hi Tom, can you try manually setting the negotiation on the sunray and sx300 to 1000f?

-Tom
Please mark answered for helpful posts

-Tom Please mark answered for helpful posts http://blogs.cisco.com/smallbusiness/

portmannth · ‎11-05-2013

Hello Tom,

Well no, the SunRays are very dumb devices and they do only auto negociation. Moreover, the SunRay 2 I am using only has an 100base-T interface (I also made some other, earlier, tests with an SunRay 3 which has an 1000base-T interface, but I suspect that the SunRay CPU could not cope with the network traffic (lots of pause frames received on the switch (c2960s)), because the video quality was much worse compared to an 100F negociation).

Tom