SPS224G4 THROUGHPUT PROBLEM

[vlutsenko]

Good day,

during using  switches Linksys sps224g4 faced with the following problem.

When the end user is connected in 100 Mbps port of the switch and rtt  more than 10 ms to a remote server, the download speed from server in  one stream does not exceed 2.5 MB.

When rtt less than 5 ms, the download speed in the one stream 11 MB.

When switching user from 100Mbps to 1Gbps port, download speed from  remote server grows up to 11-14 MB.

When we connect user to 100Mbps port of  another manufacturer switch in  one stream download speed reaches 10MB.

Tried to configure next parametrs:

1. Checked for firmware 1.0.2 and 1.0.6, the situation is identical.

2. Enable, disable negotiation on the client port.

3. Turn on / off qos.

4. Turn on / off flow-control.

5. Turn on / off jambo-frame (I do not know how it could have an  influence, just in case).

6. During traffic analizing on the users port were found no differences  between downloading from servers with rtt 5 and 35 ms.

Best regards,

Vyacheslav Lutsenko

ISP Freenet

[Tom Watts]

Hi Vyacheslav, to the remote server can you run a trace to see if there is a bad route hop or a spike on a particular hop? In which manner is the switch deployed? Is it serving as an aggregator or an access to CPE?

-Tom
Please mark answered for helpful posts

[vlutsenko]

Hi Tom,

device is used like an access to CPE.

Here is result of the trace to the remote server:

                                                                                    Packets               Pings

Host                                                                             Loss%   Snt   Last   Avg  Best  Wrst StDev

1. 188.231.188.33                                                            0.0%    28    2.9   3.3   2.7  13.1   2.0

2. 109.251.255.38                                                            0.0%    27    5.5   6.1   5.3  10.9   1.7

3. 77.88.212.149                                                              0.0%    27    5.6  10.1   5.5 120.9  22.2

4. 166.63.204.197                                                            0.0%    27   35.7  37.6  35.7  54.6   4.4

5. 192.2.21.89                                                                  0.0%    27   35.9  41.2  35.8 111.4  16.4

6. 195.27.183.143                                                             0.0%    27   34.5  34.7  34.4  38.4   0.8

It is no matter from what remote server make a test, I know alot servers with good thoughput and all time I get the same result. More over, our company have 2 different uplink providers, and situation is the same in both of them.

If it will be helpful to understand the problem, I can attach to the post switch configuration file and dumps of wireshark, which were made during speed test. 

Thanks for your help.

[Tom Watts]

Are you able to get a utilization report from the uplink providers? It may be an over utilization on the aggregate lines.

-Tom
Please mark answered for helpful posts

[vlutsenko]

Good day.

It's not a problem of external channels utilization because of next points:

-the problem exists all time and not depends on part of day;

-we have 2 independent external channels and problem exists on both of them

-we have level-2 switches of other vendors, and they have not that problem

On my mind, there are some problem with queues algorithm.

Best regards,

Vyacheslav Lutsenko

ISP Freenet

[Tom Watts]

May just need to replace the switch. The replacement SKU should be SRW224G4-k9-

Here's is the links to spec it out

http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps10898/data_sheet_c78-610061.html

http://www.cisco.com/en/US/docs/switches/lan/csbms/sf30x_sg30x/administration_guide/78-19308-01.pdf

-Tom
Please mark answered for helpful posts

[vlutsenko]

It is a good idea, I would have done, if it was only one switch in our network.

But unfortunately, we have installed a big count of switches this model.

Best regards,

Vyacheslav Lutsenko

ISP Freenet