Instant On - Wired

Expand all | Collapse all

1930 24G, slow TTFB, something wrong

  • 1.  1930 24G, slow TTFB, something wrong

    Posted 03-04-2021 04:43 PM

    just got my self a 1930 24G running my homenet with a AP22 and AP11.
    So far so good but one thing is really weird and I'm short before returning the switch:

    All Websites everything I open on the Cabled connection is really slow. It's not the Bandwidth; Once the connection is there everything is fast but the "handshake" initial time until the website load (Time to first byte) feels like on a 56k Modem. 

    It's definitly a Problem with the 1930, I have my old switch (Zyxel 24P) sitting next to it on the same router. If I just unplug my Desktop from the 1930 into the zyxel everything is quick again. If I move the cable down to the 1930 again the problem is there immediatly.

    It's cloud managed mode and on Version 1.0.4 no errors and the desktop is not even in a VLAN it's just a plain native LAN.

    No clue whats wrong.

    It's not just a slight "felt difference" it is very standing out difference. On the Zyxel a big website loads in seconds. Same site on the 1930 takes 30-40 Seconds to complete because Each individual request takes ages.

    Here are two Screenshots whre you see the huge difference of the same website, only thing I change was moving my cable from 1930 to zyxel:

    And here the Zyxel:

  • 2.  RE: 1930 24G, slow TTFB, something wrong

    Posted 03-04-2021 05:11 PM
    Ensure the connection from the router to the 1930 is using a connection a 1g,  not 100 HD or FD. Test another cable with the 1930.  If you are still having issues then reach out to support. They can review configurations and logs for you.

  • 3.  RE: 1930 24G, slow TTFB, something wrong

    Posted 03-07-2021 03:31 PM
    I will contact support. It's really weird.
    The Switch is connected 1Gbit. Already changed the cables but still something wrong.

    I found a clue that there is something with "Network Security" enabled. (ARP Protect / DHCP Snooping) as soon as this is Disabled everything is smooth.

  • 4.  RE: 1930 24G, slow TTFB, something wrong

    Posted 03-18-2021 03:43 PM

    That means that the server is very slow to respond.

    TTFB stands for "time to first byte" and it shows the amount of time that the server took to receive the request, do any server-side processing it needs, the respond with the first bytes of the page.

    Take a look at the server to optimize queries, speed up file access, reduce load, scale by increasing the number of processes, or any number of other things.

    Brown Michel

  • 5.  RE: 1930 24G, slow TTFB, something wrong

    Posted 03-18-2021 04:16 PM

    thanks for the explanation but I know what TTFB Stands for. I'm a dev and woring in IT since 20 years.

    If you would read my post fully you would understand that it is NOT a server issue. It's caused by the Network security feature in the Switch, randomly dropping packages so that only after several TCP reinit a connection can be established.

    In Comparison on the clean line or another switch the load is fine. Happening also on internal iperf TCP connections over the switch to another VLAN. So clearly a switching issue.

  • 6.  RE: 1930 24G, slow TTFB, something wrong

    Posted 30 days ago
    I would hazard a guess that you have an issue here that is very specific, but I couldn't tell you what it might be. Something in the code is most likely doing a "60 second thing" though. I can't think what would cause that on the server itself. 

    FaceTime App
    Pierce Michel

  • 7.  RE: 1930 24G, slow TTFB, something wrong

    Posted 21 days ago
    something is really strange with that 1930 ARP Protection and IPv6.

    As soon as I enable ARP PRotection / Network Security feature on my default / main LAN (VLAN 1)  I nearly instant got problems with IPv6.
    Is the ARP Protect interpreting the RA correctly or does it need to have a DCHPv6 distributing the addresses?

    The weird thing is that it is not completly dropping the IPv6 traffic but more or less throttel it. If I have a MTR running to -6 I have about 70% Packet loss. If I disable the NetSec feature the loss immedialty start to dissapear. IPv4 no Problem at all.

    My Network only uses RA to distribute  IPv6 no DHCPv6 but this shouldn't be any trouble for a modern switch and arp protect / source guard?

    For now I keep the NetSec disabled.