Page 2 of 3 FirstFirst 123 LastLast
Results 11 to 20 of 25
  1. #11
    Untangle Ninja sky-knight's Avatar
    Join Date
    Apr 2008
    Location
    Phoenix, AZ
    Posts
    23,617

    Default

    I'm just being a pedant, the server is fine. It's just that when rating a router, the "internet" speed is largely irrelevant. The number you want is the throughput limit. Because that number is the aggregate total traffic the unit can process before it hits the wall. When you have two interfaces this limit is very similar to the Internet speed. But when you're a partner like me building servers that do this, I need to know how many of those gigabit interfaces you can utterly saturate before the server falls down.

    This is even more true for those monster servers that you got with 10gbit interfaces in there...

    I just hate the "gigabit" term, because it means so many different things to different people. And I'm being stupid about it at the moment, so don't mind me, enjoy your server that works wonderfully.
    Rob Sandling, BS:SWE, MCP
    NexgenAppliances.com
    Phone: 866-794-8879 x201
    Email: support@nexgenappliances.com

  2. #12
    Master Untangler deleted_account+152373@untangle.com's Avatar
    Join Date
    Sep 2016
    Location
    Malta
    Posts
    455

    Default

    Quote Originally Posted by sky-knight View Post
    I'm just being a pedant, the server is fine. It's just that when rating a router, the "internet" speed is largely irrelevant. The number you want is the throughput limit. Because that number is the aggregate total traffic the unit can process before it hits the wall. When you have two interfaces this limit is very similar to the Internet speed. But when you're a partner like me building servers that do this, I need to know how many of those gigabit interfaces you can utterly saturate before the server falls down.

    This is even more true for those monster servers that you got with 10gbit interfaces in there...

    I just hate the "gigabit" term, because it means so many different things to different people. And I'm being stupid about it at the moment, so don't mind me, enjoy your server that works wonderfully.
    Yes dude I understand what you were trying to point out , but hey its sometimes fun to have a little debate right? lol I know you always mean well but thanks for pointing some stuff out thats how we learn and we can never learn enough or know it all
    I like to listen. I have learned a great deal from listening carefully. Most people never listen

  3. #13
    Master Untangler Sam Graf's Avatar
    Join Date
    Feb 2016
    Location
    Michigan
    Posts
    675

    Default

    Quote Originally Posted by sky-knight View Post
    I just hate the "gigabit" term, because it means so many different things to different people. And I'm being stupid about it at the moment, so don't mind me...
    Not stupid, I don't think, because you're absolutely right. It'd be nice if there was a truly satisfying way to describe Internet connections speeds, something consistent with the technical reality. Something other than "Oh, wow" or "Ugh."

  4. #14
    Untangler
    Join Date
    Dec 2014
    Location
    Baton Rouge, LA
    Posts
    54

    Default

    Now do a DSLReports Fiber test, custom settings 32/32 stream, randomized uncompressed. You'll feel the heat coming off that little J1900.

  5. #15
    Master Untangler deleted_account+152373@untangle.com's Avatar
    Join Date
    Sep 2016
    Location
    Malta
    Posts
    455

    Default

    Quote Originally Posted by thecomputerdude View Post
    Now do a DSLReports Fiber test, custom settings 32/32 stream, randomized uncompressed. You'll feel the heat coming off that little J1900.
    Sure it would crawl it down.

    This is a simple test on the j1900 with simple stuff, hey add IDS/IPS and it would crawl even more. My real server is I3 Kaby lake and it laughs at these speeds with all test including the one you proposed
    Last edited by Chrismal; 11-09-2018 at 05:51 AM.
    I like to listen. I have learned a great deal from listening carefully. Most people never listen

  6. #16
    Untangler
    Join Date
    Dec 2014
    Location
    Baton Rouge, LA
    Posts
    54

    Default

    Quote Originally Posted by Chrismal View Post
    Sure it would crawl it down.

    This is a simple test on the j1900 with simple stuff, hey add IDS/IPS and it would crawl even more. My real server is I3 Kaby lake and it laughs at these speeds with all test including the one you proposed
    The point is, Speedtest uses packet compression and specifically enables certain packet sequences to artificially creep their reported speeds up. It's a marketing gimmick. Now, if you actually got bad speeds, then there's a serious issue

    And it's not important for a barebones setup like you've got for testing, but throw an encrypted workload on there and experience the benefit of AES-NI...or rather what happens when you don't have it haha.
    Last edited by thecomputerdude; 11-09-2018 at 01:17 PM.

  7. #17
    Untangler
    Join Date
    Dec 2014
    Location
    Baton Rouge, LA
    Posts
    54

    Default

    Quote Originally Posted by Chrismal View Post
    @thecomputerdude

    Dude what the hell are you going on about? shit I did a speed test and that's it who asked about AES-NI, encrypted, traffic speedtest trash, If you read my other post I wrote I also Have the I3 Kaby Lake that is AES-NI If you didn't know so yes I did experience it , now If you really think my speed test is crap just move on to more interesting posts
    It's not crap, it's a perfectly good representation of speed. However, I doubt your use case even touched that J1900 and just ran straight across the perfectly capable Intel NIC. In other words, you have an extremely optimized workload that is extremely avoidant of the processor

    You did mention you weren't averse to a debate, so.... I wanted to see if a J1900 could theoretically carry gigabit speed when it's more involved in the matter. I don't have any J1900 systems I can test myself. For Science!

    I ran the same test (same rack setup) across a much beefier processor and only got a <0.1 load out of it with Speedtest. I ran a DSLReports test (customized) and saw a ~0.4 load.
    Last edited by thecomputerdude; 11-09-2018 at 07:10 PM.

  8. #18
    Master Untangler deleted_account+152373@untangle.com's Avatar
    Join Date
    Sep 2016
    Location
    Malta
    Posts
    455

    Default

    Quote Originally Posted by thecomputerdude View Post
    It's not crap, it's a perfectly good representation of speed. However, I doubt your use case even touched that J1900 and just ran straight across the perfectly capable Intel NIC. In other words, you have an extremely optimized workload that is extremely avoidant of the processor

    You did mention you weren't averse to a debate, so.... I wanted to see if a J1900 could theoretically carry gigabit speed when it's more involved in the matter. I don't have any J1900 systems I can test myself. For Science!

    I ran the same test (same rack setup) across a much beefier processor and only got a <0.1 load out of it with Speedtest. I ran a DSLReports test (customized) and saw a ~0.4 load.
    Dude I apologize It was a long day and saw things the wrong way, should have went strait to bed ) hope we are still cool.
    Doing some more tests on the J1900 DSL reports with the custom options drops to around 650mbps, also tested the speed internally by pluging in my zfs server on the wan side and doing a file transfer, also iperf I could not get over 700mbps max but it mostly stayed on the 550mbps side, if you have any other taughts or ideas to try out please share
    thanks for chiming in really
    I like to listen. I have learned a great deal from listening carefully. Most people never listen

  9. #19
    Master Untangler deleted_account+152373@untangle.com's Avatar
    Join Date
    Sep 2016
    Location
    Malta
    Posts
    455

    Default

    Here is a test running HTOP in ssh


    DSL.PNG
    I like to listen. I have learned a great deal from listening carefully. Most people never listen

  10. #20
    Untangle Ninja sky-knight's Avatar
    Join Date
    Apr 2008
    Location
    Phoenix, AZ
    Posts
    23,617

    Default

    That's normal, that device should bottle up between 600 and 800mbit. Which is "almost gigabit". I have a new appliance with a trailing letter D, because it's not G... because it's "almost" a G.

    Yes, I was that tired while naming it... but I didn't know how else to do it in a model number. You need an i3, with good interfaces, and a good bus to get a real gigabit through. The Atoms / Celerons just can't quite make it. But, they are close.

    Meanwhile we have gigabit to homes, and you need enterprise level hardware to fully support it. Fortunately, the backbones in the US arne't wide enough to do gigabit anyway so we have time before real performance suffers.
    Synical likes this.
    Rob Sandling, BS:SWE, MCP
    NexgenAppliances.com
    Phone: 866-794-8879 x201
    Email: support@nexgenappliances.com

Page 2 of 3 FirstFirst 123 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

SEO by vBSEO 3.6.0 PL2