SD-WAN

 View Only
last person joined: 9 days ago 

Ask questions and share experiences with SD-WAN and Session Smart Router (formerly 128T).
  • 1.  NTP Configuration

    Posted 04-04-2019 05:11
    Hi All,

    Please provide some detail on how NTP is currently being deployed on nodes for customers, as well as if this might form part of conductor services in the future to make things simpler? 

    Regards,

    ------------------------------
    Morne Vermeulen
    Engineer
    ------------------------------


  • 2.  RE: NTP Configuration

     
    Posted 04-05-2019 12:35
    Hello @Morne,

    As of 4.0, keeping the proper system time via NTP has become a requirement on our 128T platform to ensure analytics are stored properly, so this question is perfect on timing.

    The 128T software today leverages NTPD in order to ensure the system time is kept. Within the router configuration you can utilize an FQDN or IP address. Here's some example below:
    amorris@aws-vm.conductor-field-eng# show ntp router seattle-site-02-lanner
    Fri 2019-04-05 16:15:27 UTC
    
    Node: grimlock
    
    ======== ================== ================= ========= ====== ====== ====== ======= ======== ========= ======== ============
     Status   Time Source        Ref. ID           Stratum   Type   When   Poll   Reach    Delay    Offset   Jitter   Tally Code
    ======== ================== ================= ========= ====== ====== ====== ======= ======== ========= ======== ============
     active   +up2.com           58.148.140.87           2   u         1     64       1   90.833   -27.028    1.910   candidate
     active   *vps6.ctyme.com    216.218.254.202         2   u         2     64       1   29.728   -21.401    2.759   syspeer
     active    triton.ellipse.   128.252.19.1            2   u         1     64       1   73.888   -27.292    1.317   reject
     active    ns2.switch.ca     206.108.0.131           2   u         2     64       1   40.845   -17.898    0.000   reject
    
    Completed in 1.31 seconds
    amorris@aws-vm.conductor-field-eng# show config running authority router seattle-site-02-lanner system ntp
    
    config
    
        authority
    
            router  seattle-site-02-lanner
                name    seattle-site-02-lanner
    
                system
    
                    ntp
    
                        server  0.north-america.pool.ntp.org
                            ip-address  0.north-america.pool.ntp.org
                        exit
    
                        server  1.north-america.pool.ntp.org
                            ip-address  1.north-america.pool.ntp.org
                        exit
    
                        server  2.north-america.pool.ntp.org
                            ip-address  2.north-america.pool.ntp.org
                        exit
    
                        server  3.north-america.pool.ntp.org
                            ip-address  3.north-america.pool.ntp.org
                        exit
                    exit
                exit
            exit
        exit
    exit
    
    
    amorris@aws-vm.conductor-field-eng# show config running authority router conductor-field-eng system ntp
    
    config
    
        authority
    
            router  conductor-field-eng
                name    conductor-field-eng
    
                system
    
                    ntp
    
                        server  38.229.71.1
                            ip-address  38.229.71.1
                        exit
    
                        server  104.155.144.4
                            ip-address  104.155.144.4
                        exit
    
                        server  104.156.99.226
                            ip-address  104.156.99.226
                        exit
    
                        server  45.79.1.70
                            ip-address  45.79.1.70
                        exit
                    exit
                exit
            exit
        exit
    exit
    
    amorris@aws-vm.conductor-field-eng# show ntp router conductor-field-eng
    Fri 2019-04-05 16:25:29 UTC
    
    Node: aws-vm.conductor-field-eng
    
    ======== ================== ============== ========= ====== ====== ====== ======= ======== ======== ======== ============
     Status   Time Source        Ref. ID        Stratum   Type   When   Poll   Reach    Delay   Offset   Jitter   Tally Code
    ======== ================== ============== ========= ====== ====== ====== ======= ======== ======== ======== ============
     active   +clock.team-cymr   204.9.54.119         2   u       412   1024     377   26.871   -3.000    0.707   candidate
     active   *4.144.155.104.b   216.239.35.8         2   u       239   1024     377   29.206   -2.678    0.607   syspeer
     active    104.156.99.226    .INIT.              16   u         -   1024       0    0.000    0.000    0.000   reject
     active   +dfw0.clover.mat   129.7.1.66           2   u       491   1024     377   29.696   -1.411    0.642   candidate
    ​

    This NTP sync traffic will utilize the "management" port and thus follow the Linux routing table.

    Note, if using the FQDN approach, the mgmt interface must be able to resolve these FQDNs. The easiest way to do this is outlined in this article which cover both cases of out-of-band or in-band mgmt interfaces by setting up global DNS servers.

    #NTP #Analytics #4.0.0
    ------------------------------
    Adam Morris
    Sales Engineer
    WA
    (206) 617-4999
    ------------------------------



  • 3.  RE: NTP Configuration

    Posted 04-09-2019 04:01
    Thanks @Adam,

    However, I was more leaning towards a way of deploying this at mass scale. I understand that Linux and 128T have different routing tables and that the DNS functionality should sit on Linux (because that's where we'll configure NTP and where 128T draws it's system time from). 

    But from my perspective, if we had to deploy 200 sites for a project, how would I make this easier on the rollout team? Having to create a global-dns.conf file with those contents on every device that we need to ship hardly seems scalable.

    On a different note, this also seems like it would work on a "manual" management interface, but not with the interface created when conductor services are enabled? From my knowledge, ticking the "conductor" slider on an interface, automatically adds a /32 route in the linux routing table along with creating a kni interface that traffic would be sourced from and sent to the conductor. Which means this config will not work with conductor service?

    Also - is there any way to set up NTP servers on an authority level, and not on a per-router level? Something where I can specify 4 servers at an authority level, and I could maybe change a single device if that device alone had different NTP needs. 

    Regards,

    ------------------------------
    Morne Vermeulen
    Engineer
    +27 (0) 10 141 8512
    ------------------------------



  • 4.  RE: NTP Configuration

     
    Posted 04-10-2019 08:17
    Hi @Morne, to answer your concern regarding scaling installation process to a large number of sites, there are several ways in which you can tackle this goal.  Number one, we offer custom SALT states, including global-dns configuration, which make it easy to automate the process of adding DNS servers/services, define firewall zones/rules, set time-zone, and so forth.  This is discussed elsewhere on Interchange.  Number two, you can configure your Linux services that are repeatable and do not vary, and create a snapshot image of it to blast on each machine.  We have quite a bit of experience with these processes as an organization, and can help you with essentially "templatizing" your installation.​

    ------------------------------
    Gene Shtirmer
    Sales Engineer
    Randolph NJ
    (973) 610-5676
    ------------------------------



  • 5.  RE: NTP Configuration

     
    Posted 04-10-2019 20:02
    Hello @Morne, as @Gene pointed out, today we use a variety of devops tools and/or system blasting techniques to enable an early version of "management over forwarding". These techniques have been successfully used on most production branch deployments today. We can work with you to provide the solution that meets your needs either way.

    Also, we are planning to implement the full "management over forwarding" natively in the product in the short term. It is currently scheduled for a 4.x release which will come out later this year. In the interim, you can use the built in "conductor over forwarding" feature (network-interface -> conductor -> 'true') to achieve the early "management over forwarding" as well by using steps 1-5 here:
    https://community.128technology.com/HigherLogic/Security/SAML/localSAMLLoginService.aspx?binding=HttpPost&ReturnUrl=https%3a%2f%2fcommunity.128technology.com%2fcommunities%2fcommunity-home%2fdigestviewer%2fviewthread%3fGroupId%3d43%26MessageKey%3d13dbbb4f-b6df-4627-bb5b-0b97d18c3ec1%26CommunityKey%3d1cca2e49-ba26-471a-9522-f5ccd96c86fe%26tab%3ddigestviewer
    128technology remove preview
    View this on 128technology >

    In regards to having the NTP servers be at the "authority" level, I will reach out to your account team and get this added as a feature request as I agree this has its advantages (especially when using a FQDN).​​

    ------------------------------
    Adam Morris
    Sales Engineer
    WA
    (206) 617-4999
    ------------------------------



  • 6.  RE: NTP Configuration

    Posted 04-12-2019 10:18
    Just for references i opened a feature request for that specific request. (ID:28021)

    ------------------------------
    Sebastian Hofmann
    Sales Engineer - EMEA
    ------------------------------