Forum Discussion

chris_noon_3316's avatar
chris_noon_3316
Icon for Altocumulus rankAltocumulus
Feb 06, 2018
Solved

BIG-IP DNS and LTM on 1 node

We are looking at a new F5 BIG-IP DNS and LTM deployment. It seems (based on forums) that the BIG-IP DNS configuration must be performed in the common partition as it can cause issues in an added partition. LTM I know has no issues in added partitions.

 

My question is...

 

Should we buy 1 high-end box and have the BIP-IP DNS run in the common partition while the LTM instances run in added partitions?

 

Or is it recommended we buy a BIP-IP DNS stand-alone device?

 

Or can BIG-IP DNS now run in partitions without problems?

 

Does anyone have F5 documentation supporting the best practice for this type of design?

 

application delivery
BIG-IP
BIG-IP DNS
  • chris_noon_3316's avatar
    chris_noon_3316
    Feb 09, 2018

    I spoke to a friend I have working at F5 and these are his thoughts:

     

    There are serval options for my deployment type and no set best practice design. It is about scalability, growth and what I am most comfortable with.

     

    The GLB is typically deployed either on a separate piece of hardware to the LTM devices, or run within a separate vCMS (virtual instance) on the same tin and not within a separate partition (https://support.f5.com/kb/en-us/products/big-ip_ltm/manuals/product/vcmp-viprion-configuration-11-4-1/1.html).

     

    The GLB typically runs in active/active mode utilizing the distributed sync groups. This is similar to a HA pair running active/active where the configuration is synced between the devices to ensure config consistency.

     

    Based on the above you can either buy 1 physical, 2 physical, 2 Virtual or 1 physical and 1 Virtual.

     

    This doesn’t give you a clear path/answer to follow for best practice design, it’s a follow your mind/heart approach with what you a) feel comfortable with and b) feel is most robust/scalable.

     

    Based on this information I will be looking at 3 options:

     

    • 2 VMs, 1 for GTM and 1 VM for the LTM partitions
    • 1 physical for GTM and 1 VM for the LTM partitions
    • 1 physical utilizing vCMS

    That being said, anyone reading this forum with similar design questions can select anything they want as it depends on a case by case basis and your business decisions.

     

10 Replies

Sort By
  • Surgeon's avatar
    Surgeon
    Ret. Employee
    Feb 06, 2018

    Who told you that GTM can not work in partitions other then common? What you are going to achieve wit partitions?

     

    Partitions designed to segregate access to the big-ip objects based on user roles.

     

  • chris_noon_3316's avatar
    chris_noon_3316
    Icon for Altocumulus rankAltocumulus
    Feb 06, 2018

    Hi Surgeon,

     

    I read a few forum/blog articles about people having issues running GTM, now BIP-IP DNS on none common partitions. I was hoping this is no longer an issue.

     

    We are looking to have 1 partition as a GTM, 1 partition as a Web/Application LB and 1 partition as a DB LB. Leaving the common partition with minimal configuration.

     

    Does the about sound like it will work?

     

  • Surgeon's avatar
    Surgeon
    Ret. Employee
    Feb 06, 2018

    What do you mean by telling "I read a few forum/blog articles about people having issues running GTM, now BIP-IP DNS on none common partitions"

     

    Can you share mode details about that?

     

  • chris_noon_3316's avatar
    chris_noon_3316
    Icon for Altocumulus rankAltocumulus
    Feb 07, 2018

    Hi Surgeon,

     

    I have read this F5 Dev conversation and it appears there is no issue with what I plan to do: https://devcentral.f5.com/s/feed/0D51T00006i7QpDSAU

     

    I have read this F5 Dev conversation and it mentions issues deploying GTM (now BIG-IP) on anything other than the common partition: https://devcentral.f5.com/s/feed/0D51T00006i7WrTSAU

     

    My understanding is that GTM and LTM can be run in partitions without issue. I was just trying to get confirmation from someone with more F5 experience.

     

    I suspect the article with issues is either old version of code or strange configuration.

     

  • chris_noon_3316's avatar
    chris_noon_3316
    Icon for Altocumulus rankAltocumulus
    Feb 07, 2018

    Understood, thank you.

     

    So these are my questions:

     

    • Can I create these objects in the Common partition? Then utilize them in a BIG-IP DNS partition?

       

    • Or should the Big-IP DNS be configured and run from the common partition while other LTM services run from their own partitions?

       

    • Or is it best to buy a separate Big-IP DNS device and run it all within the common partition and have no other partitions?

       

  • Surgeon's avatar
    Surgeon
    Ret. Employee
    Feb 07, 2018

    Chris, in order to answer this question, can you answer what you are trying to achieve with partitions?

     

  • chris_noon_3316's avatar
    chris_noon_3316
    Icon for Altocumulus rankAltocumulus
    Feb 07, 2018

    We need 1 x BIG-IP DNS and 2 x LTM F5's, per site.

     

    The Big-IP DNS will push traffic to either our Production or DR site depending on availability. This will be 1 partition.

     

    The LTMs will serve a Web/App layer (this will report to the BIG-IP DNS) and a DB layer. This will be 2 separate partitions.

     

    Totalling 3 partitions.

     

    We would like to run the Big-IP DNS and the 2 x LTMs on one device per site.

     

    Is that the information you are looking for?

     

  • Surgeon's avatar
    Surgeon
    Ret. Employee
    Feb 08, 2018

    This is not exactly I am looking for. Can I know the reason of using partitions? Why did you decid to use partitions?

     

  • chris_noon_3316's avatar
    chris_noon_3316
    Icon for Altocumulus rankAltocumulus
    Feb 09, 2018

    I spoke to a friend I have working at F5 and these are his thoughts:

     

    There are serval options for my deployment type and no set best practice design. It is about scalability, growth and what I am most comfortable with.

     

    The GLB is typically deployed either on a separate piece of hardware to the LTM devices, or run within a separate vCMS (virtual instance) on the same tin and not within a separate partition (https://support.f5.com/kb/en-us/products/big-ip_ltm/manuals/product/vcmp-viprion-configuration-11-4-1/1.html).

     

    The GLB typically runs in active/active mode utilizing the distributed sync groups. This is similar to a HA pair running active/active where the configuration is synced between the devices to ensure config consistency.

     

    Based on the above you can either buy 1 physical, 2 physical, 2 Virtual or 1 physical and 1 Virtual.

     

    This doesn’t give you a clear path/answer to follow for best practice design, it’s a follow your mind/heart approach with what you a) feel comfortable with and b) feel is most robust/scalable.

     

    Based on this information I will be looking at 3 options:

     

    • 2 VMs, 1 for GTM and 1 VM for the LTM partitions
    • 1 physical for GTM and 1 VM for the LTM partitions
    • 1 physical utilizing vCMS

    That being said, anyone reading this forum with similar design questions can select anything they want as it depends on a case by case basis and your business decisions.