Forum Discussion

Deon's avatar
Deon
Icon for Nimbostratus rankNimbostratus
Nov 22, 2011

LTM Object Naming Conventions

What naming conventions do people have in use for LTM objects - particularly VIPs and pools? When it comes to a VIP or pool name, where is the balance struck between cramming as much information into the name as possible (port(s), DNS Name(s), etc) and an approach where one simply names the object according to the intended use or service it provides?

 

 

Are there any Blogs on the topic here on DevCentral that I might have missed or not found in my search attempts?

 

 

 

Thanks

 

-Deon

 

  • Hi Deon,

     

     

    I like the format of:

     

     

    [app name]_[protocol]_[object type]

     

     

    like:

     

     

    www.example.com_http_vs

     

    www.example.com_http_pool

     

    www.example.com_http_monitor

     

     

    This allows you to quickly identify what app and object type an object is for via the GUI, config files or in a tech.out diagnostic file.

     

     

    In v11, (nearly) all objects have a description field which you can modify at any point. You still have to modify object names by editing the config file(s) and reloading though. So it's easiest if you come up with a standard and follow it :)

     

     

    Aaron
  • Deon's avatar
    Deon
    Icon for Nimbostratus rankNimbostratus
    I do like the simplicity of [app name]_[protocol]_[object_type]. I guess maybe I can refine my question to what do people use for the [app name] portion of the name? Reason for me starting the discussion is that I am starting to look at the possibilities of using multiple pools within one application (VIP or vs) using ProxyPass to select from possibly many pools based on the URI in the request. A dns-like name for the pools works fine where there is a one-one relationship between the vs and the pool. A dns-like app name for a pool is not quite a good fit in the case where a pool from one app is used within another app or maybe there is a pool that is default to no virtual server and thus has no dns-like name. I suppose in this scenario anything meaningful can be used for the [app name] portion of the name. Is it common to use the dns-like name in most cases and then in these situations where it does not apply or is misleading to use an alternative for [app name]?

     

     

    -Deon

     

  • Hamish's avatar
    Hamish
    Icon for Cirrocumulus rankCirrocumulus
    Virtual servers and pools I tend to standardise on FQDN-port. That way if people know what they're connecting to (URL) then I know which VS it'll be hitting.

     

     

    network VS's tend to be named net-net.num.dot.ed-masklen. e.g. net-0.0.0.0-0 is the default address. Where they conflict (e.g. two defaults depending on src clan, it gets an extra descriptor between net- and the ip address. e.g. net-wireless-0.0.0.0-0 (Default network VS for a wireless VLAN). I don't currently have any network VS's for specific ports. But they'd be something like net-0.0.0.0-0-port

     

     

    Profiles tend to be default name and a short descriptor to show why they're different. e.g. fastL4-tmo86400 for fastL4 with a 1 day idle timeout.

     

     

    Whatever you choose though. Be consistent. And have an index document. And diagrams. A diagram is worth it's weight in gold. Or more.

     

     

    H