Best Way to Configure Backup T1 Interface

[Patrick McHenry]

Hi,

I was wondering what was the best practice is for configuring a backup T1 interface? The primary interface is a DS3 to HQ.

Could it be possible with a DS3 to have an interface show as up but the DS3 not function?

I was thinking if this is the case - maybe I should configure a IP SLA pinging the other side and when that fails, I could run an EEM script to bring up the T1 and when the pings come back I could shut the T1.

Any ideas or docs on how to set up?

Also, any docs on how to set up the T1 interface?

Thank you, Pat.

[John Blakley]

Pat,

Can you tell us a little about how you're set up? Do the 2 interfaces connect to the same ISP? Do you run a routing protocol or static routes? You may be able to do something as simple as using the "backup interface" command on your primary interface or you can use floating statics with tracking. What type of card do you have to configure for T1?

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Thanks for the response

The interfaces connect to two different providers but, they are both dedicated links. I will be running eigrp on this branch router and the router at HQ. I'm not experienced with T1 interfaces, I believe they are PRI - it's not in front of me but, I can confirm tomorrow morning. What should I look for with the T1s as far as type?

Thank you, Pat.

[John Blakley]

In my experience, sometimes you have to configure the card with a "card type t1" command in order to get your interfaces, but then there are some cards that when inserted automatically give you your serial interfaces to configure. If you do a "show run" you'll see "card type in slot not configured" or something like that if you need to configure it manually.

Are these internet links or links to an mpls vpn? If their vpn, you could keep both up and just set the metrics where your ds3 is always preferred or you can set the ds3 to be the primary and the t1 to be the backup:

DS3 interface:

backup interface s0/1

T1 interface (s0/1):

ip address 192.168.1.1 255.255.255.0

The T1 will stay down unless the router sees that the DS3 went down and then the T1 interface will be brought up. You can't shut the ds3 interface though to test it because it has to be "up/down" I believe for the failover to work. If you shut the interface it will also bring down the t1. If this is an internet link for both, you'll need to take into account all of your natting and how things will change when going over the T1 instead. It's late and I hope I'm making sense

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Thanks John - sorry if you should be sleeping.

I think I will have to do the card type t1 command as I had to do card type t3 to get the DS3 interface to show up. These are not Internet connections, they are dedicated circuits. Not sure if they are MPLS vpn - I just know that they are a layer 2 connection and I will have a /30s between the HQ and the branch on the DS3 and the T1.

My concern is this - if I use the metrics method or the backup interface method, and the DS3 is not working (meaning the brach cannot reach the HQ via the DS3) couldn't the interface be still up or am I wrong - if the branch loses a layer 3 connection with HQ it will show as down and the T1 will take over?

Also, instead of setting metrics - won't the metrics appear automatically and the metrics for the DS3 will be better so EIGRP will prefer that route (which is what I want)?

Thank you, Pat.

[John Blakley]

Pat,

That could happen. So you want to take into account the far side going down yet your connection to the ISP is still showing up. You may have to use sla and eem in order to do this like you had originally stated. Tracking will let you track routes or interfaces via glbp/hsrp, etc, but I don't believe it's going to be possible to bring up an interface based on a tracked object. Your eem script can take advantage of the tracked object though, so I'd create an icmp sla that pings the opposite end of the link and then your eem script can act on the failure of that object.

If you are using static routes only between the two sites, you could, for example, set your default route to your DS3 and track on that (same icmp sla). You could also have another default route that has a higher AD and when the tracked object fails, the tracked default static route will be removed from the table and the default that points to the T1 will be put in its place. Let me know if you'd like to do the sla with eem and I can help you with that if you need....

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Thanks John,

If the branch DS3 link loses Layer 3 connectivity to the HQ (which is what I am concerned about) but retained Layer2 connectivity with the carrier, wouldn't the T1 take over because of EIGRP? Shouldn't EIGRP sense theL3 connection is down and immediately route through the T1? Or, could it be possible, like you and I said that the router would still route packets through the DS3 even though they would be getting blackholed.

I would rather not configure an IP SLA and a EEM script if I don't have to but, if that is the only option then I will. I'm looking for immediate change over during a link fail and I don't thinka EEM script can neccessarily do that. I guess I could have a combo of both - if the link goes down EIGRP will move the traffic to the T1 and if the branch link stays up but the HQ is down, the IP SLA would trigger the EEM script.

Thank you.

[John Blakley]

Pat,

If you shut down the far side interface as a test, do you see your side go down? Technically, eigrp would prefer the next path. If you do a "show ip eigrp topol" do you have multiple entries for the same route (at least 2), and if so, do you have single entries in your routing table now? You'd need to have at least 2 in your topology table for it to immediately fail over, but in working conditions you'd really only want your traffic to go out of your ds3 if you're not wanting to load balance. Load balancing would require that you have identical metrics through two different paths (your ds3 and t1). Being that the t1 is a lot slower than your ds3, I'm also going to assume that all of your routes would only go through the ds3 at this point in time.

In short, yes, if eigrp is engineered correctly, I would say that it would use the t1 as a backup but I could lab this up to just double check.

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Yes I need to test this before I head out to the branch which is probably 800 miles away. I'll have to use completey different addressing as the branch is up at this moment.

As far as the T1 interface - I did the command as you suggested and now I see the T1s under show run. the default config is:

controller T1 0/2/0

framing esf

linecode b8zs

cablelength long 0db

!

Is this the only config the interface will need other than an IP and mask?

Thank you

[John Blakley]

Pat,

It should be the only config. Since you see that, you should have "int s0/2/0" now and be able to configure you addressing information and everything else you need.

I'll try to lab this up today. Do you know if your two sites share the same subnet on their wan interfaces? If you look at a route to the other location on this router with the DS3, does it show the other site as the next hop for a subnet that the other site owns?

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Thanks John,

yes, the HQ WAN and the branch WAN connection share the same subnet. The DS3 link is one subnet and the T1link is another.

I don't see an option to configure and address on the T1:

WM-1002-WAN1(config-controller)#?

Controller configuration commands:

  bert            Start BERT test

  cablelength     Specify the cable length for a DS1 link

  channel-group   Specify the timeslots to channel-group mapping for an

                  interface

  clock           Specify the clock source for a DS1 link

  default         Set a command to its defaults

  description     Controller specific description

  exit            Exit from controller configuration mode

  fdl             Specify the FDL standard for a DS1 data link

  framing         Specify the type of Framing on a DS1 link

  help            Description of the interactive help system

  linecode        Specify the line encoding method for a DS1 link

  logging-events  LOG T1 Events

  loopback        Put the entire T1 line into loopback

  no              Negate a command or set its defaults

  shutdown        Shut down a DS1 link (send Blue Alarm)

  yellow          Yellow Alarm Configuration

Thank you

[John Blakley]

Pat,

You'd configure your address under the standard interface now. Usually with those cards, you won't see the standard interfaces until you've done the "card type" command as you've done. Now you should be able to do a "show ip int brie" and see the interface s0/2/0.

To add the address just do the norm as in:

int s0/2/0

ip address x.x.x.x x.x.x.x

If you share the same subnet on the other end, I have a feeling that you'd see the interface go down on the HQ side if you were to shut the other side. If that's the case, you could still go with the backup interface command or just tweak eigrp.

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

So I still need to tweak EIGRP? Won't the metrics decide who is the better route by default and only use the T1 when the DS3 route is down?

So, I should see both routes with the sh ip eigrp topology but only the DS3 with the sh ip route, correct?

thank you.

[John Blakley]

So, I should see both routes with the sh ip eigrp topology but only the DS3 with the sh ip route, correct?

Technically, yes The metrics should be much preferred for the DS3 over the T1. If that's the case, you may not need to do anything at all, but it all depends on what the routing table says.

And as a side note, if in fact you shut down the other side and you see the HQ side go to a up/down state, then EIGRP would fail over to the T1 if it does have a route in the topology table to fail over to. Otherwise, eigrp will query for the route and the failover won't be as fast as if there were 2 routes in the topology table.

HTH,
John

*** Please rate all useful posts ***

[Patrick McHenry]

Thanks John,

But I still need a floating default route, correct?

Is this a fairly simple config. Do you have a good doc?

Thank you