Wednesday, July 8, 2009

Gonna Need a Bigger Boat: The Fish MPLS-TE Learning Example


Basics
The FISH topology is a widely used tool for understanding the power of MPLS Traffic-Engineering. The principal is that traffic originating in A or B will normally take the shortest path via the route through C-D-G, assuming all circuits have the same BW/Cost. The exercise is to utilize the available, and typically idle, BW capabilities of the C-E-F-G path.

Specifics
The process to activate MPLS-TE is seemingly long, but not overly complex.

Prep the environment:
The first step is to make sure that MPLS is running on your network with full reachability provided by either OSPF or IS-IS. Most basic config to accomplish this is config on each router:


Router(config)#router ospf 100
Router(config-router)#router-id w.x.y.z
Router(config-router)#network 0.0.0.0 255.255.255.255 area 0
Router(config-router)#exit
Router(config)#mpls ip
Router(config)#mpls ldp router-id Loopback0
Router(config)#mpls label protocol ldp
Router(config)#inteface serial 1/0
Router(config-if)#mpls ip


Next, the routers along the path require that MPLS Traffic-Engineering be enabled globally and then on the interfaces. Each interface that has the potential for have a TE Tunnel traverse it will need to have the option activated.

Router(config)#mpls traffic-eng tunnels
Router(config)#interface serial 1/0
Router(config-if)#mpls traffic-eng tunnels


Prep the IGP:
Traffic Engineering information is carried by the IGP routing protocol, in this case OSPF. In this simple example the entire network is in area 0 within the ospf process.

Router(config)#router ospf 100
Router(config-router)#mpls traffic-eng router-id loopback 0
Router(config-router)#mpls traffic-eng area 0

OSPF uses the Opaque LSA's to carry TE information. The IS-IS protocol requires the additional step of using WIDE METRICS in order to carry that information.

Define the tunnel's path:
There are multiple ways in which you can The tunnel starts off similar to a GRE tunnel in terms of configuration, but the similarities beyond the name are few.

RouterC(config)#ip explicit-path name CEFG enable
RouterC(cfg-ip-expl-path)#next-address 12.12.23.3
RouterC(cfg-ip-expl-path)#next-address 12.12.34.4
RouterC(cfg-ip-expl-path)#next-address 12.12.46.6

Build the tunnel:
The tunnel starts off similar to a GRE tunnel in terms of configuration, but the similarities beyond the name are few.

Router(config)#interface Tunnel1
Router(config-if)#bandwidth 100000
Router(config-if)#ip unnumbered Loopback0
Router(config-if)#mpls ip
Router(config-if)#tunnel destination 12.12.6.6
Router(config-if)#tunnel mode mpls traffic-eng
Router(config-if)#tunnel mpls traffic-eng path-option 10 explicit name CEFG


Get traffic into the tunnel:
There are three ways of getting traffic into the tunnel:
  • Policy Based Routing
  • Autoroute Announce
  • Static Routes

For simplicity we will allow traffic from A to route normally and use policy-based routing to forward traffic from B to follow the Traffic-Engineered path over the long path CEFG. Here we will match traffic coming from the loopback of B (12.12.7.7) headed to the loopback of G (12.12.6.6).

RouterC(config)#ip access-list extended TRAFFIC-MATCH
RouterC(config-ext-nacl)# permit ip host 12.12.7.7 host 12.12.6.6
!
RouterC(config)#route-map TE-BCEFG permit 10
RouterC(config-route-map)#match ip address TRAFFIC-MATCH
RouterC(config-route-map)#set interface Tunnel1
!
RouterC(config)#interface Serial1/3
RouterC(config-if)#description RouterC link to RouterB
RouterC(config-if)#ip policy route-map TE-BCEFG

Verification:
To verify the tunnel is operational simple show commands provide all the information you need.

Originating router:
RouterC#sho mpls traffic-eng tunnels

Name: RouterC_t1 (Tunnel1) Destination: 12.12.6.6
Status:
Admin: up Oper: up Path: valid Signalling: connected

path option 10, type explicit CEFG (Basis for Setup, path weight 192)

Config Parameters:
Bandwidth: 0 kbps (Global) Priority: 7 7 Affinity: 0x0/0xFFFF
Metric Type: TE (default)
AutoRoute: disabled LockDown: disabled Loadshare: 0 bw-based
auto-bw: disabled

InLabel : -
OutLabel : Serial1/1, 16
RSVP Signalling Info:
Src 12.12.2.2, Dst 12.12.6.6, Tun_Id 1, Tun_Instance 10
RSVP Path Info:
My Address: 12.12.23.2
Explicit Route: 12.12.23.3 12.12.34.4 12.12.46.6 12.12.6.6
Record Route: NONE
Tspec: ave rate=0 kbits, burst=1000 bytes, peak rate=0 kbits
RSVP Resv Info:
Record Route: NONE
Fspec: ave rate=0 kbits, burst=1000 bytes, peak rate=0 kbits
History:
Tunnel:
Time since created: 3 hours, 5 minutes
Time since path change: 3 hours, 5 minutes
Current LSP:
Uptime: 3 hours, 5 minutes
RouterC#


Intermediate router:
RouterC#show mpls traffic-eng tunnel

LSP Tunnel RouterC_t1 is signalled, connection is up
InLabel : Serial1/0, 16
OutLabel : Serial1/1, 16
RSVP Signalling Info:
Src 12.12.2.2, Dst 12.12.6.6, Tun_Id 1, Tun_Instance 10
RSVP Path Info:
My Address: 12.12.34.3
Explicit Route: 12.12.34.4 12.12.46.6 12.12.6.6
Record Route: NONE
Tspec: ave rate=0 kbits, burst=1000 bytes, peak rate=0 kbits
RSVP Resv Info:
Record Route: NONE
Fspec: ave rate=0 kbits, burst=1000 bytes, peak rate=0 kbits


Traceroute from Router A:

RouterA#R1#traceroute ip

Target IP address: 12.12.6.6
Source address: 12.12.1.1
- Truncated -
Type escape sequence to abort.
Tracing the route to 12.12.6.6

1 12.12.12.2 [MPLS: Label 20 Exp 0] 80 msec 20 msec 32 msec
2 12.12.25.5 [MPLS: Label 19 Exp 0] 328 msec 8 msec 16 msec
3 12.12.56.6 28 msec * 60 msec
RouterA#

The path Router A's loopback traffic takes to Router G is via the shortest path. We have not placed the traffic into the TE tunnel thus standard best path routing applies

Traceroute from Router B:
RouterB#traceroute ip

Target IP address: 12.12.6.6
Source address: 12.12.7.7
- Truncated -
Type escape sequence to abort.
Tracing the route to 12.12.6.6

1 12.12.27.2 28 msec 16 msec 8 msec
2 12.12.23.3 [MPLS: Label 16 Exp 0] 176 msec 136 msec 56 msec
3 12.12.34.4 [MPLS: Label 16 Exp 0] 168 msec 28 msec 52 msec
4 12.12.46.6 60 msec * 32 msec
RouterB#

The path that Router B's loopback takes is via the TE tunnel. The traceroute shows the additional hop through the network.

Advanced Topics
I am going to save the more advanced topics for a later post. The more advanced topics I would like to cover include:
  • Priorities - Setup and Hold
  • Dynamic Tunnel Pathing
  • Affinity bits
  • Autoroute Announce
  • Taking suggestions - What would you like to know about

Friday, April 24, 2009

All quiet on the western front

So I have been away for a while. Been on twitter some. Work projects have zapped 110% of my time for the last couple of months combined with a big after work project that should be wrapping up in the next two weeks. Until then I thank you for your patience and I am looking forward to getting back to everyone with some study material for the Service Provider IE.

In the mean time find me on Twitter(@routetarget) and say hi. Also feel free to @routetarget with any technologies for the SP lab that you would like for me to take a stab at for a blog post.

Tuesday, January 6, 2009

Vendor Neutrality: Gotta get 'em all

I am now in possession of the two of the big three IE training companies workbooks and am working on getting the third. I have done a cursory glance through them and will be starting to run through them over the next couple of weeks.

I understand that the groups preparing these docs have to cater to a wide audience, particularly as there is no prerequisite to have R&S knocked out. But seeing as lots of the material in the beginning is fairly basic relative to the R&S, I will review, but generally skip it.

First up is the IP Expert Service Provider Workbook and Proctor Guide. I am skipping through the first couple of chapters and jumping straight to the OSPF section for "Core Routing Protocols".

Cheers,
Route Target

Thursday, December 18, 2008

Busy Week in Review

Study, Study & Panic?
Less than eight weeks out already, could it be that soon? Alas it is, but I have a plan... Not to be confused with having a flag. Also on a better note, I have been reviewing CCIE Dreamer's blog and glad that I am not the only person who has decided to go down the SP path after already having R&S knocked out.

A word on Support, and I don't mean the TAC...
For those of you who are done with all forms of school and are not married, I don't want to hear a peep from you about not having enough time to study. For those of us who are married/decided to go back to school/raising kids or otherwise have family commitments will understand my jubilation at my next remark. My wife has agreed to me becoming the invisible husband for the next two months. I told friends after I passed the R&S and it will be so this time as well, while it is possible to accomplish a CCIE of any track without support from work, I have yet to see a married man or woman who did not have some support from their spouse in accomplishing getting their digits. That being said, I know several people working on getting either their first or additional IE's and each has said the same thing independently, so there has to be some level of truth to it.

Back to the Plan...
I see it as having about five good weekends in addition to evening studies to get the inconsistencies corrected in my mental database. I think that I am about at the 1st Normalized form now.

Sunday, December 14, 2008

New Lab Date Set

Time to get serious about studying again:
Currently scheduled to sit the SP lab in RTP during Feb. So I will be buckling down and taking some long over due weeks off for training purposes at work. As bad as the economy may be, there always seems to be more work.

Tools, tips, and Tannenbaum:
At this point, I have both the CCBOOTCAMP and the IPEXPERT Service Provider Study guides and have loosely mocked the scenario's up with Dynamips. For anyone not familiar with Dynamips and Dynagen for lab prep, do yourself a huge fovaor and take a look. I'll be spending a little more time on getting that prepared in the next couple of days. With the holidays rapidly approaching the intent is to get as much of the basic review done before the end of the year and then January be spent performing mock labs.

Question to you:
Other than the two that I mentioned, does anyone have experience with the SP material from other vedors that you would recommend? While, ideally, this would be my last attempt at taking the SP lab, I may have time to go through one more workbook or guide prior to February. One of my co-workers who is already a multiple, the show-off, is recommending that I look at Internetwork Expert as he has been impressed with their material for other tracks. Any others?

Good luck to all in your studies & heres to less than 60 days and counting,
RouteTarget

Sunday, November 16, 2008

The beginning can be a difficult thing...

Welcome one and all!
So this is the beginning.  Not the easiest things to do, but will see what we can do on that front anyway.  

Does the world really need another blog?  I am gonna have to say yes.  The great thing about the internet is that there is something out there for just about everyone.  If you happen to discover that it is not out there for you, well then... Go create it yourself.

So here is what I wanted to do with this blog:
1) Focus on CCIE Service Provider issues.
2) Help focus myself for an attempt at the same.
3) See where I can try to help the community as a whole.  

I have taken some inspiration from a couple of the other CCIE bloggers out there and figured that if I could help others in the same way, well it will have served #3 above.

To be Anonymous or not?  I am going to say yes for the moment and here is why:
I really want to stay impartial to hardware, software as well as training material out there.  There is no totally pure Cisco, Juniper, Nortel, Foundry, F5, etc... shop out there.  That being said, I am going to do my best to remain impartial.  We all have to deal with other carriers, vendors, customers and the like.  I want to try and remain honest to those I serve and that I rely upon to get my job done.

I am open to comments and will see what I can do to respond threads when appropriate.

Thanks and look forward to seeing what we can do to have more people get a second, or maybe even third, CCIE.

Route Target