Archive

Archive for February, 2010

EIGRP – the variance command

February 27, 2010 8 comments

I was reading the variance concept again as I didn’t feel comfortable with my level of understanding. It is now much clearer to me and I want to share my notes.

The variance command is used to balance traffic across multiple routes that have different metric values. If we have one link with lower metric and as a result it is being over utilized while other connections are idle, we can load balance using the variance command.

Router(config-router)# variance multiplier

The multiplier value, from 1 to 128 is used for load balancing. The default is 1 and it indicates equal-cost for load balancing, as a result path is chosen by metric values.
Multiplier define the range of metric values that the EIGRP process allow for load balancing.

variance command - example

variance command - example

In this example these are the FD values:
B -> 30
C -> 20
D -> 45

By default, when variance=1 the only path used is via router C, because the lowest FD is 20 and only this path have this low metric.

By configuring variance 2 the required FD is 2*20 (=40) and the path via router B is now less than the requirement and can be used to load balance traffic from router E to network Z.

This is an easy to configure load balancing tweak and a good point to remember for the exam.

the mother of all Snowmageddons

February 26, 2010 1 comment

Rofi Neron is covered with snow and waiting to be plowed

snow. again.

snow. again.

This record braking storm bring us a record snowfall of 9.4 inches at Central Park NY today (This breaks the old record of 7.5 set in 1874), An all time monthly record snowfall of 35.9 inches at Central Park NY (breaks the old record was 30.5 inches set back in March 1896), The first Triple-double (3 storms with at least 10″) and we’re about 10 inch from the snowiest ever storm in NYC history!

And if you think this is impressive, situation in Brooklyn is much worse…

Routing Your Commute

February 24, 2010 1 comment

When you live in a Subway city like NYC every morning is full of routing decisions.

Most lines have three areas in their AS:  Brooklyn, Manhattan and either Queens or The Bronx. Our four autonomous systems.
These four autonomous systems combine hundreds of miles (distance),  stops (hop count) and cars (packets). There are also millions of commuters (is it broadcast or many-to-many multicast?) that need to be transferred from one station (router) to another.

Like every good network, the MTA system have multiple backups and while some trains run on the low-cost 56K local lines, others have T3 segments on the express line.

Lucky me, my train uses the local track and make all stops. If it was a RIP system I would definitely have lost my way as I have more than the allowed 15 hops…

This morning, while riding to work and reading my (endless) OSPF notes, I had a blurry moment when my notes came to life. My F train stopped at Broadway-Lafayette station where two of its express mates have a stop. Whenever I get to this station I look across the platform and check if the express train is there. If I see it, I usually hop over and save myself 6-7 minutes – eternity in morning commute time and priceless in routing terms.

For a reason I can’t explain, maybe because I woke up on time and had enough time, a good sit and interesting OSPF reading, I’ve decided that the express will do without me. I stayed on my train to discover that it is waiting and waiting. Finally another express train stopped and this time I had no doubt  in my mind, I crossed the platform and waited for the doors to close. To my surprise the local train left and we didn’t move for few more minutes.

Eventually the train moved and used the local line (which explained the delays), I was late as usual and left with great admiration to those routing protocols – how are they doing it all day, every day and don’t get tired???

NYC Subway - LED panelAnd while at this subject,
am I the only one thinking the new LED screens on the latest trains is one big routing table?

BSCI – Cisco IOS DHCP server

February 23, 2010 Leave a comment

One of the topics Cisco removed from the ROUTE exam is DHCP;  it’s about time they fix this weird decision. Why it is on the BSCI to begin with is a mystery but it does not change the fact that I have to study this topic.
Here are few notes about using Cisco IOS as DHCP server:

There are three DHCP assignment methods:
Dynamic
Automatic
Manual

There are two binding methods:
Manual binding – MAC address to IP address
Automatic binding – not kept on the DHCP server (router), usually kept on TFTP server
DHCP database agent (TFTP) is configured using ip dhcp database command

When you enable the DHCP service using the service dhcp command, DHCP service uses the following packet types:
DHCPDiscover ->broadcast from the client (always the same type)
DHCPOffer ->DHCP server send back IP address and parameters (coming from IOS)
DHCPRequest ->client say which IP address it accepted (choose from all the responses)
DHCPAck ->DHCP server approve the IP assignment and provide detailed parameters

DHCP discovery requests are broadcast and we know that routers block it. To bypass this router limitation the command ip helper-address is used.
ip helper-address
translate broadcast to multicast to allow traffic via router, it should be configured on the interface that receive the broadcast.
This is a useful command out of the DHCP world so keep it in mind!

This is a basic example of DHCP configuration (include basic DHCP values):
ip dhcp pool IOS_DHCP_SERVICE
network 10.1.1.0 255.255.255.0
domain-name mydomain.com
ip dhcp excluded-address 10.1.1.1 10.1.1.19
dns-server 10.1.1.10 10.1.1.11
netbios-name-server 10.1.1.10
default-router 10.1.1.1

If we want an interface (DSL or Cable) to receive a DHCP address:
ip address dhcp

Other than the extra overhead on the router, the main downside for using a router as DHCP server is the day-to-day management. I do not think many people would argue that using Microsoft’s DHCP server is much easier. Microsoft servers present an easy to use interface that provides information on all the clients and Cisco does not. The following command allow you to export the binding information from the Cisco router to a TFTP server as a text file:
ip dhcp database tftp://10.1.1.15/dhcp-bindings.txt

I hope my exam will have basics only on DHCP and as long as they keep in the above scope, I think I’ll do just fine.

OFF TOPIC – Dante’s Internet

February 22, 2010 1 comment

Paul Tassi of unrealitymag.com published this hilarious post that every blogger can identify with. This is the center of the post:

BSCI – Multicast notes

February 22, 2010 1 comment

I was going over few of my multicast notes today.

Multicast is a smart(er) way to send the same traffic to many recipients (taking over Broadcast in IPv6). I’m not sure how big it is in the exam but here are few notes that I find important.

There are three ways to stream many receivers:
Unicast ->one to one
Broadcast->one to all
Multicast ->one to many

Multicast uses UDP and a Class D range of 224.0.0.0 – 239.255.255.255 and this is how the range split:
Globally Scoped – addresses 224.0.1.0 through 238.255.255.255 (Internet)
Limited Scope – addresses 239.0.0.0 through 239.255.255.255 (local organization\group)
Private Range – addresses 224.0.0.0 through 224.0.0.255 (used on local LAN)

IGMP (Internet Group Management Protocol) is a protocol used for multicasting, allowing clients to subscribe to multicast stream.
There are 3 versions of IGMP:
IGMPv1 – when a client stop listening the router still send stream
IGMPv2 – host send leave message when it doesn’t want to get stream
IGMPv3 – host send join request for the stream on the server

There are three group related concepts:
1. Router does not have to be a member of a multicast group to send stream to the group
2. Router must be a member of a multicast group to receive multicast data
3. If a packet is sent to a multicast group address, all members of the multicast group will receive it

PIM (Protocol Independent Mode) is configured on the router and set the way it will act when multicast stream will be presented:
PIM Dense mode – using flood and prune
PIM Spare mode – shared tree protocol, specify RP and all other routers access it
RP (Rendezvous Point) is a router elected to centralize multicasting in the network

I want to recommend some extra reading, though not CCNP related it has everything for the exam.

And as I always ask, if you have any comments or find mistakes don’t keep it to yourself.

TSHOOT exam topology

February 21, 2010 Leave a comment

By now you already know that CCNP has a new format. You read all the details here few weeks ago.

Jeremy Gaddis of evilrouters.net posted today a summary of the new CCNP TSHOOT exam topology that Philip Vancil, one of the exam developers from Cisco posted on CLN.

Read Jeremy’s post for more details, topology maps and topology document.