Monday, July 29, 2013

CCNA review update and roadmap

I have my test scheduled and I'm pushing toward the goal now.  My plan for the next few weeks is to:

  1. Lab exercises daily on GNS3 for routers and within my switching lab for L2 topics
  2. Daily review and update of my blog reviews to solidify those areas of study
  3. Extra concentration on areas where I lacked, including but not limited to:
    1. Switching:  STP, RSTP+
    2. Switching:  VTP
    3. Routing protocols
      1. OSPF
      2. EIGRP
    4. IPv6
    5. WAN protocols
    6. NAT
    7. WiFi
  4. Test-taking process and review
It will be somewhat challenging to keep focused, but I'm up for the challenge. I have several other technologies that are requiring my attention at work that I need to spend some serious concentration on currently that I'm intending to blog about in the coming weeks following my exam, also.

I have been working with EMC VNX SANs for about 6 months now.  This is a topic that will likely garner much of my attention and study in the coming months as deployments become more complex and the requirements grow in scope.  As these deployments grow from simple, direct-attached Fibre Channel installations to FCoE deployments with varying methods of replication and file/block services, I have much in which to be competent.  Since I'll be studying the technology, I will likely map that to a certification.  Stay tuned...

I am needing to certify on Fortigate firewalls, since I am going to be serving as a product specialist in that line for my company.  Stay tuned for Forti-stuff...

No rest for the network engineer, apparently.

Wednesday, July 24, 2013

Frame Relay - Address Mapping

Address Mapping in Frame Relay Circuits

A router needs a method to map a DLCI to an IP address, similar to the way ARP maps a MAC address to an IP address.  This is accomplished in two ways:

  1. Inverse ARP - 
    • Default config for Frame Relay
    • provider uses LMI to send a list of DLCIs to the local router, and the local router sends Inverse ARP requests to all other DLCIs, asking for their IP address.  All DLCIs respond with IP address.  
    • If you have multiple DLCIs, your interface becomes a multipoint interface, however - Inverse ARP does not work with subinterfaces
  2. Static Mapping
    • Complete control, but more config
    • Assign IP addresses to DLCIs
    • R1(config-if)#frame-relay map ip 192.168.5.2 503 broadcast 

Frame Relay - NBMA, Split-horizon and Subinterfaces

A little more info regarding frame relay circuits:

Frame Relay is a NBMA (Non-broadcast Multi Access) network, meaning that it does not allow broadcast traffic between devices.  This will cause a problem if you use distance-vector routing protocols such as RIP, because the split-horizon loop avoidance mechanism will prevent updates from going back out the interface on which they came.  In a hub-and-spoke configuration, the spoke offices will not get updates about each other.

To avoid the split-horizon issues created in this instance, one could disable split-horizon and rely on the other loop-avoidance mechanisms (not recommended) or use subinterfaces.

Subinterfaces are logical interfaces attached to a single physical interface.  The router sees and treats them as separate interfaces, however.

Multipoint subinterfaces do not solve the split-horizon issues and are essentially useless in Fream Relay circuits.  They simply add multipoint interfaces to a logical interface, recreating all the same issues as a physical subinterface.  Meh...

Point-to-point subinterfaces allow a single, physical interface to hold multiple ports.  They are created using a "dot" - instead of Serial 0/0 we now have Serial 0/0.10.  Each subinterface can connect a PVC of its own, and the router treats each like a different port.




Tuesday, July 23, 2013

Frame Relay 101

It does seem to be a little silly to be reviewing Frame Relay in such depth for the 640-802 CCNA exam, but I suppose that's one of the reasons Cisco is retiring the exam and old guys like me are cramming to get this taken.  Anyhow...

Frame Relay is a technology that is packet-switched instead of cell-switched like ATM.  The fundamentals are that each router on either end of the Frame Relay circuit connects to the provider network, and there is a virtual circuit inside the provider network that makes those routers think they are directly connected.  Doing a traceroute from one end of the virtual circuit to another shows only one hop, even though there are undoubtedly several devices within the provider network the packet traverses.

The 3 Major Designs are:
  1. Hub and Spoke - all branch offices connect to a central site.  The main benefit is cost savings, while the main sacrifices are redundancy and tandem switching (routing all traffic through the main hub)
  2. Partial Mesh - key sites have multiple virtual circuit connections
  3. Full Mesh - all sites connected to every other site, which is expensive.
 PVC (permanent virtual circuit) is an always-on virtual circuit, no set up time

SVC (switched virtual circuit) is an on-demand virtual circuit

LMI (Local Management Interface) is the language used for the provider to send status info about the state of the virtual circuit to the router

DLCI - because serial interfaces have no MAC address, they use a Data Link Connection Identifier to determine which interface with which to communicate.  DLCI numbers are locally significant in that, when traffic is destined for a remote IP address, it goes out a local DLCI instead of heading to a destination DLCI like in the case of using a MAC address. 

CIR (Committed Information Rate) is the minimum speed the provider promises to you at all times.  Each PVC has its own CIR, and anything over that minimum rate is marked "De" or discard eligible - which means it can be dropped if there is not sufficient bandwidth.

BECN (Backward Explicit Congestion Notice) alerts the sending device that it should slow down.  BECN goes away from the congested resource to the sending resources

FECN (Forward Explicit Congestion Notice) is data used to generate traffic back to the sender to tell it to slow down, and is sent to the receiving device by the FR provider.

PPP authentication and compression

PPP is a simple encapsulation to configure on Cisco routers, with the ability to add multiple features once enabled.  TO start, the only configuration necessary is to enter into the interface configuration mode and type:

R1(config-if)#encapsulation ppp

Do that on both sides of your WAN link and you're good to go.

The CCNA exam requires feature configuration for PPP links, however, and mainly CHAP authentication.  CHAP is an authentication method that sends MD5 hashes of a shared password between configured peers.  The configuration steps are:

  1. Configure a hostname for both routers
  2. Create a user on each router with the hostname of the peer, being certain the hostname is case-correct.  Also be certain that the password is identical for both peers.  As stated above, username and password are both case sensitive
  3. Enable ppp encapsulation on the interfaces of both peers
  4. Enable CHAP authentication on both peers
Voila, that's it.  Before you enable authentication on both sides, a debug will show:

 R2(config)#
*Mar  1 00:09:57.055: Se0/0 PPP: Authorization required
*Mar  1 00:09:57.075: Se0/0 PPP: No authorization without authentication
*Mar  1 00:09:57.075: Se0/0 CHAP: I CHALLENGE id 1 len 23 from "R1"
*Mar  1 00:09:57.091: Se0/0 CHAP: Unable to authenticate for peer
*Mar  1 00:09:58.051: %LINEPROTO-5-UPDOWN: Line protocol on Interface Serial0/0, changed state to down


And the other troubleshooting step would be to show the brief interface statistics:

R2#sho ip int brie
Interface       IP-Address   OK? Method Status      Protocol
FastEthernet0/0 10.2.1.1        YES manual up       up 
Serial0/0       192.168.2.2     YES manual up       down
FastEthernet0/1 10.20.1.1       YES manual up       up 
Serial0/1       192.168.3.1     YES manual up       up 


Here you see on Serial0/0 that we have a status "up" and Protocol "down" which is the classic case of layer 1 being satisfactory but layer 2 is the problem.  This is always an issue with the L2 protocol, whether it is a mis-matched encapsulation type or misconfigured authentication or compression.

Compression

Configuring compression on a PPP connection is quite simple, also.  The caveat is that you need to do it on both ends or the link will once again fail (you will see this with an up/down status on the connected interfaces).  

Monday, July 22, 2013

Random thoughts on WAN connectivity

Reviewing WAN connectivity this morning.  Here are some items of interest to me...

T1 interfaces use an RJ-48 connector.  While looking like an RJ-45, the RJ-48 uses STP cabling, a different pinout and voltages.  Don't be fooled!


PPP replaced SLIP and offers encrypted authentication and support for protocols other than TCP/IP.

ATM is cell switched while Frame Relay is packet switched.

PPP Sub-layers:
  1. ISO HDLC:  common ground for devices using PPP, enables base features of PPP
  2. LCP (Link Control Protocol):  Where all PPP features are enabled.  These are:
    • Authentication - PAP or CHAP
    • Compression
      • Stacker - replace continuous streams with codes
      • Predicter - replace data with indexed codes
    • Multilink - bundle multiple lines into a single management, load-balanced line
    • Callback - security method to dial remote users back at a predefined number
  3. NCP (Network Control Protocol):  common interface for L3 protocols

Friday, July 19, 2013

Random items of review regarding Access Control Lists

Just a few things I picked up yesterday that I wanted to jot down before I forget again:

  1. Fast path switching in NAT/PAT is indicated by an asterisk when running a show ip nat translations command.  This has nothing to do with NAT, but indicates that the switching was done based on the initial packet asks for a port and subsequent packets use that same port.
  2. access-group is used in applying ACLs to an interface, access-class is used in applying ACLs to lines
  3. Dynamic Access lists allow a router to create an ACL based on user credentials.
  4. Limit the number of telnet sessions in the line command, such as line vty 1 to limit to a single session
  5. Static NAT allows connections from the outside and are always in the NAT table
  6. Always pay attention to the source and destination in an ACL, especially where ports are concerned - do not limit based on source port, but rather on destination.
  7. Extended ACLs should be placed on the interface nearest the source, whereas standard ACLs should be placed nearest the destination.

More thoughts as they arrive...