CCDA models

Hierarchical models allow you to.
save cost
Make it easier to understand
Modular network growth
Improved fault isolation.





core
* fast
* highly reliable
* redundant
* fault tolerant
* Low latency
* Avoid CPU intensive and classification
* QoS   using the classification done by other devices.
* limited diameter hop

Distribution
*aggregate
* filter
*route

Access
* classify QoS
* Rate limit
* arp inspection
* PoE
*  Trust


In this model, the Distribution handles the RVI or IP of the Vlan and will route between vlans.

Because of STP one link is disabled from the access to the distribution.
So you are paying a load of money for a "backup link"



If you drop the Layer 3 border to the Access layer.

Then you will gain more links.
However the Layer 3 licenses cost more.
You will also be limiting the Vlans to be local to the switch.


Cisco 6500 VSS- supervisor 720-10G
can run VSS.

VSS is proprietary to Cisco and makes the switches pretend to be one switch.
Must use VSS capable Sup720 or SUP-2t.




The above was the simple model.
This is the Enterprise Architecture Model.

Let's have a look.
Basically the old model was the Core---<>--- Distribution----<>---- Access.
Since now we have a dedicated Datacenter   we can also expand it to
Core----<>---- Datacenter Distribution -----<>----- datacenter access layer

For the WAN  we can also create an Edge Distribution   and   an  Enterprise Edge (access)

So in reality our network looks like 3 triangles

The smaller the network the more layers you can consolidate.



Enterprise Edge.

In this case they split Ecommerce and Internet into two modules.

The E-Commerce will have
Firewalls - obvious
IPS and IPS on switches -  IPS is the new must
Web,application and database -  for Ecommerce


Internet
Will have firewalls
Http servers
Ftp
Email
DNS


WAN
Will have MPLS/Frame/VPN site to site which you can use
to connect branches or to datacenter



Remote Access VPN can use PSTN dial up
or the Internet to allow VPN access for users.











Connecting to your provider.

one router does not provide local router redundancy.

One ISP does not provide  ISP redundancy.


Entyerprise WAN technologies.
MPLS   multiprotocol label switching
Metro Ethernet using QinQ
Leased line
SONET   , SDH
PPP
Frame Relay.
ATM
cable/DSL
Wireless

Cisco says MPLS will map the DSCP QoS marking into the relevant MPLS

Service Provider Edge.
Matches your  Enterprise Edge module.

Enterprise Branch Office.

Usually a router and a switch.

The connection back to the Enterprise Edge can be
1. MPLS using the  WAN
2. VPN  to lower the cost using the INTERNET















Enterprise Data center or Cloud

Teleworker is a very small office or one person.
Usually you can use an ISR  or an ASA to collapse everything into it.

Borderless pillars are.

Access layer High availability.
Workstation to Router
Server redundancy in the FARM
router redundacy within the network compenents
Link media in the access layer.

Workstation path discovery can be using.
ARP  -  find out the gateway
VRRP  HSRP  GLBP  all are the same   two devices create a virtual gateway.
ICMP                 router discovery
VSS  - this is the two chassis that behave as one
Explicit -  I can explicitly configure the gateway manually on the NIC>
RIP   -   yes you can run RIP on the Server ?? not the smartest



HSRP

VRRP is the same thing as HSRP  except it is not proprietary.



GLBP is the Cisco improvement.
It allows BOTH links to stay active thereby increasing the available bandwidth.
Load Sharing
Multiple Routers      up  to  1024   Virtual Groups   on each interface.
You an PREempt a router with a higher priority.
Authentication can be used.


Server redundancy
Use
FEC and GEC     etherchannel bundles.
Use Dual NICs


Load balancing in Cisco can be changed by.
maximum-paths   this will change the number of links the router will use from the default of 4.
Maximum is 6.
All links must have the same cost in order to load balance.

EIGRP can use Variance to load balance   (bullshit)

Process switching     load balances packet by packet.

Netflow Switching load balance by destination   which is much better as the packets   will not have a chance
to come out of order.

Mesh is   N(N-1)/2
Full mesh
Keep broadcast at less than 20%

Make sure a Backup redundant link is not using the same last mile circuitry.

So
Workstation to Router      use   HSRP,VRRP, VSS
Server redundancy      use    daul attach NIC ,  FEC,  GEC bundles
Route Redundancy       use the     load balancing and high availability of more than 4 routes
Link Redundancy     use     a variety of WAN.