07-13-2005 01:47 AM - edited 03-02-2019 11:22 PM
I need to understand the pros and cons of using a collapsed core architecture as opposed to separate access,distribution and core ...thnx
07-13-2005 02:36 AM
Pros:
Core, aggregation, and peering layers collapse onto one or few large-scale, unified systems, creating more scalable, more available, and more easily managed Points of Presence (POPs).
Infrastructure convergence is made possible with Layer 2 and Layer 3 services delivered by multiservice edge platforms or directly by the core routing system.
Service convergence (to IP/MPLS) is made feasible by the improvements in network scalability, availability, and service awareness resulting from both the overall POP simplification and the advanced capabilities of the routing systems that make this simplification possible.
Cons:
Availability has been delivered at the cost of redundant routers and interconnecting links, doubling capital costs and the number of elements under management.
POP scalability reaches an upper limit quickly as the number of interconnecting links required to support redundancy and tiering increases.
Scaling link speeds between routers often involves disruptive equipment upgrades, and slots that could be generating revenue are instead used for intra-POP connections.
Network operations and management are made overly complex by the number of individual systems that must be monitored and maintained.
Provisioning is difficult because of the number of routers that must be configured to deploy a new service, particularly where service-level agreements or protection for high-value or real-time traffic must be implemented.
Routing complexity and the number of peers that must be managed by each router increase convergence times, adding to the potential for customer-detected service disruptions caused by router failures, link failures, or operations and maintenance mistakes.
Differences in service awareness and service application capabilities in routers deployed at different tiers can preclude support for real-time services, security, multicast, and traffic prioritization.
Credits: various CCDA/CCDP Ciscopress texts. For details and more, checkout http://www.cisco.com/en/US/products/ps5763/products_white_paper09186a008022da42.shtml#wp1002953.
Josef.
07-13-2005 03:42 AM
hi.thanks alot for ur response.i wanted to know about collapsed cores with regard to local area networks
07-13-2005 02:38 AM
Hi,
We genrally use Collapsed core acchitecture for Small or medium business, where generally you dont have a large traffic flow and you can have your one L3 Switch or router serving the both distributin layer access-policies, QOS, Routing and other services and the Core layer services like traffic shaping, PBR etc.
When desginging the collpased core you have to take traffic, application and traffic patters into account and are desined for small/ mdeius m=busineses as I said. You have to take the number devices which would aggregate on the core depending on the users.You device should be efficeint in processing the traffic when you have a collapsed core design so you have to go through a lot of Docs and other tools to get the most suitable device.
HTH,
-amit singh
07-13-2005 03:48 AM
Hi thanks alot for ur response.Am looking at about 150 users.Can a 2970G-24TS be used as a distribution layer switch in a nw of abt 70 people
07-13-2005 04:49 AM
Hi,
I wouldn't go with a distribution layer for 150 users as its not necessary, you don't need the scalability.
In many designs a distrubution layer is a nice way to aggregate a whole load of access layer switches and perform inter VLAN routing away from the Core, this leaves the Core to deal with fast switching to the server farm. There are many different Cisco designs using Access, Distribution and Core layer which can be see in the following document:
The collapsed core takes away the seperate distribution layer and consolodates interVLAN routing and fast switching to the server farm into one layer.
Unless your company has the potential to grow massively, a collapsed core design will be more than adeque for a network of 150 users, plus you are saving money by not adding an additional layer that in your environment is not necessary.
I would probably look at having 3 x 48 port 3750 switches (WS-C3750-48TS-S or WS-C3750-48PS-S) stacked in the access-layer and 2 x 10/100/1000 3750's (WS-C3750G-24TS-E1U) providing the core and distribution layers.
Each 48 port access switch has 4 SFP GBIC ports and so does the core 3750's so you can create up to an 8 Gbps cross stack etherchannelled uplink between the access and the core layers. You probably won't need anywhere near this much bandwidth but you can calculate what you need by looking at the applications/servers that will be connecting to the Core layer, and any external/internet connections.
HTH
PJD
07-13-2005 06:06 AM
Well, swinging back to the original question: both Collapsed core and the separate layers (or "Enterprise Composite Model" as Cisco prefers to call it) are just different ways of viewing network infrastucture design.
Almost any switch above a 2950 can carry out distribution functions. 3 x acess 3750-48s and 2 x 10/100/1000 3750s for distribution for 70 users will easily hit AU$100,000 by the time you are thru with works. Your manager won't like you ;-)
2 2950-48s (96 users) with 2 Gigabit uplinks and a 3550-24 (or 3560) as a "distributor" is more than enough, giving a total of around AU$15,000 which would make you much more amenable to your manager. All this, of course with the caveat of future expansion: are you a high growth dynamic organisation or are you zero growth Not-for-profit organisation?
As always what you need depends on what you want to do. As you can see, we all have different views, ways of skinning the proverbial cat!
Josef.
07-13-2005 06:22 AM
Blimey are you sure?!?
In the UK these switches will cost approx:
3 x WS-C3750-48TS-S = £6,981.00
2 x WS-C3750G-24TS-E1U = £4,756.00
If its only 70 users then 2 x WS-C3750-48TS-S = £4,654.00
As you can tell i like the 3750's :)
Good Luck
Paddy
07-13-2005 11:12 AM
Thnx alot
07-13-2005 03:30 PM
Yep Paddy,
Those figures were quoted to us by a Cisco Partner (rolling my eyes). Of course there was no way could we have gone with that...
Josef.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide