Netowrk flow very slow ,collision / broadcast domains

i will give you more info next week about the network , i m not yet there , but i got a draft of what's going on there no details . but typically , there's a federator switch ( cisco ) and it"s a flat network with no subnets . i m looking just for a desing to solve this kind of situation before i ll get there and get hands dirty :).

hi all , sorry for the late .
here's some details about the network . there about 700 users over a 9 floors bluiding .each floor have access switchs cascaded (each one is connect to the next switch ) till the last switch connected to both federator switches configured in active/passive mode .
they want to implement Vlans ( about 10 vlans ) with 802.1X authentication .with a dHCP server .to make the network more manageable  and optimised .
also they want to connect the servers farm to the federator switchs and make them works in activ/active mode .
this is a draft of what i proposed  based on entreprise network desing standards , i m open to any other suggestion .
thanks .


 

If the switches are currently daisy chained, then that likely explains the network slowness and collisions.

thanks for all suggestions ,
did the servers really need to be in a vlan ? i m thinking about letting them in the native vlan on the switch .

@ r0ssc0: the routing will be in the core , since the edge switch are only L2 , what 's the need for vrrp you mention ?  i think stp is enough isnt ?

actually the switch at the edge are chained ,we wan to unchaine them to optimize the desing

It's a good practice to put servers in their own VLAN so that later on, you can add ACL's or QOS easily.

you are correct you dont need to use vrrp and could  just use a routing instance matched to the root of each vlans  spaaning tree.  Best practice however in this type of network would be to use the dual cores for reduncacy  and deploy vrrp so that when a core switch goes down the routing instances will failover to the other switch inline with spanning tree.