Best way to connecting switches together to reach the best performance?

we got 3 Floors and almost 11 Switches. we are going to connect these switches so got the best performance. so this is how we got that.

Network Diagram

My Questions are: Is this Plan can get the best performance? Is it different to use 1x48 port Switch instead of 2x24 port switches? if no, Why?

Any Suggestion!?

From serverfault

• No network daigram - so sorry, nothing to see here.

  For best performance in general:

  • Make sure you ahve a central hub with a high end switch, and
  • Make sure that has more bandwidth than the rest.

  If feasible - dont know your switches - connect them in a dual layer ring. Like 11 switches, each 2x10gb.... so two rings of 10gb. Or... each has 2x10gb to central switch.

  n general you reach a limit of how good performance you get with pure switching - which is when routing / layer 3 switches come in handy (like extreme networks).

  But for anything more specific, we need the diagrams AND the perforamnce requirements.

  : sorry! I fixed it

  : We got SMC and D-Link Switches
  From TomTom

• Ok, putting that in another answer...

  • Unless you can run a 10gbit link to the server room down to at least switch 2/4.... I would go with separate 1gbit links from EVERY SWITCH to the central one... and then hope to go 10gibt there. Otherwise you run serious congestion possibilities, at minimum between 7/9 and the server room.
  • I assume you dont have central management and higher level routing (i.e. layer 3) functions - SMC and DLink are pretty cheap budget stuff (use them myself), but heavy lifting is not their strong side.
  • Given the plan there is not a lot you can do. run as many cables as possible, then fix thigns up in the server room. Dual link 1gbit to every switch from a central one, or single / dual 10gbit, otherwise the indicated 7/9 may kill you. You aggregate 7 switches over that link.
  : my problem is that my manager don't pay money for new cabling system

  TomTom : Well, then go 10gbit on the central links. Otherwise you simply are limited. The main trunk line will overload - and basically make everything slow. Cabling should be damn cheap. Most expensive is the time for the people to put it in.

  3dinfluence : If the manager won't pay for cabling...you think he's going to go for 10Gbit which is most likely going to require new switches, 10GB fiber gbics, and running fiber?

  TomTom : Well, so what you ask? Best performance will be crap. You have a network setup where all the switches wont make a difference because you try to get all the traffic through a small little 1gb line... if the traffic overloads that (and it will most likely), that is it. No options without spending money. Btw., 10gb... would bun on the same cables (cat 5). Depending on the switches you may be able to put modules into yours. Whoever did the planning did not care about perforamnce. Well, sorry to say, but you have to live with that. Send your maange the link to this page for explanations.

  TomTom : OR: maybe your users just surf, use office... and the 1gbit line survives ;) But then... given the hugh amount of switches behind that one bad link... it is highly unlikely.

  : :-) Thank you very much TomTom, for your great suggestion, I am going to give it away and give my manager this page :-)

  TomTom : Really do it. Sometimes management wants the impossible. Reality, sadly, does not change ;)

  Chris S : You have to be very careful throwing around phrases like "you need 10GB fiber" without having the slightest clue what kind or the amount of data that's flowing over the network.

  TomTom : Well, first i never told fiber. Second - look at the number of serial switches and end user machines. There is one linke that even at 2gb full duplexwould just be way too pone to overload, pretty much regardless what people do on their workstations - unless it is ONLY surfing around. There are 7 switches connected to that one line. With 24 ports per switch this means 168 machines. Make the math yourself. Plus, as I said - 10gb is available over decent copper.
  From TomTom

• You currently have a Tree topology. You should really stive for a Ring or Star topology to help minimize points of failure. Wikipedia has a decent article to help identify what you're working with.

  For bandwidth and redundancy I would double up the physical links for each logical link (so if you're running a star, two links from each spoke to the hub switches). Make sure you're switches at least support spanning tree protocol if not something better like LACP to aggregrate links.

  If you stick with the tree topo I would highly recommend that any parent "node" be two switches, linked, with redundant connections to the child "nodes". For instance. #5 in your diagram should be two physical switches that are linked to eachother. Then all child nodes 6 and 2/4 should have two physical links each (one link to each of the #5 physical switches).

  You might also want to clarify what you mean by "best preformance". I read that as "highest availability with acceptable bandwidth"; where you might actually mean "best bandwidth, to heck with availability". It's easiest to just list your priorities in order: Availability, Bandwidth, Minimal Latency, Cost, Maintainability.

  I always stick with 24-port (or less where appropraite) switches as the number of failed ports should the switch fail is minimized and the cost difference is typicaly minimal.

  I know it's not what some people like to hear, but avoid the cheap consumer-grade switches like the plague. They simply don't have the features (STP, LACP, Warranty, Management, Configurability, etc) usually and that's the stuff that will bite you hard as your network grows. 3Com, ProCurve, and Cisco all make excelent switches and the cost is almost always justified by time saved screwing with problems.

  : Thanks indeed Chris, my priorities are Bandwidth, Minimal Latency, Cost, Maintainability, Availability. and the best performance is best bandwidth. The question is we want to reduce the cost of network cable by this system too? Is the star topology can get this?

  Joseph Kern : Also remember that you cannot have more than 5 switches chained together. "100m per-segment (Cat 5), no more than 5 segments".

  Chris S : @Joseph: That standard was written a long time ago, mostly for use with hubs. Modern switches alsmot always learn MAC addresses and direct packets along a directed path. Again a good reason to get business-grade equipment, which usually maintains a larger list of MAC endpoints. Also, VLANing by floor or something similar would solve this problem. Or getting stwiches that support medium distance stacking (as many ProCurve units do for instance), so the multiple switches logically act as just one.

  Chris S : @user: assuming you're not planning on growing significantly in the near future, a star topology is probably the best route. If you have a pair of switches with enough switing fabric to support the whole network then use them for your hub. Remember, redundant links are your friend (so long as your switches support them well).
  From Chris S

• A few general rules to follow are:

  1. Use the Three-Layered Hierarchical Model
  2. Avoid daisy chaining switches together if possible
  3. Use port trunking to reduce bottlenecks
  4. Use port turnking and/or RSTP and redundant paths for fault tolerance
  Chris S : The three-layered model is great for large buildings and campus networks, but is generally not ideal for networks under 250 users.
  From ITGuy24