-
Simply replacing the hubs with switches is a bandaid approach that will once again fall off and require another bandaid.
If all they want is a simple change of hardware, then sell them some switches and be done with it.
But I would not go that route. Not at all.
I would sell them on allowing me to fix the problem once and for all, after a study of traffic volume and priority, security needs and future projected growth needs. Rome wasn't built in a day and neither is a lasting solution.
The best approach to fix their problem is to create a paper plan based on employee gripes, traffic analysis, existing network configuration documents and a couple of building walkarounds with a knowledgable manager. If you want to really fix this problem, you have to get a "lay of the land", ask alot of questions and that means interfacing (connecting) with the people that are gonna live with your solution. Nobody knows the network gripes/needs better than the people who use it.
You use that paper plan to segregate traffic using switches, Vlans, routers with the existing wiring.
It's always a safe bet going that:
1) Too many employees are Internet surfing
2) That alot of traffic comes from Spyware/Adware/Non-work related...
3) That the network was not properly planned and documented (think congestion points)
4) That if they keep bandaid fixing it, one day it'll simply implode (figuratively speaking)
5) They have no idea of the costs incurred by company downtime based on a faulty network infrastructure.
Before you give them a proposal, I'd do my best to convince them a network analysis is the only way to go, before equipment is put into place.
Remember if you sell them a high dollar solution that later needs revamping, customer satisfaction and trust goes down the tubes.
Just my quick thoughts.
-
This question is strangely familiar to a question in the CCNA Lab companion workbook
-
ZT3000 does make a good point. But right now the OBVIOUS problem is the hardware is not up to par for the work load. Monitoring the network will be MUCH easier once it is segmented correctly and you have proper hardware in place. One that is done you can drop a sniffer box on the network somewhere and monitor each vlan on its own (I said this all in my first reply BTW) Networking monitoring and policy management is a beast on its own, and changing the hardware is far from just a bandaid. Its a pain to use unmanages switches and hubs if you really want to do GOOD monitoring. especially in set up where everything is in one domain. You cant control nearly as much and you cant fix a problem when it is found nearly as easily.
I do fully agree that the company should do a full network analysis/audit before dropping this kind of cash. that way you can get an optimum solution for the problem rather than a generic one.
-
Would anyone recommend any of the cat 29XX or 35xx switches. Or any other specific recommendations
-
the 2900 series are great, we use them alot where i work. they are great mid range switches, we have them so sgment each floor of our main building, and also use them at remote sites often.
-
Ok, like many have said, the prefered setup would be managed switches on the floors, configured with vlans, with a layer 3 core switch to connect all the access switches, and route between vlans. Preferably you'd have gigabit uplinks; while fiber is nice but pricy, unless you have really long distances to go, good quality Cat5e or Cat6 cabling for these would do fine.
That being said, I'm most certainly sure that the suggestion of 4/5/6000 series is WAY overkill and WAY over budget (concidering they're still running hubs!)... I myself have used a catalyst 3550 series which are featurefull but much less pricy. And while the cisco brand is nice and sounds leet, consider alternatives like the HP procurves (1/4th the price of ciscos per port) or Foundry Networks for "brand" alternatives or even d-links (which work nicely with a more than sufficiant featureset without the brand price) if money doesn't allow.
Ammo
-
D-Link sucks... cisco is prcey but is also the industry standard. 3com does make some cheaper alternatives, and we use those also, specifcly the superstack series for lower end work. The reason I suggested the cat5000 series is becasue it leave you alot of room for expansion, but the 3000 series would be sufficient.
-
The cat5000 series is a bit out of the price range i was after. Thanks for the suggestions guys. I'm sort of comparing prices at the moment. What would be a decent price for a 29XX 48 port switch?
-
Id check ebay, we have been very successful in picking up cheap gear there among other places.
-
New equipment is all over the net. Throw a stone and you hit 25 resellers and the rebound get's the used/auction sellers.
I prefer established sites that have a reputation for dealing with corporate clients who demand more from their vendors...the problem is you usually don't get the rock bottom price from these sites. http://www.cdw.com/ is a perfect example.
There are a LOT of good suggestions here. ZT3k makes a good counterpoint, but you really need to think about your audience. If you are in the unfortunate position of being the IT monkey aka digital janitor aka tech dude who we hired as an intern and won't go away... Anyway, if you are in a position like this, they probably aren't willing to listen to a lot of input on the results of your usage analysis and long term reccomendations. It's the right way to go, but if it's not what they want to hear, and you don't have the ability to change their mind, then SirDice has given you some clear cut examples of what to consider. You definitely want switches on each floor, with vlan'd segments and top-of-the-tree switches/routers at your PoP. Servers shoudl absolutely be on a separate segment, and while you're at it you should consider where you would place sensors and such (if you don't currently have IDS in place.)
Best of luck, and let us know how things go.
