jlazzaro/spidey - yes, I've been burned by 2900/3500/3550 and various XL switches. The 3750s are WAY better than their previous products. It's one of the few Cisco switches I would be comfortable recommending. The only caveat I know of is that they don't do well if you turn on a lot of features - but nobody really does.
xSauronx, obviously you have to work with the opportunities you have available to you. But when Dilbert starts hitting too close to home, it's time to look around. Dumb management can sometimes sink a company quick, but more often tends to sink a company *very slowly*. Remeber that time is one thing you can't make more of and you can't get back.
So I'd suggest that you learn as much as you can where you are - and even dumb behaviors are a learning experience, often times learning what *not* to do is more important than learning what to do. A lot of the experienced insight around here comes from where people got burned and had to learn things the hard way. Unfortunately, in this business, experience = scars. New folks believe things like vendor claims and management ideas.
spidey07,
>cmetz likes rub my craw, but he's dead on. Just different worlds.
I'd like to think that I occasionally annoy everyone around here I play Devil's Advocate a lot, but that's partly because I'm not generally one to say "me too" - if you or somebody else already gave an answer I agree with, you won't usually see much from me. But when I disagree, then you'll be hearing the contrasting opinion. And after all, a lot of things here aren't science, they're opinion based on past experience. We give folks the advice and they have to figure out for themselves what to do with it.
>My little rant was more about the fact that a network doesn't go down. It just doesn't happen. If it does then it is bad design.
"Down" is a bad term. When people tell me "the network is down" I just want to smack them. Because that statement can mean such a broad variety of things that it doesn't tell me anything useful. I can't tell you how many times somebody's particular desktop PC has had a problem and that translated to "the network is down," or some daemon on some server wasn't running and "the network is down."
Now, if you're talking total, catastophic, it all melted down and no bits go anywhere kind of outage, I agree with you - that means somebody screwed up very bad, or was seriously, seriously cheap, or we're into "Acts of God" outage territory.
When you're building any complex system, like a network, you have to make a bunch of trade-offs, and one of them is between risk, robustness, complexity, and cost. I can build a network that will not fail to move bits between point A and point B unless a long list of major catastrophes happen. But that costs more than most people are willing to pay, and requires competent upkeep. If the budget doesn't support doing things the most reliable way, it's not incompetent design to do things the best you can, that's a deliberate design decision driven by limited budget.
And unfortunately, in the small business arena, budget usually precludes doing the best thing possible by a lot.
All that said, I still can't build a castle on a swamp. Give me buggy gear supported by guys in India who read from a FAQ, and there's only so high you can set the bar. Even the enterprise vendors have bugs and screw up - hence any really reliability sensitive network needs to have vendor redundnancy too. But that's really expensive and one of the first recommendations I see management usually cut out. Even the telcos are moving away from dual-source.
>On the topic of Cisco/foundry/extreme - I bid large networks out to all of them all the time. The total bid price is ALWAYS within 5% of each other. So stop calling Cisco expensive - capable switches with comparable features from each of these guys are going to be around the same total solution price.
My experience is that these guys will usually match price points with products that have similar enough on-paper capabilities. My experience is that the gap between on-paper capabilities and actual capabilities is narrow in the Extreme gear, wider with Foundry, and huge with Cisco. So the reason I call Cisco expensive relative to those two competitors is not because they charge more money, it's that they deliver half the performance.
(Aside: if I seem like an Extreme and Juniper fan, it's mostly because I can read their published specs and design based on them and not get burned. In Cisco land, I dare not do such a thing, I must lab test and measure, because their published specs are that far different than testable reality.)
Also, frankly, many Cisco products have just plain sucked for a long time. The pre-3750 switches are pretty much junk. I still think the 6500s are junk but they have made dramatic progress - with the sup720 and newer line cards it's a lot better than previous guts. Extreme and Foundry simply would not exist if Cisco was able to put out decent switches between about 1993 and today. At the same price point and the same capability level, you and I both know that most companies would buy Cisco over some start-up. Extreme and Foundry exist because Cisco kept releasing switches that truly sucked, could hardly get VLANs right, dropped packets like crazy sometimes for no good reason, had that awful CatOS for too many years, and implemented features in software that competitors had in hardware.
Now Cisco is catching up, and with their latest generation products Extreme and Foundry are flailing badly. Force10 shouldn't even exist except that all three let the opportunity exist.
At the lower end, compare a Cisco switch and a SMC switch. If you don't need the features of the Cisco product (granted, they have a LOT more enterprise features, but most people don't use them), the SMC switch is often a fifth to a tenth of the price for delivering the same level of performance within the features you actually use. For example, in OP's case, he could have a 48 10/100 + 2 10/100/1000 switch from SMC for about $600, or a 3750-48TS 48 10/100 + 4 10/100/1000 switch from Cisco for about $5000.
Now, you and I both know that I'm comparing apples and oranges; the 3750 is a way more capable device. But if the OP doesn't need any of the added capabilites, why should he pay 8x more? His needs are for an apple, so I'm pointing him at an apple. The more expensive device is overkill.
Cisco does have a new "Catalyst express" line that is much more reasonably priced, but it's still about double the SMC offerring. I haven't yet had the opportunity to disassemble any Cat. express boxes, but I would not at all be surprised if the guts look about the same. So is it worth paying a factor of 2x for a little bit better software modulo deliberate Cisco cripples, a 90-day warranty instead of a "limited lifetime" warranty, and the warm fuzzy brand name? Not many people think so - I don't see any Cat. express switches in the wild and the Cisco SEs I work with don't really know much about the boxes.
p0lar,
>I constantly run into this, people wanting to save $500 here or there without fully understanding the compromises being made.
Now this is true. No matter what design choices you make, you need to understand them. Too many people are just plain clueless and reckless, and that's where the trouble starts.
Was the "$300 Dell" a 27xx? Those boxes are seriously broken...