IT folks.....Analyzing network proposal for a new school....(need feedback)....

[redgtxdi]

Ok, so I'm no IT or networking guy so I'll rely heavily on opinions & questions to obtain answers here.

New school, probably starting with a few hundred students first & I don't know the potential limit. However, 3 proposals have been given to network the school as it's being built.

1. Copper Backbone – This deployment relies on CAT5/CAT6 Ethernet connectivity everywhere, and is essentially the same as our previously discussed strategy
2. Fiber Backbone – This deployment assumes fiber uplinks from the MDF to every classroom and the lab, and thus requires a core switch stack in the MDF consisting of a 24-port aggregation layer fiber switch and a 48 port Ethernet switch, as well as fiber capable switches in every classroom.
3. Hybrid Backbone – This deployment is a new idea. If we kept the same five IDFs we had originally planned and added a fiber uplink to each (rather than to every classroom), we could gain many of the benefits of intra-campus fiber without the hefty price tag increase of going to each classroom with fiber. When Jay and I from our office were meeting with you and Jay about the possibility of adding fiber runs for future use, this is what we were talking about, and this deployment would allow us to make use of those from the get go.

Copper = approx $30K
Fiber = approx $50K
Hybrid = approx $35K

Let me know what questions you might have but I'm a fan of copper, period as I think wifi will continue to progress going forward, but I may be naive.

TIA

 

[drebo]

I do extensive consulting for school districts.

#1 is not scalable enough.
#2 is overkill.
#3 is the way to go.

Designate IDFs either per building or per logical grouping of classrooms. Use copper cabling to each class room. Use either 50nm or single mode fiber to uplink to an MDF, and use at least 12 strands.

I assume the numbers you listed are for just cabling and not switches. If it includes the switches, find a new vendor because the switches he's recommending are garbage.

Stay away from Netgear, DLink, HP's low end stuff, Cisco Small Business, etc. You want real equipment. Yes, you'll be paying 3k+ per switch, but you'll thank me later when you build out the wireless network and the students need to use it for common core and SBAC testing. One-to-one and BYOD are a thing now, even in elementary schools, and it's likely to be mandated at state levels eventually.

Schools upgrade their networking equipment roughly once every 10 years, so plan for that.

 

[redgtxdi]

Thanks drebo!! This is what's listed for switches. This is a small, private school (at least to start) so I don't know how that scales to the schools you've worked with but, again, probably several hundred students to start with capacity for....who knows.

 

[drebo]

Oh god, those are terrible switches. Avoid Meraki and Dell PowerConnects at all costs.

I consult for school districts as small as one school with 2 switches and as large as 30+ schools with 35 switches at a single site.

I recommend Cisco Catalyst at all of them. For you, I'd recommend an ASA5512-X, iBoss, a Catalyst 3850 fiber core switch (if you have 4 or fewer fiber runs, you could get away with out this and just use a stack of 2960XR or 3850 copper switches,) and 2960X switches for all access layer switching. I would suggest PoE on the access switching for cameras/PoE (anyone who says an analog camera system is adequate is lying to you...go IP surveylance or just mount dummy cameras.)

Yes, it's expensive, but, again, you're going to need to deploy wireless eventually and you'll need the capability to route between your wireless VLANs and your server VLANs faster than that shitty Meraki Security Gateway can. You're also going to need the capability to controll access to the servers to keep the students from being able to hit things like Aeries. Ruckus wireless would be the solution I'd recommend...it's pretty inexpensive and does everything most schools need. Ubiquiti UniFi might work, too, but the RADIUS authentication is kind of sketch.

If you want to cut costs, buy used Cisco switches.

 

[dave_the_nerd]

Ugh, Cisco fanboys.

Used Cisco = no support contract, still more expensive. (But hey! I know a CCNP you can hire!) It's like buying Mary Kaye.

I would at least look at enterprise or commercial-grade routers/switches thought. (Juniper, for routers, for example.) If it's "Web Managed" and doesn't have a console port, that's usually a sign that it's scaled up consumer hardware.

The higher-end PowerConnects are quite capable. They actually run a licensed Cisco iOS though.

 

[kevnich2]

I really have no problem with my dell powerconnects. I would avoid the 2848 series though. Go with the higher end 5xxx 6xxx or 7xxx. Those have run for years in my offices with minor issues due to electrical and lightning spikes. We keep a few on hand as the cost was much lower than cisco and just swap them out as needed.

Other than that, I'd say drebo is right on line. It will be expensive but with how often things get upgraded, you gotta look out 5-10 years.

 

[redgtxdi]

Thanks gang! That really helps!! Although some of it is foreign to me, I definitely get the gist. I don't think everything needs to be Cisco either as I know here at my work we use quite a bit of Dell product with much success (and reliable success, at that).

I'd guess it's higher end stuff but I won't go dig up whatever it is, I just know it works.

I've edited, cut & pasted some of your comments to reply & see if they have input about it. Just like anything else, I'm sure they've got great Meraki success stories (as well as perhaps the 2848 stuff) but I'd rather clear the air up front.

Go Dell! (yes, sarcasm intended)

 

[drebo]

Ugh, Cisco fanboys.

Used Cisco = no support contract, still more expensive. (But hey! I know a CCNP you can hire!) It's like buying Mary Kaye.

I would at least look at enterprise or commercial-grade routers/switches thought. (Juniper, for routers, for example.) If it's "Web Managed" and doesn't have a console port, that's usually a sign that it's scaled up consumer hardware.

The higher-end PowerConnects are quite capable. They actually run a licensed Cisco iOS though.

A used Catalyst 3750 12-port fiber switch is $500. I could stock 10 spares and still be half the cost of a brand new one.

A 48-port gigabit PoE switch (3560E, for instance) is about 1/3rd the cost of a brand new 3650 48-port PoE switch. So, again, I stock a couple of spares and I'm still FAR ahead.

Last generation's Cisco, or even two generations ago, are still extremely relevant and good switches. A network built on 3560Es will last another 5-6 years and cost half or less than the same network built on 3650s, or even 2960Xs.

It's not for everyone, certainly, but it is a viable option and to ignore it as such is just as much fanboyism as you're accusing me.

Also, I never said Cisco was the only option. I'm a huge fan of Juniper. I would not recommend HP, however, just because I don't believe they're as capable (4200xl switches have a max of 8 VLANs for instance.)

 

[classy]

For a new school infrastructure I would go all cisco. For your access switches you can go with a lower end type cisco switch, but all my backbone switches and routers I would do cisco. Ease of finding people to support cisco as well.

 

[kevnich2]

I'm def not saying cisco isn't good. In my experience, price wise, it depends how large you are and how good of pricing you get. We don't buy a lot of switches in high quantity so our Dell's cost 1/10 of that of Cisco so we went that route and bought a few extra as spares. Works for us. If you get good deals on cisco equipment, go with it.

 

[kevnich2]

As others have also pointed out, for this being a new school, I would recommend going with cisco. If at a later time they outsource support to an MSP or a contractor, you'll have much better luck keeping your existing cisco equipment and having them support it than with going with other brands.

 

[RadiclDreamer]

#3 is the way to go

You get the high bandwidth and reliability of fiber without the cost of running fiber everywhere. Not to mention, in these prices are the prices of optics included?

Another cisco vote as well. Dell have had too many issues in the past for me to trust them, and HP config is just kludgy to me. Juniper make good gear but its much harder to find someone that can do juniper than it is to find someone to do cisco

 

[kevnich2]

I am curious on the suggestion for SM vs MM fiber? A school can't have that much distance to go that it would exceed spec for MM so why wouldn't MM be better for this application? For one, the optics are much cheaper as well.

 

[drebo]

Single mode scales better (100gb+).

Depending on the distances, OM3 might not be enough for 10gb. I've done work for schools where some of the distances were 500m+. Cisco's SR optics won't do 10gb at that distance (well, it'll probably work but not be supported.)

It just depends. The cost difference in fiber and termination is basically non-existent.

Either way, fire them immediately if their plan is to put in 62.5nm fiber.

 

[kevnich2]

Yeah we always use 50nm mm for our internal uses. Haven't heard anyone using 62.5 in a while as the distance and bw supported just isn't there. You'd have to replace pretty quickly for sure if you went with 62.5

 

[Scarpozzi]

#1 is what I'd choose copper is the way to go for your installation unless you're doing something that requires real point-to-point stateful connections..... (NOT just web traffic or local file sharing), fiber isn't required.

If you have multiple buildings in your campus, feel free to run fiber between them, string some extras and use them for VOIP/bandwidth expansions. (cheaper to run extra now than dig later)

Cat6 will do 55M at 10Gbit....Cat6a can do 100M at 10Gbit. These are typically adopted in server rooms that need lots of bandwidth...but it's only been recent that people are jumping ship from FC, Infiniband, and all the other standards for data and disk connections.

 

[imagoon]

I suspect option #3 will be the best assuming the IDF <> MDF runs exceeds 100M for copper. One thing I can't stress enough for IDF <> MDF runs is conduit conduit conduit. IT can be any type you want it to be but when it comes to replacement time you can just pull new or additional inter-closet cable with little effort.

Something I would look at is Cisco SG500 series. We have been using these as edge switches and they have been phenomenal for the price. We have several 6 to 8 switch stacks using SG500X. 48 1 gbps ports to multiple 10Gbe uplinks and 10Gbe stacking ports works well for us. In larger purchases they will beat our small site Dell alternative (3548.)

Our smaller site switch is the Powerconnect 35xx series. These also stack which helps keep config management a bit easier. They have the bonus of using the Cisco style commands so if you can do IOS you will feel pretty much at home in these.

All of these can be managed easily via Orion NPM/NCM which for us is a big bonus.

For the love of everything IT holy, don't get "web managed" switches. I am still weeding out those Powerconnect 2848 units out of the environment. Ask the to anything more advanced than "dumb switch with a web interface" and they seem to lock up and need reboots.

 

[JackMDS]

With all Due respect this is Not really a Networking issue.

The number of $$ are given by the OP. It seems that OP as well as the other participant in the thread are fully aware of the Network aspects of the issue at hand.

If the school is a solid project that suppose to last for many years to come then the 50K Fiber is the way to go.

If it is a solid project that suppose to last but it is under severe budget constrains then go Hybrid.

If the project is a fly by the pants, the school might not survive beyond 3 years and every penny counts. By all means go copper and save a "whooping" 5K on a project that in whole probably cost few Millions.

If I was the OP I'll research how much it will cost in the future to go form Copper to Hybrid or Fiber.

Or in case of Hybrid is choice, how much would cost to upgrade to Fiber.

Then present it to the people that need to decide about running the school and future expenses.

I am sure that any future upgrade would cost much more than the current differences between the choices.

 

[alkemyst]

I do extensive consulting for school districts.

#1 is not scalable enough.
#2 is overkill.
#3 is the way to go.

Designate IDFs either per building or per logical grouping of classrooms. Use copper cabling to each class room. Use either 50nm or single mode fiber to uplink to an MDF, and use at least 12 strands.

I assume the numbers you listed are for just cabling and not switches. If it includes the switches, find a new vendor because the switches he's recommending are garbage.

Stay away from Netgear, DLink, HP's low end stuff, Cisco Small Business, etc. You want real equipment. Yes, you'll be paying 3k+ per switch, but you'll thank me later when you build out the wireless network and the students need to use it for common core and SBAC testing. One-to-one and BYOD are a thing now, even in elementary schools, and it's likely to be mandated at state levels eventually.

Schools upgrade their networking equipment roughly once every 10 years, so plan for that.

EDIT, misread the recommendations...for the price difference go fiber. That's peanuts. Unless you have some severe budget issues, it will make no sense to skimp on cabling infrastructure which is very expensive to mitigate later.

I'd go with 2960S or 3560 from Cisco...if the budget was <$3k a device. If a better budget, it's hard to beat a 3750/3850. Of course, that is just from Cisco...there are other vendor's equivalents, but that's how I'd do it.

Down here some schools qualify for e-Rate. You can buy at ten to twenty cents on the dollar which is really good. However; they don't cover SmartNets, nor redundancy (like second PSU's or supervisors).

I'd definitely dual-home each IDF. A single run to each classroom from the main IDF for each area is how I deploy campuses.

Typically a set of dual 6500's at the Data Center out to a single 6500 (sometimes 4500, sometimes a stack of 3750 or 3850s) at each MDF and then I run MM fiber to each classroom with 2960's or 3750/3850's depending on port density needed.

Works well. For critical classrooms (especially where State testing is done) we set up dual fiber to the end devices in a port channel.

However, on a severe limited budget; getting the port density needed is better than none and even simple dumb switches can handle that.