Designing mock networks

egrizzly · November 2013

Hi guys,

Have any of you run into any resources on the internet out there that provide a guide to designing mock networks. Perhaps you can call these network templates. They are predesigned networks created by engineers when given a set of requirements or scenarios. The networks are mock of course and are not designed for actual clients. I have scoured the web but have not landed any good finds.

Look forward to your thoughts.

DevilWAH · November 2013

There are plenty of case studies around, and thousands of configuration guides.

But by the time you start putting ideas together in to a final design you are using real ip addresses and config that no company would want you to share. And when lobbing things up 3 or 4 routers will tell you if a network of several 100 would work.

You will find examples of networks with 2 or 3 subnet sand 4 or 5 devices, but not many people will go the whole hog and design one with hundreds of devices and thousands of end station, as there is very little benefit to be had.

egrizzly · November 2013

ok, where do you find these case studies? As I'm trying to put together a plan to service people in my local area my plan is to first pre-design mock networks/templates that fit various scenarios. If a situation then arises for instance, a business with 5 departments that needs 3 pcs in each department, then I pick the closest template/mock network, modify it to fit that business, then go out and implement it.

perhaps you could share a link of these case studies/configuration guides that you mention. Thanks

DevilWAH · November 2013

If there was a one size fits all method of designing networks then we'd all be out of a job.

networks are of course hierarchical so if you have 5 or 50 departments and 5 or 5,000 hosts the basic model is the same. You take the best practices you have learnt from the vendors and the experience from previous installations and draw up a plan. All the vendors have lots of case studies and examples on there sites.

Clients don't want to fit to some one else network design, they need a network that fits them. You should be able to go in to a customer, listen to what there needs are and sketch a clear out line of a solution on a white board that is custom to those needs. They don't want to be shown a network with lots of "but this is not quite right we would have to change A,B,C...."

Personal I think the only examples you should carry around at previous networks you have designed and installed and asked permission of the company you did it for if you can include it in your portfolio (after you have removed any of there company information of course). The rest you should be able to pull out of your head on the spot.

egrizzly · November 2013

well, I guess google will be hearing my mouse-clicks more often. There is not a whole lot of difference between a pizza restaurant with 25 employes, 3 departments, that requires a network, from a travel agency with 28 employees, 3 deparments that requires a network, plus they want wireless throughout their building.

you're saying that after designing the first network from the scratch, when you get the message to do the second network you are to start from the scratch? just because there is a difference of 3 people, and the need to toss in an access point? perhaps you can explain what you mean better. thanks

tjh87 · November 2013

Devil is saying that each company is going to have different requirements. Yes, Company A and Company B may both have 28 employees and 3 departments with a wireless requirement. This does not, however, mean that both companies have the same budget, applications, or bandwidth needs. You can't have a one-network-fits-all approach. So yes, to an extent, you should approach each customer individually. Company A may want the biggest and best network money can buy. If this is the case, you'll have a fully redundant core, distribution layer, and access layer with the latest and greatest Cisco gear and 10Gb connectivity. Company B may be a start up and won't be able to afford Cisco gear and may want to combine the core and distribution layer (very common). Therefore, your marketability should be in your adaptability.

DevilWAH · November 2013

well the high level view is always the same kind of design. lets take a generic small / medium network design. You have your core, distribution, access layer, WAN and wireless etc and the technologies you might be running on it.

This high level best practices you should have in your head and be able to scribble a variety of it in seconds based on the company's need. You should not need a template, or will it be worth having one as there is such a slim chance that a template will fit with out adjustment. The details, such as number of vlans, interfaces needing to be configured, iP address to use, routing protocols, SSID's, QOS, number and models of devices, etc, etc. your spend as much time altering a template with all this information as you would from starting from scratch.

So yes high level overview of networks you could draw up a few different designs, may be full three level (core, distribution and access), a collapse core design, the cisco enterprise architecture model, single office model, Branch office, etc.. (type network design in to Google images). That you can show customers to discuss the benefits and issues with each. Or go back over the CCNP as this covers a number of these. These give you a nice picture to discuss, but its not a "template", no such thing. I wrote a config generation tool for a global bank once, even to provide a template for a single device type was so complicated due to the number variables in the configuration. When you have 500+ devices with a standard config where only the ip on the interfaces change then a template starts to make sense. But for small medium networks with 5 - 10 core devices honestly no point.

egrizzly · November 2013

tjh87 wrote: »

Devil is saying that each company is going to have different requirements. Yes, Company A and Company B may both have 28 employees and 3 departments with a wireless requirement. This does not, however, mean that both companies have the same budget, applications, or bandwidth needs. You can't have a one-network-fits-all approach. So yes, to an extent, you should approach each customer individually. Company A may want the biggest and best network money can buy. If this is the case, you'll have a fully redundant core, distribution layer, and access layer with the latest and greatest Cisco gear and 10Gb connectivity. Company B may be a start up and won't be able to afford Cisco gear and may want to combine the core and distribution layer (very common). Therefore, your marketability should be in your adaptability.

well, I was kinda thinking like with everything being equal. its kinda funny how there's never a straight twig in networking isn't therer? by the way how are you? I just moshied onto this site after about a 2 year hiatus.

egrizzly · November 2013

It looks like you're interpreting "template" as strictly network drawings. However I'm very much referring to a process in this case. A standardized process by which to generate a network based on a set of requirements. Of course the process can be supplemented by a sample network design. The sample network design would show a visual illustration on how the process was used to arrive at the said network. Certain business requirements would call for collapsed core, full three level, e.t.c but would would still go through the same process. I say this with some confidence because Cisco best practices (which are very much similar to the industry) are process based and not subjective. other thoughts...

DevilWAH wrote: »

well the high level view is always the same kind of design. lets take a generic small / medium network design. You have your core, distribution, access layer, WAN and wireless etc and the technologies you might be running on it.

This high level best practices you should have in your head and be able to scribble a variety of it in seconds based on the company's need. You should not need a template, or will it be worth having one as there is such a slim chance that a template will fit with out adjustment. The details, such as number of vlans, interfaces needing to be configured, iP address to use, routing protocols, SSID's, QOS, number and models of devices, etc, etc. your spend as much time altering a template with all this information as you would from starting from scratch.

So yes high level overview of networks you could draw up a few different designs, may be full three level (core, distribution and access), a collapse core design, the cisco enterprise architecture model, single office model, Branch office, etc.. (type network design in to Google images). That you can show customers to discuss the benefits and issues with each. Or go back over the CCNP as this covers a number of these. These give you a nice picture to discuss, but its not a "template", no such thing. I wrote a config generation tool for a global bank once, even to provide a template for a single device type was so complicated due to the number variables in the configuration. When you have 500+ devices with a standard config where only the ip on the interfaces change then a template starts to make sense. But for small medium networks with 5 - 10 core devices honestly no point.

networker050184 · November 2013

You can find design guides on Cisco's website. Mostly the detail you are trying to get at would be company specific and usually somewhat guarded. If companies put out their templates and design info why would you hire them at that point?

DevilWAH · November 2013

No not at all, you miss understand what I am saying.

The process to designing a network is to understand what the different levels/technologies in a network are for. There are simply to many variables to construct a matrix, to formalise when for example you should use a collapsed core and when you should have distribution switches. Indeed running both is not uncommon. Types of data usage, numbers of users, security, budget, existing hardware, physical locations of hardware, etc, etc, the list goes on and on and on. two compinies with identical numbers of users and departments may require networks at opposite ends of the budget scale.

Cisco's best practice and the industry's are just what they say, best practice. but this does not mean they are always right or suitable. In designing a network you have to take only three things in to consideration.

1. Does the network support the volume and types of data, and deliver it efficiently and in the time frame require, this might be VoIP, time sensitive applications, multicast streaming, large data sets for archiving or analyses, etc.
2. Does the network keep the data secure and correctly route its from point A to point B. CAn the operator see what is going where and control it.
3. Is the network resilient, If bits fail can it continue to run, does it need to? Some companies only want part of the network resilient and save money.

If your design meets the customer needs, and at a price with in there budget then it is the correct the solution. even if it has got a few single points of failure, as long as you high light the issues to the client and discuss what effect they could have. There is no right / wrong way in IT, just good and bad ways.

If I am designing a network I propose 3 solutions.

1. The very basic design that I believe the client could get away with. Consolidating devices where possible, and with just enough resources to meet there requirments.
2. A "best Practice" design, not quite money is no objects, but a design that exceeds there need, flexible, easily managed, drawing on all past experience and putting some thing together that I would leave my current job to go be the network admin for.
3. Taking in to consideration the requirements of the client and what I may know of there budget, merging the two designs above together in to a midfield solution.

I will then sit down and discuss all three options, explaining that option A and B are at the extremes and that really option C is the one I feel they should be looking at. Having options A and B to hand while discussing it means you can refer to them. Why are you suggesting to chose for this bit of the network to go the option A route. Or why do you think it is worth spending the extra money to set up the server room network as per Option B.

Cisco best practice for QOS.. http://www.cisco.com/en/US/technologies/tk543/tk759/technologies_white_paper0900aecd80295aa1.pdf

If you are CERN copying terbytes of data across Europe from the LHC on 120Gig links to be analysed around the world. Then reserving 33% of the bandwidth for voice is defiantly not best practice. You have to understand why CISCO has designed a best practice as they have. what the limitations of it are and what is at the core of it. So when you are designing a network you can adapt them to the needs with out completely ignoring them. Implementing best practices by following a process with no thought or understanding is as bad in my view as not considering them at all.

egrizzly · November 2013

Devil you make a very good point, but of course instinct tells me there's still stuff out there that assists/guides Network Engineers in network design of many sizes. Pre-set processes/templates just speed up the design and overall business process. It's that simple. The ones who get a thorough network assessment and "design from scratch" are those businesses that pay top dollar for it. Contrary to what you may believe network consultants that work for big consulting firms do try to direct their clients to the closest template they have. It eliminates errors in network design, affords the client a speedy turn around time in case of a re-design situation, and generally reduces costs.

Although your process calls for a custom network design each time you will have a heated debate if you go into a bar full of seasoned network engineers and tell them every single network you design is done from scratch. Not that I'm trying to argue with you of course but you certainly need more information on the issue so you can look at it from the business line of sight.

networker050184 · November 2013

Of course companies are going to have templates and try to make your network fit into their normal deployment model, but again those aren't going to be something they are readily sharing with the public. You might be able to find some with some good google skills though.

Mostly for freely available info you are going to find deployment guides, best practices etc, rather than actual templates.

DevilWAH · November 2013

from experience working with a Cisco gold global partner. We did not try to fit our customers to any template. An oil company dealing with keeping all it's oil rigs, refineries and tankers in synch has no resemblance to a financial banks network. I would say the larger the network the less use a template is.

Internally to a company we would create templates for sections of there network. So should they want to deploy a new network cell in to there data centre, it was a scripted process to keep them all aligned. But between companies our team designed from experience not from scratch, and defiantly not from templates.

DevilWAH · November 2013

networker050184 wrote: »

Mostly for freely available info you are going to find deployment guides, best practices etc, rather than actual templates.

Exactly,and I think that's all you should need to quickly and effectively design a net work. For small networks you can put together an overview aligned with best practices in a very short time. For a global customer with 50,000 servers and 200,000 workstations. You have a team working for weeks/months putting a design to gather and no template I'd going to work with that. Believe me I have sat in enough of theses design meeting. As much politics as technical.

tjh87 · November 2013

I agree with everything DevilWAH is saying here. OP, if you are looking for a template to go off of, it would be this:

Core
|
Distribution
|
Access

Everything else is going to be up in the air (even the above model in some cases). So if what you are looking for is best practices, then Cisco books have all of that. As far as a one-size-fits-all template for building customer networks; not going to happen. There are just way too many variables. And just like Devil said, consultants and architects in the real world DO build each network from scratch based on the customer's needs and best practices. Those best practices being the "template" in your scenario. So I don't know what bars you are going to.

egrizzly · November 2013

...well that would be quite surprising to me. I am quite sure that these meetings are necessary for large enterprises. I actually do appreciate your insight @ Devil so don't take the bar comment personally. yeah, "processes" and "best practices" would suffice for me and go as far as satisfying what I'm looking for. I have heard about the politics from an experienced friend and you're right, their is a lot of it.

Designing mock networks

Comments