Global Scale Network Design with Malcolm Booden – ZNDP 092

ZNDP 092: Global Scale Network Design with Malcolm Booden

Zig Zsiga: [0:00] Global scale Network design with Malcolm booden episode 92.

Welcome back my friends nerds geeks and ziglets out there we have another episode of the zigbits network design podcast where zigabytes are faster than those gigabytes hey we strive to provide

real world context around technology what

up everybody I hope everyone is doing great zsiga here and welcome to episode at 92 of the zigbits network design podcast

once again I'm zsiga I'm here to help you with network engineering Network design and network architectures and today we are discussing,

global scale Network design helping me today is my good friend and well-known industry expert Malcolm booden.

[0:48] Hey real quick Malcolm has been on the show a couple times Malcolm is a seasoned networking security professional with 20 years of experience in the IT industry he has worked for various Enterprise customers directly along with isps internet service providers and,

integrators he is currently the chief executive

officer and Lead consultant at MMB networks and he specializes in network and security design and Delivery for customers ranging from small to large Global Enterprises and that's all we're talking about today is global

Enterprises global scale Network design

Malcolm is a Cisco certified design expert and holds a various other certifications from Cisco for Dan and Palo Alto networks aside from Consulting

Malcolm has developed a training offering and let me let me this is extremely important I want to foot stop on this he has built a training offering focused on real world and real life Network design,

this is extremely critical for all of you to hear if you want to check out his training you can go to training dot m and b Network stop,

I'll have the link in the show notes as well but he has a real world training program to teach you how to do Network design and wider Network design and all these things that we just don't get out there today,

so go check it out hey Malcolm thanks for joining me again buddy I know it's been some time how are you doing today.

Malcolm Booden: [2:14] I'm doing very well my friend how are you.

Zig Zsiga: [2:17] Oh man I am on Cloud 9 it is a great day getting a lot of things done making things happen I'm super pumped and excited for today's show and just so everyone knows we're going to talk about global,

scale Network design if I said that correctly please feel free to correct me throughout this process welcome as always,

that's what we're going to talk about today I'm excited.

Malcolm Booden: [2:40] Yep another way in freezer is probably packed massive Networks.

Zig Zsiga: [2:44] Big massive networks are you go extremely large massive networks that that's what we're going to talk about today so before we jump in though as always.

Can you give us a quick rundown of who you are what you do that kind of stuff The Cliff Notes version is always.

Malcolm Booden: [3:02] Sure I'll keep it brief because of actually appeared on one of your previous shows and easy to find on the internet on LinkedIn LinkedIn / n / Malcolm booden all lowercase and Twitter,

the same Alias Malcolm booden.

I've been on networking and Whitey for about 20 years the majority of that and networking since about 2003 graduated with a degree in network computing.

What Enterprises in the UK social partners isps.

[3:44] And various things that I th awesome companies etc etc and and for the last four years or so of.

Been working for myself I suppose I have a smoking so and say call them in the networks and we work with.

Large Enterprises mainly funnily enough so what a small organization but we plug skills Gap I suppose and design mainly.

Across Global Enterprises ranging up to Fortune companies so as pretty interesting what the last few years and that's really the.

The basis for today's discussion and I'm ççde certified as you well know and for anyone who doesn't know Zach and I go way back they probably about 2015 when we were studying for the ççde together.

Zig Zsiga: [4:37] Seven years ago right as I do that math right six years ago seven I don't I can't math I can't math today so but

welcome again but I appreciate coming back today and we did do a show

it's been a couple years few years back in 2018 just so everyone knows it's episode 25 it was designing for data center migrations and application Mobility it was a great show so if you

didn't listen to that show I highly recommend going back in time have a good listen to that show Malcolm and I talked about how to migrate between data centers it was a great

deep dive Network design focused show so feel free to

go look at that show listen to it quick link for you is zigbits dot text / 25

all right buddy so today we're going to talk about global scale Network or large extreme networks or whatever the wording is right we're going to talk about some Network stuff so.

I'm gonna just start what what is the difference from your perspective between a small or smaller scale Network design and a global scale Network design.

Malcolm Booden: [5:46] All right so there's probably a lot of different answers to that question and let me let me put some clarity around my background and working I suppose specifically in the integrator space,

or the networking agree or space and value-added reseller space as a engineer consultant and moving into,

pre-sales and systems engineering and doing a bit of all of the above most of those places their main focus at the.

[6:18] Anna graders and isps are work thought they were ranging,

probably small and medium Enterprises and those skills of projects.

Some examples of projects that you would get maybe replacement of a course which,

but of course which is perhaps service in twini.

Sweats tax at the axis we are or maybe addiction or core rule out with,

all sorts of weird and wonderful protocols and talking Nexus like multi V DC with oh TV and.

Calculating the right amount of ports and things for when you you know taking an account so I guess that's being a bit specific but I guess the.

[7:06] I'm getting out there as what I find with smaller organizations you can easily walk into our project,

and discover that is really really complicated defects a business problem so if you've got,

Cisco Nexus with multiple vdcs and or TV and vrx within the BBC's and you have to migrate from a flat Network for example,

that's maybe like a year-long project or six months or a year or even maybe longer depending on the complexity but the reason that is so long,

as

the complexity you've got so many different protocols so many different layers so many things that you can break and all you may actually be doing is refreshing and end-of-life piece of cat.

[8:03] All right so that that might be a small Network design but a very complex Network busy.

Another example might just be if you're changing out the corporate firewalls for example.

And that could be a long process I mean might take three months it may take six months,

if you have to plan a lot because you think I'll pay two firewalls but again and the lovely what would the network virtualization if you have firewalls regardless of vendor,

and it has the capability to have one big box or up here of big boxes,

that is then what July used in multiple different firewalls and that firewall or those firewalls and the virtual firewalls.

[8:54] Are your corporate segmentation policy and what I mean by that is your everything via the firewall and perhaps you want to move away from that dependency and moved to something like,

Sgt zor.

D any scenario overlays or something so go even if that was a small Network that would be really really complicated.

Know what I find in comparison with global Enterprises and one thing I actually didn't mention,

but those examples it's normally you have a couple of data centers and it might be an country.

[9:36] I know the u.s. is in the u.s. is like multiple Scotland's where I'm from but you know if it's where then stay for example and in your case is like,

then that would perhaps have a couple of days indoors and.

The skill of a you're thinking two data centers and the sites that need to communicate with the data centers and a region effectively.

[10:02] Where things get a lot of butt or quite a lot different I should see and in our praise.

Scale or large Enterprise scale as the you effectively have that,

footprint that I just described but you have it in every corner of the globe and effectively replicated in every corner of the globe and when I see a corner of the globe I mean Global region.

So not like a region like a state or a small country I'm talking like asia-pac.

Amia North America and Canada are normally group together and then Latin America.

[10:44] And if you then start with no even got anywhere near the tip of the iceberg yet but when you start thinking about that.

I'll just mention a few things before,

I think a breaking and hand it back to you are things like compliance quality of infrastructure and the ground for one connectivity.

Chef Peng getting things to say through customs.

Because if you're if you're running a project from a certain country like.

[11:20] Somewhere in Europe like the Netherlands or the nordics or Germany somewhere it's easy to get in and out of easy to get in and out of,

if you're running a project from the then it might be difficult to get cat n,

to other countries and other parts of the world or you might have to Source locally and things like that that's going to knit some detail so I suppose that's where we're talking about scale,

instead of like effectively being in one location where you might do very simple or very complex,

projects regardless of what that is when you go global scale just remember whatever you do in region 1 you're going to probably have to replicate and Region 2 3 and 4,

and perhaps in multiple places within those other three regions and then the cloud know just to add that curveball.

Zig Zsiga: [12:17] Add the cloud on to that

I mean I killed there's so much to unpack right there's so much to talk about here I think the definition of the example you gave of smaller scale still

complex and times right still things that our capacity could potentially be complex and take a while to complete,

like that core migration right it's an end-of-life device we're going to you know

refresh that device it's a very complicated there's a lot of things going on it running on it and now we have to migrate it right but that could be one physical location in one state or one lo que no one one area but then.

Your transition to the global aspect where you're breaking up the design elements

by region right and that region is traditionally based on the regions in the world you said like a mirror I think you said asia-pac I don't want to speak for you but is that what you said.

Malcolm Booden: [13:18] Yes ome also Europe Middle East and Africa normally group together US and Canada or North America and Canada Latin America,

so Brazil Argentina Chile and all these countries and then but sometimes as well depending on which organization you're working with.

If you go to Global Enterprise one they may actually group.

North America Canada and Latin America as the Americas so there's actually only three Global regions such as Europe.

Asia-pacific always on its own because it has some very stringent.

Requirements of and compliance in certain countries and then I made a cop but there is effectively for Global regions but some organizations just group the Americas together.

So yeah that would be the that would be the mean Global regions that was referring to.

Zig Zsiga: [14:19] Well that's great that's great right because I I didn't know that so I'll be honest I didn't I didn't know that honestly that people group,

the Americas together at times

so you mentioned like things would have to be done so whatever you're doing like region why you might have to do it in Region 2 and any of these circumstances are there like if you make a change a design element change in one region does it have a negative

on another region.

Malcolm Booden: [14:47] I'm going to say it depends.

Zig Zsiga: [14:49] It depends it came.

Malcolm Booden: [14:51] Just what I can do you know what are we like 10 minutes and then I've see that the pins already but there you go.

Zig Zsiga: [14:57] I was waiting for it was waiting for I got a timer somewhere in here.

Malcolm Booden: [15:01] It does as dependent on what the what the design changes.

So one of the things for example that you might think about let's talk about when there's an exam effectively 11 and 1 and sd1 at the moment and have been for the last few years,

but some of the things that you might consider they are.

F you go right back to the start of the design and you see or yeah we want to do an SD one that sings can eat we can get overlays and we can get segmentation and then we can do this and that and the next thing.

The first thing that you might see as I know but there's a not as well how are we going to deal with China.

And because there's the great firewall there and those challenges with getting certain types of traffic and encryption and things like that and that.

And there's also other countries turkey and Russia have their own white compliance can separations we're at other countries probably have the same thing where you have to break out of country to the internet in that country.

For example I think in Turkey if you get an an ISP Link in Turkey.

You wouldn't be allowed to use web proxy and Frankfurt to Tunnel over 2 and then break out Frank for years to break out from the a ESPN talking.

[16:28] Or that was certainly a requirement that came up in one of the designs that Google designs I was working on but don't don't don't.

Totally hold me to that that's just something that I design constraint that came up so forget the country that could be a constraint that comes up therefore if you change the design in that country.

At definitely or if you follow the compliance I should see in that country that's definitely different if everybody else is going through.

Web proxy over a GRE tunnel you know one of the different vendors that's available for doing that and then provide and Sassy services.

That could that could be one consideration and then the other one might be.

[17:17] I don't know if you are looking at will get on and that's a little bit like the skill and aspect of it in a little while but,

just think about traffic flaws when you've got four regions traffic flows,

if you're on an HD one environment and I'm using HD one because as a really interesting.

People think or some people might think it's easy because you just point the controller and you collect bits actually when you when you start getting into underlays and overlays and,

things like then it's actually I think it could be more complex but that's subjective.

Where I was going with that is that if you.

[18:00] Go right back to the start of the design does a simple designs decisions really at the face of it that you have to make a decision about traffic causing a global design I let's see.

Customer wants to do a nasty one deployment and their strategy is complete internet everywhere.

That's great oh good now the net you start peeling off a layer and ucok Global Internet deployment and the underlay nor mpls is being used.

What do we have to think about and in the end the internet design.

Those different tiers of internet connectivity which equals different quality of internet connection which equals different.

Types of performance which equals different types of traffic flows and so on and so on now.

When you then stop looking at that you also have to think about stuff like let's see we've got four regions.

The four that I mentioned we split out North America and Canada and Latin America of Europe and in Asia Pacific and we see everything is on the internet.

And in true overly Style.

[19:17] A seat and Europe wants to talk to another seat in Europe.

Then what do we what do we what do we decide here that's fine we want we keep that n region so Europeans a one talks to European site to and it never leaves leaves Europe.

For that communication however.

F we want a say in Europe to talk to a seat in Asia pack,

how are we going to how do we want to achieve that.

Do we want to just leave it to chance and say go directly over the Internet so you might have like sd1 controller in your Datacenter for your control clean the control plane tells you if you want,

hi hi.

Insight in Europe you want to speak to that so you can asia-pac bold a tunnel to this IP address and it tells you what IP address to go to and then you could build a tunnel directly over the internet.

[20:24] Another way that you could do it as if you have all your traffic in Europe talking to each other directly.

Well if you want to then talk to Asia Park then you re-enter the Datacenter in Europe,

across a backbone Network to a person or an age of part and then over perhaps a local mpls connection,

that may mitigate some of the challenges are opened you know the great firewall and these compliance needs.

[21:02] And that's just one example that's it one design decision that you need to meet God think about and it's around about traffic cause but even within Europe you may see.

Okay I need to be careful about the supplier that I select.

Because if we go for service provider one and they've only got three Pops in France.

And then I want to.

I want to talk to a say another site in France you may have the back called the Germany in order to then speak to another seat in front if your provider doesn't.

Have that capability and that goes with then that's that's within region Israel.

Zig Zsiga: [21:50] So it sounds like there's a lesson learned here for others that might be venturing into the global,

scale Network design realm and one of those lessons learned that I'm tracking here is you have to understand your your traffic flows you have to you have to pre script how the traffic flow is going to work,

you can't just assume that and let it kind of do its own thing you really need to be clear that hey know if it's in Europe and we're going to go to you know

asia-pac or wherever you know we're going to come back into the data center use the you know

data center interconnect link of some sort and then transition over to a wide area network in the local region like you mentioned right but like I want people to understand that like.

[22:36] You actually have to take it and you have to design but you have to you have to design from a individual use case,

every flow for your traffic and I was going to get into Malcolm like just internet access,

like I think I think you're getting into even like how do you go from one location to another location but like if we take one region,

right and that region has IP addresses is probably an IP address scheme I'm a assuming right there's probably some sort of translation happening on some sort of device Nat or Pat or combination thereof,

so you know how would you do just just basic internet access between all four regions.

Malcolm Booden: [23:22] So basically access those full rate so the other thing that may or may not help or ID this discussion as if you start thinking about any axis.

As we know what I suppose is going to add a seminar well as a firewall that you break out and either seems you somewhere or you.

Like a web proxy when I see somewhere or it seems you to the end on it and there's something the on-prem web filtering capability or the.

The cloud-based web filtering and solutions we are,

wow thoughts either have clients or some mechanism so if you're wrong and you're not even on the VPN so wait.

You know umbrella and take functionality is you skill or Palo Alto press my foot in there all these vendors of all got Solutions are different ways shapes and forms of delivered on that.

And the basically if you're talking about a laptop getting on in the end on it those normally either a client or some sort of configuration in the registry or something like that and literally I went booze gay but that's that's kinda how that's achieve.

[24:39] If we're talking about one seats were really are starting to see the chef.

And general terms towards Robin and on a break either feeling on it on the sites or.

[24:53] A hybrid model depending on like you some some companies.

For example production and Manufacturing and process control networks meat just like that Comfort I have a nanny SLE between private,

a private socket but put everything else they ponder over the internet now when we think about that.

The main traffic flows.

The I see sites as you can see that are adopting internet first strategies.

[25:28] Are the bones from the site.

You basically go via a cloud web proxy or what I GRE I don't accept the GRE tunnel do a web proxy and that that provider.

Effectively you're going to someone else's data center instead that your own and it's a service.

And then the the policy that you have in your instance of that.

- well I mean z scaler umbrella Palo Alto or whatever the technology is.

And that's that the and then they can do things like SSL decryption,

and things like that which takes a headache a way of having a firewall and every site and things like that but going back to the traffic flows so that's an outbound web browsing directly from site but via someone else's data center no.

[26:21] The other traffic flow in the other direction to the site's doesn't typically happen.

The IC enemy there might be some applications that are hosted at local seats that traditionally they were accessed directly.

If you were going and redesigned not Network you would probably want to bring that through your central point or migrate that service to the cloud or whatever.

No the clothes is another one because you will see the clothes.

Actually before I move on at that because it's all kingdoms are related.

The instance that I gave or the example that I give we are you go via their smudger call cloud service to do your web filter then that's all great.

On average I would say.

Any particular vendor but on average having spoken to quite a few different than those on that type of.

[27:26] Secure access Services Edge or sassy model as it's being called in general terms as normally for each vendor around about 50 pops or up to about 50 pops,

across the globe that have these services.

And then when you start investigating the services that they can do not all of those 50 pops so that's maybe 50 pots across 30 or 40 countries.

Zig Zsiga: [27:56] Not everywhere.

Malcolm Booden: [27:57] And what I've seen as that.

Or some of the feedback that I've seen from some customers and I won't mention any vendors but from what I've seen is there's not enough pops for us to do this globally so it's another skill and thing so it might be.

Country one might have two ports in order to deliver like secure internet or secure access Services Edge web proxy,

via their pork and it might have two or three Pops in that country but the country next door to it.

I'll just use I'm not picking on Spain and Portugal but let's just see Spain and Portugal are right next to each other and.

And those those are.

Proxy or a port in Barcelona and there's a port in Madrid but there's no port in Lisbon or the one in Lisbon is cut down so if somebody wanted to use that particular vendor service in Portugal they would need to go into a spin.

[28:58] And then and then yeah and then in certain countries that have the compliance restrictions which are not need many more countries have been want to offend anyone I'm just using them as examples but you get the you get the point that basically.

Yo

Zig Zsiga: [29:16] I'm sorry I just,

basic internet access is complicated like like like I just want to make that clear that everyone like basic internet access from a global scale design can be very complicated and you have to treat it,

individually for each region right like and I mean in the u.s. I don't know if you have to do the URL filtering option.

Actually I've never seen that welcome before so that's that's quite interesting that that you're seeing that a lot

in different countries where you know you have to get a service and your convenience just pointing your web traffic to that proxy that web proxy and someone else's data center to get out to the internet.

But.

Malcolm Booden: [30:01] Yeah I only wear proxy so as I say it's unlocked it like king of Cisco umbrella Z scale or,

remember this the actual product or Angus Palo Alto plasma access and things like that and,

these are all things that basically you can up like web forward because multiple other things that you can do so it can do SSL decryption and.

Or not with Vote or for example so as you're going out to an accepted website at executes the web policy and the clothes but.

With decrypted traffic so you don't have to download your environment with Woods firewalls and.

There are some things like that that are capable of doing this yourself off Lord now I'm not saying that's the only way and the only correct way to do things but that's what's happening in the industry

and it's definitely happening in the global scale Network design but the challenges that are being proposed are.

Things like but not limited to your not in enough places for us to be able to use that service or the performance is suffering and one of the other things that we just talked about.

If we if we shift slightly or private slightly then we'll forget web browsing and going to websites now think about the cloud.

[31:29] The clothes as is just a funny one to me because I hear it used all the time.

And many different like the term clothes right but what is it again it's the Assassin I was somebody else's Data Center.

Right and again within a country you may have you've got regions and availability zones and availability zones are.

Did like one or two days seven hours whether in our region so London for example might be a region.

But then those two or three days and others which are the end the availability Zone and these are basically data centers which back each other up.

No

The two mean the water steaming obviously offerings and these Cloud environments those infrastructure as a service.

Software as a service and platform as a service platform as a service I never see Ava really probably exists and people probably use it but my personal experience I don't see people using that.

[32:45] Infrastructure-as-a-service,

it does get used to that sleep basically you set up vpns or a mechanism to get the servers which are spun up in the cloud and then you install your own applications on the operating systems,

I don't know software-as-a-service so with it with the infrastructure as a service you might see be it's pretty obvious that you'd be very specific with your meatball,

I need additional in London and then one in Amsterdam or Brussels you know I or Frank for or one of these types of large hub,

places which are Central and then you have several all sorts of things in the u.s. and Canada and Asia and,

and so on in Brazil and things like that now when we start looking at that.

There.

[33:40] The infrastructure was a COS it's about it's kinda a network control as Network architect because we can architects in right well that's close to us that's course the US that's not close to us,

so we'll choose these data centers.

And we are kinda and control because we have to sit that connection up whether it's us I may be sick VPN whether it's an ASD one whether it's our daily connect of some kind we have to set that up.

When it comes to software as a service I go to a website to access that whether it's controlled or not controlled.

[34:20] And the owner and what I'm finding as these Global Enterprises.

We are maybe ten five or ten years ago even even going back to 5 years ago.

You buy a Sox application or for vendor or supplier and you just use it.

Know what I'm finding is that an rfps and,

and new network designs cost the most support especially Global Enterprise customers they will challenge these companies to see,

where are your applications who's that.

Right they'll ask that question to the software as a service supplier and then they will also say to the ISP who's providing their internet link where is your pop what did I say in our is you're Poppin.

And they then start if you want if I want to get to Microsoft azur.

If I say this all SAS application vendor where is your application hosted in the sea Microsoft Azure Frankfort.

[35:33] And then I go in there and this particular day is in or in Frankfort and also this particular day is in or in Paris then I would be saying.

I want my internet connection to comment about this in or as close to it as possible and then for you to have a cross connect.

Of some kind or some mechanism and place already or the capability to route to I'm taking the fastest the fastest route to get to my key applications.

Global Enterprises are asking those questions that and I think that when you and I were going through the CCD and there was a hole that service provider Technologies and that's him appraised Technologies the emergency.

Absolutely.

Zig Zsiga: [36:20] Hard hard core merging so no I think going through that process shows,

a lot of complexity and also interdependencies as well I like that you do you define SAS and infrastructure as a service and platform as a service.

You know so so I was going to ask some questions but I think I'm going to walk through this this,

thought process of my head here and then maybe questions will come out of it so let's say I have a SAS application and that's ziggs SAS application.com

right we'll just make it ziggs you know awesome SAS application.com.

Malcolm Booden: [37:02] I think you're onto something there me.

Zig Zsiga: [37:03] Yeah right.

Malcolm Booden: [37:04] Mark get it Market at then released.

Zig Zsiga: [37:08] Released right yeah make millions so I have this now,

and I think I think on,

from a user's perspective all they care about is that I'm going to type that in my browser of choice on my computer of choice and all I care is that it works right it's,

get sufficient it works I can get to it complete my job or you know I'm not doing my job and watching videos whatever right I'm watching YouTube Netflix so.

The implications on the back end is what I want to talk about like you mentioned where the application resides eyewear is it hosted in a specific data center certain regions like I want to know like how are you people even getting to the

location right like how are we validating that the data is actually at that location as well because there's the back of the applications data right.

I mean.

I don't know any application doesn't include data and so I'm even thinking like the implications here of a SAS application that is global scale how do you replicate that data.

Malcolm Booden: [38:17] Yeah I think that's a shame I don't know to be honest like I don't really know the answer how that data is replicated because I think that's more in an into the realm of being a cloud gooder.

Zig Zsiga: [38:30] Ah see ya.

Malcolm Booden: [38:32] But what I would say is that how I understand these things are that in order to get to the course.

We're talking about Technologies the same old stuff as we've always thought about like anycast DNS combined.

[38:53] That said it might be you know I don't know I honestly don't know the create the answer that.

Zig Zsiga: [39:00] Yeah it's a question to his top of mind.

Malcolm Booden: [39:02] That would be something that I would I would ask our supplier Angie.

Zig Zsiga: [39:08] There you go everyone.

Malcolm Booden: [39:09] Global for anyone that's less than and if you're going through a SAS application.

And process and select and then going through a bad that an RFP and you're the network team there's a question for you to ask,

that's that's application is global how do I know nor that I'm going to the right place but even

even just even an access in the right data and then that opens up a whole world that other things about compliant again that that this will be something that you'll see over and over and over compliance standardization.

Across because that's what big Global networks need compliance and standardization in order to operate effectively and stay up and be predictable as the holder actually going to behave.

Zig Zsiga: [39:57] And then data sovereignty we didn't get into like data sovereignty so.

Malcolm Booden: [40:01] With us that's going to be I was going next and when you're when you're talking about so there's there's the question some questions like how the how where is my data located that's one question,

how is it secure if it's a if it's a global platform and it's a SAS application how is it secured from other tenants on the platform same platform.

How do I make sure that I get the most optimal route and this goes back to me talkin about.

They're seeing where is your application hosted to the application to Veda an NEC to this the ISP.

Where is your pop what's the route that my traffic takes in order to get to this Microsoft Azure Data Center.

Microsoft Azure additional 1 2 3 how do I get from this site that the center over urine or internet connection we're not even talking about private,

networks year we want to know who your appeared with how we to that specific location.

Maybe not for every application but for key occasions or key services or key locations like these are the cloud data centers or gcp awas,

these are what we need to prioritize how do you ensure that your solution gets us there in the most efficient manner.

Zig Zsiga: [41:22] Yeah I mean so I think that is extremely valuable those questions right especially for people like you know,

Network Engineers security Engineers that are going through the process of identifying you know where are we going to you know have this SAS application or where are we going to have this

infrastructure that we're hosting right that's an infrastructure as a service offering and and where is our data going to be right these are all the things you have to think about compliance is huge I mean compliance is one of the biggest ones because each country

has different compliance standards,

and as a company if you are that big global company now you have to you have to follow all those different standards for your data

for your applications I mean it's crazy the implications here and you're living this Malcolm you're living this every day right that's what you're doing that's what you've been doing for the last three and a half four years this is this is awesome

awesome.

Malcolm Booden: [42:21] And it's really and with all that's taken into consideration one of the things I learned quite quickly going from that,

that you know that Network and agree or space where it's just like crazy like you know you that you got maybe got three four five projects small really complex projects that are going on and it may not all be on you you may be part of a team,

but it's late.

Right that's done next right that's done next next next next then you're jumping about all over the place and then mix them all that those like all these little intricacies that you have to account for,

and beg Global organizations with a standard tidy standardization sorry and.

Protect the Bella E I suppose of the network but there's like there's a spanner that can be thrown into the works with that which I'll come back to in a second,

but the key objective really I suppose whether Bank large Global Network and the preference as that.

Steve Bella and predictability is key for Kang I should stay because and repeatability.

So where is you have these like small intricate.

Designs and implementations and small to medium Enterprise space.

[43:51] And it's not always like that but I'm just seeing you more likely to get them there and I'm not saying that you never get them in the large Global but what suppose I'm seeing is that.

You could have,

less feature-rich stable and repeatable and easy to operate and troubleshoot.

Network that does the job.

[44:16] I'm not saying it's not setting the here though I'll late but it's doing the job and that thing that you said about put my laptop on I can get to my stuff.

It's acceptable and I can do my job that's more what Global Enterprises are after,

and they know what they've that's what they strive for is that they want they know what's going to happen and the certainty because if you have I know it's a large global.

Enterprise and I say I say going don't make mean our production line,

who shefte mates are not for the work 8:00 p.m. to do a 12-hour shift and because the network is down,

because some crazy thing is being implemented rather than just something simple,

I know they can walk it what as those 20 or 30 or 40 or 50 people just get sent home for 12 hours or could just get sent home for 12 hours and that means that 12 hours worth of product or.

Whatever that's being produced at that location has not been,

and then they'll be a monetary value attached to every single product of the lines even if it's.

[45:33] If it's 30 products are made in a 12-hour period.

And the old cost the sale price was a thousand pound that's the only Grand that's been lost.

[45:49] Because the network the network was too complicated to troubleshoot and I suppose you can apply that to any company but

Global Enterprises that's why you see they strive for you know it's the same so if something goes wrong at one site you've probably seen it before or you know it's the one link or you know it's on SFP or you know it,

or the church that configuration isn't standardized with which PF something must have changed something,

or it might look different because an exception has been made because they needed to do something specific on the site this all comes back around the design documentation.

Zig Zsiga: [46:27] All the basics all the basics.

Malcolm Booden: [46:30] And all that kinda stuff but it's repeatable that's what strafe op no the spanner in the works that I was talking about is beg large Global organizations are always making.

Acquisitions so Acquisitions mergers and sometimes getting rid of,

parts of the business that are not making money so they basically just so they can go off on their own or they may just divest them for a couple years and then bring them back in and Wales these are not part of the way they are global.

Standardization or standardized design they can do whatever they want so when they are then brought into the equation that's a spanner in the works and you can and you can see all you want,

you have to have that's that's that's that's and that's done before you can connect to our network but the reality of it is that there's probably not tame and upper management will be seeing,

why is the network not flexible just accommodate for the straight away,

and that's where that's where companies are trying to get from what I've seen over the last four years that's where we're all going but I'll accept that six or eight month project and are,

and a small and medium Enterprises probably a four-year five-year project and a large Global Enterprise.

Zig Zsiga: [47:49] So I think you touched here just now on,

one of the hardest situations to design for and I would say that Acquisitions and

divest stations are the hardest design use cases

I think like building out a new technology or replacing technology isn't traditionally hard it might be complex but it's not hard but when we're talking about stripping away and entity,

like a diver station we're talking about stripping away services and in components to the services I mean a lot of times these companies don't even know what,

is included in the service and so you got to like pull it out right you gotta pull it out and then you got to when you're all done.

Both the individual companies that are left have to be able to function,

and run their business whatever that business is I mean that is truly in my opinion one of the hardest things to do from a design perspective and I mean,

what are your Lessons Learned I'll ask that question for divest stations and an acquisition so far.

Malcolm Booden: [48:57] Oh man the Lessons Learned I don't know there's so many things that I think that.

One thing I'm like Jama complete advocate for design documentation but.

As networks are getting more complex I mean when I started,

when I turn the Cisco networking Academy in some year some time ago which are not are not real not name right but it basically had four.

Zig Zsiga: [49:27] I'll just couple years ago.

Malcolm Booden: [49:28] Yeah so so there is there is four semesters in it right there was one design local area networks there is one design.

I think there was actually a whole semester Ali it weeks on like the OSI model and then there was,

I don't know there was another one that was just I just remember it being like Philly at Ridgely proper how a crimper cable.

I didn't make up make a crossover with the scissors straight through how you use a Multimeter up to troubleshoot problems,

I learned about protons neutrons and electrons electromagnetic interference and repeaters and attenuation and all these things right so if you think back about that and well.

I'm glad I learned about that because it's all the physics behind what we do and you you really go into that level of detail deep but it's good to know like to understand from at especially from a design perspective.

However know how I would summarize forget the protons neutrons and electrons thank you that is translating to all we had to really worry about is routing and switching,

so all we have to worry about is routing switching the SNR.

Spain Leaf Wireless unified Communications.

[50:54] IP cameras we even have like left on the network now whether you're as normal,

and all these things come into come into the equation right so if we go back to.

If we go back to the question and I'll try to answer that what what is the when I was going with that is that.

The previous scenario that youngest.

Explained written in switchin you could rate a design document manually.

And then if something changed you could update it quite easily probably.

[51:39] If you go to the second scenario and that's only some of the stuff that we have to deal with these days and you have to document the network systems so I'm not talking about

the CCTV system is all the digital signage systems their separate the but they just go on the network so the network design has two

accommodate for all the stuff line One Security Wireless decent or CCTV digital signage,

Bowden Management systems and critical critical systems they make the on separate networks but they still have to be documented in the network design.

If something she is if you do that and you get all which we do we do still do we create hlds LL these for all these things and we document that we go through the project we get it signed off it goes any operations and it never gets updated again,

and then something happens or not may not even be read again you know or section might be read like because that's the bet that keeps on feeling or whatever.

My thing.

And Lessons Learned and we are I wanted to take my consultancy so you don't mock my company mnb networks as all right it's all around.

[52:51] For if you go and a new customer it's about understanding what you've got.

Rachel discoveries and I don't mean like putting like one of these old Legacy tools on I mean finding a tool that can do everything from.

Discovery asset inventory in the lifecycle report and.

So you can see what's on your network or two documentation configuration compliance though some of these are bells and whistles and I don't know if you've heard of a thing in there in the manufacturing Industries I called Ed you're old when.

Zig Zsiga: [53:25] I have not now.

Malcolm Booden: [53:27] So it's effectively if you've got a manufacturing process or something and an industrial Network you can because these and those environments are alike.

There's all those things like there's life at rest rather than.

Losing money or you lose money and you could potentially if I machine malfunctions and some days I'm gets chopped off as a result kind of thing,

and but you can effectively have a digital twin to test things in the process before you actually do it on the production,

production line for example know the some tools out there know the network,

I've caught and Donna - and those tools out there and some people are calling it digital digital 20 your network and what it effectively is,

as you can see you can do it discovery on your network like all these Discovery tools or these monitoring tools of being able to do for.

Probably since the at was only routing and switching these.

[54:25] But but no they can they can do that and they can get the naked probably put like you know there's seat looks like this but no will see in tools that can actually document Leah one two and three.

Leader 123 information firewall rules and then effectively create a digital twin so that you could then say,

and CD if I'd done this if I made this change,

and it takes all the config and all that from the database if I make this change I can effectively taste the change on the digital twin and then if it works like it's alright I can press go and then roll out into the production environment,

then stole wait to see if it works and if not automatically roll it back.

I know when you talk about like mergers and Davis the jars and things like that that sort of approach and I'm not.

I'm not going to say like that's that's how you do it but you know I think that hovind cools is as a very minimum very bare minimum to actually go and interrogate the network can understand what you've got.

As just a more imperative now than ever because I've I've done and the reason that this doesn't sound a lot over my 20 or 30-year career of my be done.

[55:39] Donald between 5 and 10 or thats or networks and the reason for that is because it takes three six months to,

gather them get the tools and gather the information let it run Wicked the netflow this in mp the device left and then if it's multi-vendor,

you may only have like a tool from a specific vendor that can give you loads information on there,

cat but they can just tell you that there's something there for the rest of the kit effort happens to have an SNMP stroke so.

[56:14] You know I think like the discovery piece and understand and so that's one thing and then you know tools that can Auto generate documentation right or at least diagrams at layer 1 2 and 3 level so that you can understand,

what would the impact be F that's happened,

I think that's really industry has to go and you can probably do some of that with like you know Python scripts and programmability in that as well so you don't always have to buy a big corporate to in order to do that but,

I think that that's that's where I think that the answer your question is real it's really as you said it's really hard to do,

diabetes and it's time-consuming,

and I think that the way that we have to look at that as we have to flip it on its head and say let's Embrace these new technologies or new approaches to working that you know there's a lot of people doing a lot of good things that aren't there,

programmability species across all vendors including like that includes definite and like everyone's got their own little.

Piece to say about that so it's change and I'm seeing in my customers the global customers that I'm working with that's know.

Is the biggest shift in networking in my career the last thing that I remember being this significant was VoIP.

[57:36] And I include I include.

[57:42] Clothes in that because I don't think Khloe's new I think Chloe's just something that's been around I mean I've got a post on LinkedIn the other day,

that because I was thinking about it and I was like this is truly we have no they'll board or manager proxy server right at the first place that I work one for the whole company 2,000 users,

so used to go back set the proxy and your and unexplored on the point to this one Novell border manager right and then that you stay,

do something don't know what they done something,

if people who just wanted to bypass it and they have an internet connection if we just take effect but like button off in the internet options but if they knew how to do it but anyway that's what we used and then we make it to the first thing we are we still had the text box but,

service code message Labs so message Labs had what they called powers and they had like web voters,

they had to into in Europe to Towers one in Frank for and one in Amsterdam,

but we had the vet a virtual virtual IP or I got you a DNS name or whatever that basically went to Frankfurt first inning and that was in a bit 2007 so I don't think clothes clothes is actually new,

but I think that the shefte that is happening in the networking industry.

[59:10] And the software approach and you know all the stuff about the service provider say stuff being used for segmentation like vxr and,

and things like that all coming in but really the programmability is Aid and getting things done instead of going around 50 boxes and eventually.

Doing that with an invented a and the python script whatever it is you need to do whether it's reading right into the device or whatever,

that's not totally totally changing the way that we work and as I said the last thing I can remember that had that big a chef was going from analog and digital.

Phone light so you use the same the same cat ports but you'd have these like dongles that you plugged into the top five ports in the wall and then you connect the telephone and then you have a separate point for your network cable,

I do remember when we implement it nor tail voice over IP.

And it was like what the the phone can go on the same cable is a network that was that's the biggest shift I think that know what's happening in the networking industry,

it's the biggest shift since then and my opinion.

Zig Zsiga: [1:00:24] I would agree and I want to I want to make a couple comments on the analog shift so you know

before before VoIP was a thing right before the phone network moved into the networking network network a network Inception,

before that happen right I don't I don't know how important the network was.

[1:00:52] I don't you know that that shift I think was the beginning for most environments most companies to prioritize the.

And now I mean now,

it is a commodity like it is it is a utility it is you know you have power yeah water and you have the network you have your internet right you have your your network must stay online all the time and those are just in my opinion unstated requirements now,

you must have a network and it must be up and running and it's a matter of how much availability do you need now like it's how much not that you need it,

you need it like it's but how much do you truly know how much you gonna pay for it.

And that's a huge shift right that's a huge step to VoIP started that and here we are now we're we're talking about cloud and honestly just applications and services right because that's the underlying thing with,

network is getting users to their applications VoIP is an application as well write it runs on top of the network it's a different type of application so the network is literally just getting you

our users to the applications and the data to the application and then back to you and that's really the underlying implications of the network nowadays.

I want that was my one thing right and then my next thing is.

[1:02:11] So we talked about Acquisitions we talked about divest stations and I they are the hardest thing to design for truly they are and it takes a lot of,

time and effort and energy to design for both of those situations but you mentioned something and I wanted to talk about this so a lot of times when you're having an acquisition

the company your business is going to force you to do a short-term plan what is that short-term get get that new that new,

application that new devices new offering online right now

and so you're not going to be able to do all the pretty things that you want to do that you know the normal standardized design things there's going to be single points of failure it's going to be things that aren't optimal maybe you're going to be spanning layer 2

maybe you're going to be doing some funky routing or leaking of routes to get it to work.

Malcolm Booden: [1:03:05] But a good example on something that I've done for that in your major that I worked on,

which was a huge in a multi-billion pound,

or multi billion dollar acquisition I won't mention any more than that but it was two organizations I was on the acquired side but it was in the region of,

hi tens of millions and billions sorry not millions billions High tens of billions of for the acquisition and.

That was,

I read and about 350 seats or 400 seats on I want to say and 700 seats on the acquiring say so we went from thirty five thousand users,

400 sites to then being part of our 1100 user.

[1:04:02] No that's not right.

Zig Zsiga: [1:04:02] The sights right.

Malcolm Booden: [1:04:03] Let me let me let me let me know my mask again so.

[1:04:10] 330,000 I'll get it I'll get it eventually 35,000 the only 5,000 users on acquired side.

70,000 users on acquiring side.

A hundred and five a hundred and ten thousand users and total after the merger.

We are completely different systems completely different networks completely different PCS and this is where that's that's chat could go on for days.

Zig Zsiga: [1:04:44] Yeah seriously.

Malcolm Booden: [1:04:46] Because you don't even like I never thought about it until I was in the position and we.

We're at the point where were like oh yeah we can do - that's in the network however we can't really because.

The Antonis PCS so if you go down to the actual PCS and laptops until their reimaged with acquiring or,

the company who's making the acquisition their desktop image then they won't be allowed to go on the new network now,

what we had to try it in the interest of time I'll try and condense us about that but if you think that you had to completely indifferent mpls Networks.

We had Eleven Hundred sites so the 700,

imagine with a four hundred and a hundred and ten thousand employees the 35,000 with the 75,000.

And we had two completely separate networks one,

it was a pure mpls the acquiring company so the acquiring company had 700,

sites with effectively active standby mpls no went on it the other say the other company,

what quite Forward Thinking and we had.

A variant of sd1 let's see I never lie let's see if you 10.7 friend right.

Zig Zsiga: [1:06:16] Yes.

Malcolm Booden: [1:06:19] And they had overlays and you had a net Bricco say and things like that and you had our wear one of these web proxy things for people to break out.

And they also had the name PLS the merger was that.

[1:06:38] We had to agree or find a way to interconnect so we had three NN is globally,

one in each of the global regions so the America is what actually,

North America in Canada and what time but for the purposes of the nni what Tom had to use the North America and Canada nna and then there was an NA and is your part and then a and,

Europe Inez nnis don't do things if you think of like smaller company and bigger company.

Some survivors from the smaller company if you think of it on the left hand side the small company and the bigger company is on the right hand side and there's data centers on each side.

How it started off as,

we are going to move that's specific application from the left to the right and to one of our data centers so the company who was who was making the acquisition.

And all you need to be able to all as I'm using it I'm using air quotes yet for all that needs to happen here as that we need a HTTP connection to that to that server.

So we are like firewalls and then Network the network cannot connect so that's not too bad right then.

What we had was.

[1:08:02] We want users from that seed to be able to work in those offices and that said to work in those officers so they wanted to then start working between offices.

I will be going to do that okay we need to configure the wireless LAN controllers at those sites to have a GRE tunnel across to the radius server and other network

but that has to go through the firewall at the any and I and all that kind of thing so then you see the complexity start stiff enough of it.

Zig Zsiga: [1:08:34] And this isn't stuff you want to do though like this isn't stuff you want to do from a design perspective.

Malcolm Booden: [1:08:39] Tactical these are tactical I mean we also we also had to re IP the whole network on the left hand side.

[1:08:47] And on top of the airfoil IP conflicts but that's probably one for another day but you know so G see it starts off as right we just need to access that server overheats DPS and that Datacenter fate,

the next requirement was no we need some users going saw some execs going between both directions,

okay that should be we've got an SSID out from the controller and then we you know create this like tunnel over to the radius server on opposite sides so it might be a son once a,

and then clear pass on the other so the so for you and SSID,

Bill Advertiser and I think it gets against your clearpass and on the other side we will advertise already SSID and authentic uh over the nna to the ace,

and then the third thing that we got into a nest was already rolling so it was like no we need you to do that so now we need you to do this and the business requirement is

make the application available was the first one

the business requirement for the second one was we need the exact solemnly execs to be able to go between for me important meetings and things like that and then the third one was,

right now we're going to start doing Bolden consolidation and rationalization,

two of those two offices one for company and one for Company B both in the same city for example they will merge into the bigger office or the one that's got the longer lease.

[1:10:13] The problem that we had as they were both they could be in the same building for all we know but they're on different mpls Networks.

Zig Zsiga: [1:10:21] Oh you gotta get new mpls oh wow.

Malcolm Booden: [1:10:24] So what we've done is on the on the Strategic mpls Network them stay in,

we got the provider to create a vrf for the company that was being acquired,

I did not then you can communicate by can be tied to those routes where communities and then only accepted their roots with those communities enter old Network,

and then from the other side because we didn't we that other mpls was going away what we've done instead there is GRE kind of goes in the opposite direction because then we didn't have to get the provider to actually do the.

Zig Zsiga: [1:10:56] To do anything yeah.

Malcolm Booden: [1:10:59] Do they at the core configuration so that's years the merger acquisition date scenarios and again that's only it's only the tip of the.

Zig Zsiga: [1:11:08] Yeah I mean there's three examples,

there's three examples right there and there I mean the first one might be an easy one right it might be actually an easy one potentially potentially but the other dudes like

okay we have to make it work that's the short-term goal right the business wants it done now right that's that's the key and so you've got to make it work,

you don't always have time to go back and do it right either so it's.

Malcolm Booden: [1:11:32] Well they're not and that's not to say that because the next step after that was like really pee the whole network and yeah,

yeah I didn't actually get an empty because they couldn't order the new mpls link that was on the acquiring company who is making the acquisition you couldn't move the site onto their networking tool the sacred been re I peed.

You know what the reality everything even OT Networks.

Zig Zsiga: [1:11:59] Wow wow hey Ben,

Malcolm this has been truly great buddy I knew I'd get you back on here someday and we'd have a good conversation and I did not know that we'd be talking about global global

scale that were design though any any kind of last-minute thoughts opinions questions views for the listeners.

Malcolm Booden: [1:12:26] I think I think that in the the takeaway is probably as I see those so much more that you can talk about but we need to take keep it within their but then the team skill the key takeaway is that I would say our,

if you ever get the opportunity to work on a global scale.

Network ticket because it might be like even if you're in a service provider or a bar or whatever you get modes of variety and you get that opportunity believe me you will learn,

a law so don't think because it's an Enterprise and you're jumping over that you're like I'm leaving All The Cutting Edge State hey you will learn a lot and you learn,

different things and you definitely grow from it and then the other sort of key takeaways that I would say as,

planning an understanding,

understanding your applications which is always been what applications of you go I don't know that's pretty much the standard question for the last 20 years late,

or 15 years that you get and it's like I don't know but it's never been more important,

but they understand your applications and the floors then no especially in any network but especially in global networks or think between the regions,

and also think that one size is never going to fit all region you can have a framework but you're never going to have like one-size-fits-all design and that's never going to vote very away from that.

[1:13:54] And then the other thing is yeah the scaling aspect to it and you're going to have you're going to have some fun and games,

and it takes a lot longer but it's the final thing probably is like stability excused predictability and stability,

is key but again not doesn't mean to say that you're just going to get to work on Borden stuff you're still going to have to evolve the network.

[1:14:23] Two new technologies but in a stable way but,

absolutely recommend working on these types of networks to anybody if you ever get the get the opportunity and that's coming from someone who used to leak the intricate.

Netiquette.

Less poor TV type designs across multi vdcs and I just moved to this and it's like it brings different challenges so,

that would be my closing comments and if anybody wants to continue the conversation as I say I'm quite easy to find on.

Social media so thanks for posting me like is great to catch up and we'll need to continue the conversation if I see here at Cisco live or something like that.

Zig Zsiga: [1:15:07] Yeah for sure maybe we'll do another one if you're up for it I know you're really busy with everything else going on this could have been like a three-part episode man right like let's be real this could have been like a three or four part episode

so you already said where everyone can find you I do want to highlight real quick you you do have a course can you go ahead and talk a little about your poor your course real quick.

Malcolm Booden: [1:15:28] Yeah so we'll be in the last year or so so it's 2821 moment and we have developed,

I'll train and site really and we have a couple of courses ones called Network design fundamentals it covers hit the high points it covers,

Gathering requirements.

And we also cover Gathering requirements high level designs low level designs how those are structured how to rate them gives you examples templates things like that,

and then also to other topics that people sort of raise their eyes at really Network design fundamentals but migration plans and test plans so plenty of examples and the

I didn't we also have a one design course which has.

And one Technologies so the difference between mpls L 2 VPN LT V PN point-to-point vpls all that and then it covers like using those all together,

HD1 design Global scaling,

of one networks and then one migrations and things like that so that's it that's it in a nutshell it's just training door mnb Network stopped it but you can put the put the link in there.

Zig Zsiga: [1:16:44] Yeah no I think it's awesome and you know I think there's a gap of the industry for a lot of this stuff that a lot of this training and really truly teaching people Network design so for everyone that's listening I will have all those links in the show notes if

interested in learning about Network design I mean we go take a look at Malcolm's training,

it truly is amazing stuff that's out there right now hands down one of the best and Malcolm's are great great instructor,

he's been a great friend to four years so maybe I'm biased man I don't know maybe I'm a little.

Malcolm Booden: [1:17:19] I appreciate that that is begging me up I'll buy your paint I'll buy your paint later on oh good.

Zig Zsiga: [1:17:26] No man we're just passionate about this stuff right so I'd rather have more Network designers right like I want more people in this field like

this is this is great stuff once again welcome thank you for everything I really appreciate it taking the time I know it's late your time I think this has been great

I hope you have a good night man thank you.

Malcolm Booden: [1:17:46] Like very much to speak to you again soon.

Zig Zsiga: [1:17:48] Hey friends nerds geeks and ziglets that's going to close out today's episode of the zigbits network design podcast where we talked about global scale Network design with Malcolm booden

Today's Show notes will be at zigbits. X / 92 my friend Malcolm has just launched his neck,

design course so if you are in the realm of learning to learn real world real life Network design you should go check out his course right now it's training.

[1:18:18] M & B networks Dot

if you want to have live Network design conversations join the zigbits Discord Community there are a ton of Highly skilled experts ready to help you with your network design questions

go to zigbits dot tax / Discord to join it's a 100% free community

if you like today's episode let us know you can find more zigbits network engineering Network design and network architecture content including technical podcasts monthly webinars YouTube videos and a dedicated community,

on this court all this content I just mentioned is 100% free content you can find all of this and much more at the zigbits DOT,

follow us on Twitter at zigbits and find us on LinkedIn sign up for our free Weekly Newsletter the network design digest filled with the best network design content in a network engineering today

that's at zigbits dot text last newsletter as always I appreciate you and I thank you for listening.

[1:19:20] Don't forget my friends to attack your goals attack the day attack your life and make progress until next time bye for now.