WEBVTT

0:00:03.160000 --> 0:00:06.520000
 Hello and welcome to this video, which
 is one of a series of videos, which

0:00:06.520000 --> 0:00:11.200000
 is going to give you a review and refresher
 of quality of service features

0:00:11.200000 --> 0:00:16.840000
 and concepts as pertains
 to the CCNA 200-301 exam.

0:00:16.840000 --> 0:00:20.880000
 So we're going to start with the concept
 of classification and marking

0:00:20.880000 --> 0:00:24.860000
 with QUS. Now, once again, I'm assuming
 that by the time you get to this

0:00:24.860000 --> 0:00:28.740000
 video series, this bootcamp series,
 you've already gone through several

0:00:28.740000 --> 0:00:33.720000
 other courses pertaining to the CCNA,
 one of which relates to QUS.

0:00:33.720000 --> 0:00:36.180000
 So this is really
 a review for you.

0:00:36.180000 --> 0:00:42.120000
 So let's start with what is classification,
 types of classification, and

0:00:42.120000 --> 0:00:43.480000
 then we're going to
 go through that.

0:00:43.480000 --> 0:00:44.980000
 All right, so what
 is classification?

0:00:44.980000 --> 0:00:48.480000
 So in the world of QOS,
 here's your objective.

0:00:48.480000 --> 0:00:54.260000
 You have identified, or you can predict,
 one or more points in your network,

0:00:54.260000 --> 0:00:58.020000
 and by point, I mean interfaces in
 your network, that have gotten, or

0:00:58.020000 --> 0:01:00.120000
 you think will get congested.

0:01:00.120000 --> 0:01:03.920000
 Congestant, meaning there's so much traffic
 trying to leave that interface.

0:01:03.920000 --> 0:01:07.280000
 That interface does not have enough
 bandwidth to transmit everything it

0:01:07.280000 --> 0:01:08.080000
 wants to transmit.

0:01:08.080000 --> 0:01:12.360000
 So it has to store stuff in memory buffers,
 and then at the worst case,

0:01:12.360000 --> 0:01:16.600000
 if the congestion is persistent and doesn't
 slow down, those memory buffers

0:01:16.600000 --> 0:01:18.400000
 ultimately get full.

0:01:18.400000 --> 0:01:21.540000
 And now, any other packets that come
 into that router or switch that need

0:01:21.540000 --> 0:01:25.020000
 to go out, that one interface, have
 to be dropped, because we have no

0:01:25.020000 --> 0:01:27.840000
 more memory buffers to
 store those packets.

0:01:27.840000 --> 0:01:34.000000
 So the idea behind QOS is that if you
 don't have QOS and that type of

0:01:34.000000 --> 0:01:36.760000
 thing happens, you don't know
 what's going to be dropped.

0:01:36.760000 --> 0:01:40.140000
 It could be your voice traffic, your data,
 your video, your mission critical,

0:01:40.140000 --> 0:01:41.960000
 you have no control over that.

0:01:41.960000 --> 0:01:45.400000
 So we want to give a little predictability
 and control, and that's what

0:01:45.400000 --> 0:01:47.240000
 QOS is all about.

0:01:47.240000 --> 0:01:51.200000
 And the first step of that is identifying
 your traffic, saying, so you

0:01:51.200000 --> 0:01:54.980000
 would start in your mind in the design
 process, thinking to yourself,

0:01:54.980000 --> 0:01:57.820000
 okay, how do I want to
 classify your traffic?

0:01:57.820000 --> 0:02:01.700000
 So in my mind, I might come up with
 some ideas like, all right, well,

0:02:01.700000 --> 0:02:04.800000
 my voice over IP is mission
 critical to me.

0:02:04.800000 --> 0:02:08.800000
 If my IP phones go down or my IP phones
 get choppy, that's going to affect

0:02:08.800000 --> 0:02:12.920000
 sales, which is going to affect my
 bottom line, my bottom dollar.

0:02:12.920000 --> 0:02:17.760000
 So I want my IP traffic, my voice over
 IP, to have the highest priority.

0:02:17.760000 --> 0:02:21.520000
 Then let's start making some other classifications
 of traffic like, okay,

0:02:21.520000 --> 0:02:24.340000
 maybe my DNS traffic
 is second importance.

0:02:24.340000 --> 0:02:28.560000
 Maybe my file servers are third importance
 and so on and so forth.

0:02:28.560000 --> 0:02:32.960000
 So now you've come up in your mind about
 different types of traffic that's

0:02:32.960000 --> 0:02:36.220000
 going to be going through those congested
 interfaces and you've sort of

0:02:36.220000 --> 0:02:37.740000
 classified them.

0:02:37.740000 --> 0:02:43.160000
 Now what you need to do is you need to
 translate what's in your head into

0:02:43.160000 --> 0:02:47.400000
 the Cisco iOS command line into the routers
 and switches so they can properly

0:02:47.400000 --> 0:02:49.180000
 identify the traffic as well.

0:02:49.180000 --> 0:02:53.800000
 And then they can do something, they can
 do different QOS actions to different

0:02:53.800000 --> 0:02:56.300000
 categories of traffic.

0:02:56.300000 --> 0:03:01.140000
 So we must divide traffic into classes
 and a class of traffic will receive

0:03:01.140000 --> 0:03:02.560000
 the same type of QOS treatment.

0:03:02.560000 --> 0:03:07.740000
 So for example, voice over IP traffic might
 be defined as a class of traffic.

0:03:07.740000 --> 0:03:13.220000
 And then once we identify, all right,
 how can I let the router or switch

0:03:13.220000 --> 0:03:17.320000
 know that this packet you're looking
 at right now is voice over IP and

0:03:17.320000 --> 0:03:18.840000
 this one is not.

0:03:18.840000 --> 0:03:24.160000
 Well, what's distinctive about a voice
 over IP packet is it the UDP port

0:03:24.160000 --> 0:03:26.280000
 number? Is it something else?

0:03:26.280000 --> 0:03:29.860000
 So once you can identify that,
 you can tell the router switch.

0:03:29.860000 --> 0:03:33.380000
 This is what I want you to look for like
 via an access list or something.

0:03:33.380000 --> 0:03:37.660000
 And when you spot this, this belongs
 in the voice over IP class.

0:03:37.660000 --> 0:03:42.360000
 Now I want you to apply these QOS characteristics
 or policies to that

0:03:42.360000 --> 0:03:44.080000
 class of traffic.

0:03:44.080000 --> 0:03:48.480000
 So we need to analyze the packets to differentiate
 flows and classification

0:03:48.480000 --> 0:03:53.040000
 features. QOS features that are defined
 as classification features.

0:03:53.040000 --> 0:03:54.060000
 That's what they do.

0:03:54.060000 --> 0:03:57.440000
 They give us that ability to
 differentiate the traffic.

0:03:57.440000 --> 0:04:02.000000
 I would say they give the router switch
 the ability to differentiate traffic.

0:04:02.000000 --> 0:04:05.700000
 Just like you know in your mind how
 you want to differentiate traffic.

0:04:05.700000 --> 0:04:11.600000
 Okay. So typically speaking.

0:04:11.600000 --> 0:04:16.220000
 Step number one in QOS
 on your on your.

0:04:16.220000 --> 0:04:20.740000
 Like access layer devices is to see
 all the packets coming in that all

0:04:20.740000 --> 0:04:25.820000
 look the same and then inspect each one
 of those packets to identify what

0:04:25.820000 --> 0:04:28.420000
 class of traffic that
 packet belongs to.

0:04:28.420000 --> 0:04:32.440000
 Now that process of inspecting a packet
 looking deep into a packet, maybe

0:04:32.440000 --> 0:04:36.240000
 be looking in the data of the of the packet,
 the body for certain characteristics

0:04:36.240000 --> 0:04:38.780000
 can be fairly CPU intensive.

0:04:38.780000 --> 0:04:43.120000
 So we don't want every router, every
 switch along the path from source

0:04:43.120000 --> 0:04:45.440000
 to destination to have to do that.


0:04:45.440000 --> 0:04:49.580000
 We really only want one device, usually
 the access layer, maybe one or

0:04:49.580000 --> 0:04:52.840000
 two devices in from that
 to do that process.

0:04:52.840000 --> 0:04:57.500000
 At that point, once that device has
 said, Oh, based on the access list

0:04:57.500000 --> 0:05:01.380000
 or based on this other feature I'm
 using, that's voice over IP.

0:05:01.380000 --> 0:05:02.740000
 That's DNS traffic.

0:05:02.740000 --> 0:05:04.100000
 You know, that's telnet.

0:05:04.100000 --> 0:05:07.540000
 Once it's done that and made
 that classification.

0:05:07.540000 --> 0:05:09.320000
 Now we basically want to apply.

0:05:09.320000 --> 0:05:11.400000
 I could say the word tag.

0:05:11.400000 --> 0:05:16.160000
 We want to tag that traffic with an
 easy number to spot so that other

0:05:16.160000 --> 0:05:17.280000
 devices further upstream.

0:05:17.280000 --> 0:05:19.840000
 They don't have to do that
 deep packet inspection.

0:05:19.840000 --> 0:05:22.620000
 They can say, Oh, I see
 a number on there.

0:05:22.620000 --> 0:05:25.140000
 Number five. Well
 with number five.

0:05:25.140000 --> 0:05:28.200000
 Now, you know as a network administrator
 that number five is because that

0:05:28.200000 --> 0:05:32.320000
 first device. Recognize the voice over
 IP traffic and assign the number

0:05:32.320000 --> 0:05:36.840000
 five to that. Number three, that first
 device recognized telnet traffic

0:05:36.840000 --> 0:05:38.680000
 and assign the number
 three to that.

0:05:38.680000 --> 0:05:42.420000
 So all other subsequent routers and switches,
 all they have to do is look

0:05:42.420000 --> 0:05:44.600000
 for that number to do
 whatever QS policy.

0:05:44.600000 --> 0:05:46.320000
 They're going to do.

0:05:46.320000 --> 0:05:49.660000
 So that's what we're talking about
 when we talk about marking here.

0:05:49.660000 --> 0:05:53.700000
 Marking a packet is simply putting some
 number in there, some indicator

0:05:53.700000 --> 0:05:58.020000
 about the relative importance of this
 traffic or what class it belongs

0:05:58.020000 --> 0:06:04.240000
 to. So the most common ways
 of classifying traffic.

0:06:04.240000 --> 0:06:08.860000
 So what we're talking about here is
 when the traffic comes into a device

0:06:08.860000 --> 0:06:10.860000
 like a router or a switch
 or a firewall.

0:06:10.860000 --> 0:06:15.040000
 What are the most common methods that
 device will use to say, Oh, this

0:06:15.040000 --> 0:06:19.180000
 traffic, I need to apply this
 particular QS policy to.

0:06:19.180000 --> 0:06:21.320000
 Oh, this traffic over
 here is different.

0:06:21.320000 --> 0:06:23.980000
 I need to apply a different
 QS policy to that.

0:06:23.980000 --> 0:06:25.420000
 Well, the traffic might come in.

0:06:25.420000 --> 0:06:27.420000
 It might already be pre marked.

0:06:27.420000 --> 0:06:31.600000
 For example, some end devices like IP
 phones will already take your voice

0:06:31.600000 --> 0:06:35.620000
 over IP packets and apply a marking to
 it right there as they're creating

0:06:35.620000 --> 0:06:37.640000
 that voice over IP traffic.

0:06:37.640000 --> 0:06:40.360000
 So the traffic might
 already be marked.

0:06:40.360000 --> 0:06:44.840000
 It might be as something as simple
 as the layer three address where it

0:06:44.840000 --> 0:06:48.220000
 came from, where it's going or
 maybe TCP UDP port numbers.

0:06:48.220000 --> 0:06:51.360000
 Might be some sort of
 application signature.

0:06:51.360000 --> 0:06:54.400000
 Sometimes we have features that will
 look deep inside of a packet, look

0:06:54.400000 --> 0:06:58.760000
 into the date of a packet and maybe
 look for a certain URL or certain,

0:06:58.760000 --> 0:07:02.900000
 you know, structure of data say, aha,
 that's what I'm looking for that

0:07:02.900000 --> 0:07:04.300000
 matches what I'm looking for.

0:07:04.300000 --> 0:07:08.680000
 So there's lots of different
 ways to do classification.

0:07:08.680000 --> 0:07:12.940000
 Okay, so let's differentiate between layer
 two and layer three classification.

0:07:12.940000 --> 0:07:14.760000
 So if I'm a switch.

0:07:14.760000 --> 0:07:20.060000
 Or even a router and an Ethernet frame
 is coming into me and I want to

0:07:20.060000 --> 0:07:23.880000
 be able to look at just the layer two
 headers of every Ethernet frame

0:07:23.880000 --> 0:07:26.900000
 and say, ah, that is an at frame is
 carrying this is carrying voice.

0:07:26.900000 --> 0:07:29.440000
 Ah, that is an at frame is
 carrying this is carrying.

0:07:29.440000 --> 0:07:37.280000
 Tell that. How can we do that?

0:07:37.280000 --> 0:07:37.980000
 What are the different
 ways to do that?

0:07:37.980000 --> 0:07:43.160000
 Well, the problem we have is if you
 think about the layer two, Ethernet

0:07:43.160000 --> 0:07:45.440000
 header your standard
 Ethernet header.

0:07:45.440000 --> 0:07:48.120000
 All you have so traffic
 is going this way.

0:07:48.120000 --> 0:07:52.420000
 All right, your very first field
 is going to be a preamble.

0:07:52.420000 --> 0:07:57.300000
 Which is just going to be a static unchanging
 pattern of zeros and ones.

0:07:57.300000 --> 0:07:59.140000
 We can't touch that.

0:07:59.140000 --> 0:08:05.020000
 And then after that, we're going to
 have the destination MAC address.

0:08:05.020000 --> 0:08:07.440000
 Followed by the source
 MAC address.

0:08:07.440000 --> 0:08:11.900000
 Followed by an ether type.

0:08:11.900000 --> 0:08:15.840000
 And then then we're done.

0:08:15.840000 --> 0:08:20.440000
 Then the rest of it beyond that is
 your your layer three information.

0:08:20.440000 --> 0:08:24.440000
 Now what I've just drawn right there
 is an Ethernet two header, which

0:08:24.440000 --> 0:08:28.140000
 all IPV four and all IPV
 six packets carry.

0:08:28.140000 --> 0:08:32.560000
 So there's really nothing in there
 that a receiving device can look at

0:08:32.560000 --> 0:08:35.760000
 and say. Oh, yeah, I
 see a marking here.

0:08:35.760000 --> 0:08:40.040000
 There's there's some number or some
 valid field that indicates what this

0:08:40.040000 --> 0:08:45.320000
 is. So if we want to actually
 mark layer two traffic.

0:08:45.320000 --> 0:08:50.760000
 The only way we can do that is by adding
 a field and this takes us right

0:08:50.760000 --> 0:08:54.600000
 back to the concept of VLAN tagging,
 which is what trunks do.

0:08:54.600000 --> 0:08:59.920000
 So layer two classification can
 only happen on VLAN trunks.

0:08:59.920000 --> 0:09:03.460000
 Because whether you're doing ISL, which
 you're probably not doing that

0:09:03.460000 --> 0:09:05.780000
 Cisco's proprietary way
 of doing trunking.

0:09:05.780000 --> 0:09:08.200000
 Or whether you're doing the, you know,
 pretty much what everybody else

0:09:08.200000 --> 0:09:10.840000
 does, the standard way of 802.1 Q.


0:09:10.840000 --> 0:09:16.480000
 Both of those methods add additional
 headers or additional fields to your

0:09:16.480000 --> 0:09:20.620000
 Ethernet frame. And one of those fields
 and both types is used for the

0:09:20.620000 --> 0:09:22.240000
 priority or the marking
 of traffic.

0:09:22.240000 --> 0:09:26.900000
 So if you're doing ISL, ISL
 takes your original frame.

0:09:26.900000 --> 0:09:29.580000
 So right there, that's where your normal
 Ethernet header ends with your

0:09:29.580000 --> 0:09:33.900000
 preamble. And it puts a new
 ISL header in front of that.

0:09:33.900000 --> 0:09:37.900000
 The ISL header has a lot of stuff
 in it, but one of those fields.

0:09:37.900000 --> 0:09:45.680000
 Is called the ISL user field and
 that has three bits of priority.

0:09:45.680000 --> 0:09:48.480000
 Probably not going to see that most
 likely what you're going to see is

0:09:48.480000 --> 0:09:53.680000
 an Ethernet frame doing 802.1 Q on
 a trunk, in which case, as we know

0:09:53.680000 --> 0:09:59.340000
 as a review, 802.1 Q puts a tag in
 there and has three bits of a user

0:09:59.340000 --> 0:10:05.200000
 priority field. So if you want to do
 some sort of marking at layer two

0:10:05.200000 --> 0:10:10.460000
 on your access layer device, your access
 layer device could say, okay,

0:10:10.460000 --> 0:10:15.420000
 here came in a frame that's got something
 in it and using an access list

0:10:15.420000 --> 0:10:18.500000
 or something like that, I just identified
 this as being voice over IP

0:10:18.500000 --> 0:10:20.400000
 traffic. Let's just say.

0:10:20.400000 --> 0:10:25.640000
 So now that device, that switch, I'm
 going to trunk this to a router.

0:10:25.640000 --> 0:10:28.780000
 Or I'm going to send this over
 a trunk to another switch.

0:10:28.780000 --> 0:10:33.040000
 And if you're doing layer two classification,
 as we apply that .1 Q tag

0:10:33.040000 --> 0:10:36.660000
 or the ISL header, now we're going
 to say, oh, if it's voice over IP,

0:10:36.660000 --> 0:10:39.100000
 we're going to go into that user priority
 field, which is normally just

0:10:39.100000 --> 0:10:42.280000
 0 0 0, that's the default
 for all traffic.

0:10:42.280000 --> 0:10:47.000000
 Instead, I'm going to put a non zero number
 in there, like 101 or something

0:10:47.000000 --> 0:10:50.400000
 like that. And then the next device,
 the router switches on the other

0:10:50.400000 --> 0:10:55.040000
 side of that VLAN trunk can be configured
 to look for tags, look for that

0:10:55.040000 --> 0:11:00.180000
 user priority field and apply different
 Qs policies based on what number

0:11:00.180000 --> 0:11:02.640000
 it sees inside that field.

0:11:02.640000 --> 0:11:07.380000
 So when you're doing your marking that
 way by using VLAN tags, we call

0:11:07.380000 --> 0:11:11.000000
 that class of service or costs.

0:11:11.000000 --> 0:11:15.120000
 So if you ever see that term in a document
 or a white paper class of service,

0:11:15.120000 --> 0:11:16.620000
 that's what it's referring to.

0:11:16.620000 --> 0:11:23.160000
 That's doing markings at layer two by
 utilizing most likely, 802.1 Q tags.

0:11:23.160000 --> 0:11:28.860000
 Now the problem with 802.1 Q tags is they're
 only relevant from this device

0:11:28.860000 --> 0:11:35.660000
 to this device. So switch a can insert
 a dot one Q tag, put in a user

0:11:35.660000 --> 0:11:39.920000
 priority field of a non zero number like
 four or six, which can help switch

0:11:39.920000 --> 0:11:45.760000
 B, but switch B then he's going to strip
 that tag off before he forwards

0:11:45.760000 --> 0:11:47.500000
 it out another port.

0:11:47.500000 --> 0:11:50.880000
 Now he might when switch B says,
 oh, I need to forward this frame.

0:11:50.880000 --> 0:11:59.700000
 I just received out
 port 0 slash one.

0:11:59.700000 --> 0:11:59.920000
 So if you want to do a new product,
 you want to do a new product, or if

0:11:59.920000 --> 0:11:59.920000
 you want to do that, you want to do
 a new product, you want to do a new

0:11:59.920000 --> 0:12:02.220000
 product, but it's not going to another
 switch or another router.

0:12:02.220000 --> 0:12:06.200000
 But be that as it may switch B is going
 to apply a whole new tag to that

0:12:06.200000 --> 0:12:08.980000
 frame when sending it
 out another trunk.

0:12:08.980000 --> 0:12:12.460000
 And certainly if the if the egress port
 is not a trunk, if it's an access

0:12:12.460000 --> 0:12:16.940000
 port or routed port, now we
 don't have any tags anymore.

0:12:16.940000 --> 0:12:21.040000
 So this is only really good for when
 you want to do Qs from this device

0:12:21.040000 --> 0:12:23.420000
 to this device. This
 is not a good method.

0:12:23.420000 --> 0:12:28.020000
 If you want some access layer device to
 apply a tag that will carry through

0:12:28.020000 --> 0:12:32.440000
 to the next router, the router behind
 that, you know, 10 routers deep.

0:12:32.440000 --> 0:12:34.140000
 This is not the way
 you want to do that.

0:12:34.140000 --> 0:12:37.580000
 So if you want to apply some sort of
 a marking that will be persistent

0:12:37.580000 --> 0:12:41.620000
 as it goes through the entire network,
 then we most likely want to do

0:12:41.620000 --> 0:12:44.200000
 some sort of layer three
 classification.

0:12:44.200000 --> 0:12:48.240000
 And here we can use another existing
 field, which is the type of service

0:12:48.240000 --> 0:12:52.760000
 byte. Now in IPv4, it's called
 the type of service.

0:12:52.760000 --> 0:12:55.180000
 In IPv6, it's called
 traffic class.

0:12:55.180000 --> 0:13:01.100000
 But either way, these fields were purpose
 built to give us some bits we

0:13:01.100000 --> 0:13:05.900000
 can use to indicate the relative priority
 of a packet to tag a packet.

0:13:05.900000 --> 0:13:11.080000
 Normally by default, every IP before
 or IPv6 packet, regardless of what

0:13:11.080000 --> 0:13:15.100000
 it's carrying, voice, data, video,
 we'll have the type of service bit

0:13:15.100000 --> 0:13:17.840000
 or byte all zeroed out.

0:13:17.840000 --> 0:13:19.180000
 So here's another
 thing you can do.

0:13:19.180000 --> 0:13:22.540000
 You can go into a router switch and
 say, hey, using an access list or

0:13:22.540000 --> 0:13:25.220000
 using some other method,
 identify that traffic.

0:13:25.220000 --> 0:13:29.580000
 Once you've identified that traffic
 as voice, for example, go into the

0:13:29.580000 --> 0:13:35.340000
 type of service byte for IPv4 or
 go into the traffic class field.

0:13:35.340000 --> 0:13:40.120000
 For IPv6 and change it to a nonzero number,
 change it to the number four,

0:13:40.120000 --> 0:13:43.220000
 the number five or something
 along those lines.

0:13:43.220000 --> 0:13:48.480000
 So that's the much more common way
 of marking traffic and identifying

0:13:48.480000 --> 0:13:54.440000
 different flows or different classes of
 traffic is by modifying that field.

0:13:54.440000 --> 0:13:58.720000
 And what we'll see in the next video
 is that that field can be interpreted

0:13:58.720000 --> 0:14:00.780000
 in one of two ways.

0:14:00.780000 --> 0:14:04.840000
 So the older way was something called
 IP precedence, where we said, okay,

0:14:04.840000 --> 0:14:09.540000
 of that entire byte type of service
 or traffic class of the entire byte,

0:14:09.540000 --> 0:14:13.940000
 a few of the bits on the front end,
 I can use for marking, I can use for

0:14:13.940000 --> 0:14:17.940000
 putting a number in there to identify
 the class of the traffic.

0:14:17.940000 --> 0:14:21.460000
 With IP precedence of the
 site's IP precedence.

0:14:21.460000 --> 0:14:24.720000
 And then the newer way, well, newer,
 it's been out for a couple of decades

0:14:24.720000 --> 0:14:28.960000
 now, but the newer way was something
 called DSCP, which allowed to use

0:14:28.960000 --> 0:14:33.720000
 many more bits within that field, which
 gave you a lot more variations

0:14:33.720000 --> 0:14:41.460000
 in the quantity of numbers you
 could put into that field.

0:14:41.460000 --> 0:14:44.400000
 All right, and the last thing I want to
 talk about here in just this overview

0:14:44.400000 --> 0:14:47.520000
 is something called
 trust boundaries.

0:14:47.520000 --> 0:14:55.260000
 So if I am an access layer networking
 device, I'm a switch, I'm a router,

0:14:55.260000 --> 0:14:57.960000
 and I'm directly connected
 to some end user.

0:14:57.960000 --> 0:15:01.560000
 Okay, I'm connected to a laptop
 or a PC or a server.

0:15:01.560000 --> 0:15:06.920000
 If a packet comes into me, and I noticed
 that the type of service byte

0:15:06.920000 --> 0:15:11.280000
 or the traffic class field has already
 been populated with a non zero

0:15:11.280000 --> 0:15:16.500000
 number. Now I have to say,
 okay, wait a second.

0:15:16.500000 --> 0:15:22.220000
 The network, usually the network devices
 are controlled by your network

0:15:22.220000 --> 0:15:27.880000
 administrators, your trusted people,
 hopefully that you've hired to run

0:15:27.880000 --> 0:15:30.520000
 your network and to come
 up with an implement.

0:15:30.520000 --> 0:15:35.280000
 Good sound logical policies that
 are going to be in that network.

0:15:35.280000 --> 0:15:41.240000
 So if a router switch is going to mark
 a packet, that's probably okay

0:15:41.240000 --> 0:15:44.860000
 because some network administrator
 got into that device, configured it

0:15:44.860000 --> 0:15:48.120000
 using a well thought out
 policy and design.

0:15:48.120000 --> 0:15:52.740000
 And so whatever marking that router switch
 applies is probably safe, hopefully,

0:15:52.740000 --> 0:15:55.300000
 unless someone's hacked
 into our router switch.

0:15:55.300000 --> 0:15:58.980000
 But we don't really have a lot of control
 over what our end users are

0:15:58.980000 --> 0:16:04.500000
 doing. So some server or laptop or PC,
 when it's generating a packet from

0:16:04.500000 --> 0:16:08.980000
 scratch, if somebody on that laptop,
 for example, went into their registry

0:16:08.980000 --> 0:16:14.020000
 editor and they changed their
 registry to pre mark packets.

0:16:14.020000 --> 0:16:15.920000
 Do they know what they were doing?


0:16:15.920000 --> 0:16:18.980000
 Do we trust their reasoning
 behind that?

0:16:18.980000 --> 0:16:22.640000
 Are they trying to do something sneaky
 and maybe, you know, we don't know

0:16:22.640000 --> 0:16:24.360000
 what they're doing?

0:16:24.360000 --> 0:16:29.460000
 So with QSP have this concept of trust
 boundaries, where we say, okay,

0:16:29.460000 --> 0:16:32.700000
 interfaces typically leading
 to end devices.

0:16:32.700000 --> 0:16:35.920000
 Do we trust those interfaces
 from a QS perspective?

0:16:35.920000 --> 0:16:40.520000
 If the answer is yes, then that means,
 okay, an end user can mark their

0:16:40.520000 --> 0:16:41.760000
 packets if they want to.

0:16:41.760000 --> 0:16:45.380000
 And when it comes into the first networking
 device, we'll leave it alone.

0:16:45.380000 --> 0:16:48.120000
 We'll trust that marking and we'll look
 at that and we'll apply QS policies

0:16:48.120000 --> 0:16:50.700000
 based on what that end user did.

0:16:50.700000 --> 0:16:55.180000
 Most of the time, you don't
 trust those devices.

0:16:55.180000 --> 0:16:59.860000
 So at least on switches anyway, switches
 have this concept that when you

0:16:59.860000 --> 0:17:04.380000
 enable QOS, a lot of them, not necessarily
 all of them, but a lot of them

0:17:04.380000 --> 0:17:10.100000
 by default, instantly mark all their
 ports as untrusted ports, which means

0:17:10.100000 --> 0:17:15.120000
 that if a user's data comes in that's
 pre marked, that port will reset

0:17:15.120000 --> 0:17:18.980000
 those markings back down
 to the default of zero.

0:17:18.980000 --> 0:17:23.320000
 So it'll sort of thwart or circumvent
 whatever the end user was trying

0:17:23.320000 --> 0:17:28.020000
 to do. And then of course you can go
 on to interfaces if you want to and

0:17:28.020000 --> 0:17:29.720000
 mark them as trusted.

0:17:29.720000 --> 0:17:34.500000
 But usually you want the markings to
 be applied by a networking device,

0:17:34.500000 --> 0:17:38.520000
 not by an end user device.

0:17:38.520000 --> 0:17:41.440000
 So that's what the concept
 of a trust boundary is.

0:17:41.440000 --> 0:17:45.400000
 It's like, okay, beyond this periphery
 here, beyond this periphery of

0:17:45.400000 --> 0:17:50.100000
 ports, anything outside of that, if anything
 outside that periphery marks

0:17:50.100000 --> 0:17:52.180000
 a packet, we don't trust it.

0:17:52.180000 --> 0:17:53.540000
 We don't know why they did it.

0:17:53.540000 --> 0:17:56.960000
 So once it hits our periphery, our trusted
 boundary, we're going to reset

0:17:56.960000 --> 0:18:01.540000
 it back to zero, or maybe we'll mark
 it to what we think it should be.

0:18:01.540000 --> 0:18:04.280000
 That's what a trust boundary is.

0:18:04.280000 --> 0:18:08.620000
 All right, so that concludes this video,
 which was an introduction to

0:18:08.620000 --> 0:18:11.500000
 the concept of classification
 and marking.

