Cable Access SDN Orchestration

Cable Access SDN Orchestration

alright we’re on the air happy Friday
everyone today is Friday February 16th and this is episode 36 of get your Tech
on all things DOCSIS I’m Brady volt founder of the vote firm and nimble this
with us today is John Downey the prognosticator of cable he’s also cmts
technical lead of Cisco Systems John welcome and thanks for being back with
us John say hello I can’t believe it’s like I’m like a puppet right satellite
connection that I saw that you’re sitting on there’s a lot of delay I’m
like a mignonette just my arms get them going also be back also with us is tal
offer a ravenclaw and also director manager at air Estelle thanks thanks so
much for being back with us this month sure thanks Brady thanks for having me
and I resent that I’m a Gryffindor can’t you see that Oh indeed so um today we’re
going to be covering well let me let me get that what we’re going to be covering
a topic that is close to 2 tallis topic or actually one that you picked out it
is cable access Sdn orchestration and so
you know define its Software Defined Networking Sdn orchestration and we’re
going to get in that topic in just a minute I want to cover just a couple of
things in the news so cable apps just release the mac phi standard 4 so we’ve
been talking a lot about our v this is the mac 5 standard that came out but i
want to see if there was any any discussion so we talked about that a
little last week at any any thoughts on that no
we can we can scare no no III can start ready so yeah it’s exciting times right
– we have know so many different options we can offer to the operators and we
have the ability to suggest the integrated CCAP remote spa your most
magnified and you know going into pond later on so all of them are you know
part of the distributed access architecture trend that is growing our
industry the remote magnify is another level of pushing down the functionality
closer to the subscriber it’s got a lot of benefits of course so you know can’t
wait to see that materialize the spec is still under works so it’s important to
mention that there are Eris’s you know it’s actively a part of that as as a
Cisco of course and we’re very you know copy ously working on the details of
that expect to make it available to make it’s ready for for deployment for the
operators I think that’s gonna be a very interesting property you know offer and
value proposition when it becomes ready it’s going to change some of the
existing networks as we see them today so also in a new from oh go ahead Johnny
let me I say from our side you know we’ve pushed remote PHY begin
with for the simplicity you know they keep it simple stupid kiss principle by
keeping the file owned very simplistic fine it doesn’t really change what’s the
spec is done you keep that out in the field not much to do with it it’s very
sturdy the mean time between a long time you put the complexity more a
centralized location I think it’s funny that it’ll tie in really good with what
we’re talking about today with orchestration and automation if we do
Mac 5 which we will also because we want to be technology agnostic and support
the customer the best we can it’s more complexity will be in a node which means
if you need to update bug fixes iOS whatever firmware that’s gonna be highly
automated it’s gonna have to be right we’re gonna end up with thousands of
little cmts is out in the field right you know that the Mac has a lot of
functionality that software driven so that’s gonna have to have updates and
things like that so I think it’s this is a good same way
nope great so the other thing CableLabs is doing is they just pushed out FDX an
AK layer so first they came out with the FBX RFI now they’ve come out with the
FDX mac so that’s continuing to push FDX further along line so again we talked
about that last week there last month now we’re just continuing to see FDX
mature though again we’ll see how FDX continues to roll out so open that up
you guys want to mean that ties it my ties in just like what we said for the
orchestration again because when we do FDX for this echo cancellation idea
that’s gonna have to have some quick automation to keep updating the
equalizer like equalizer coefficients the echo cancellation so that when one
modem transmits another modem doesn’t get interference from that transmission
so your upstream downstream now is going to be full duplex meaning I could
transfer in the upstream at 500 megahertz another modem could be
receiving at 500 megahertz you know the same exact frequency but hopefully you
know they’re not interfering with each other depending on their separation
their isolation and this echo cancellation as well but that
information has to be updated it’s got some type of automation has to do that
updating and how really does that update yeah I couldn’t agree more
John that’s that’s a great opportunity again for for the automation and
orchestration that we’ve been talking about the FD expec relies on certain
clustering of cable modems to make sure that we transmit in different areas and
spectrum such that modems will not interfere with each other and to some
extent that’s gonna be down inside the you know the node or the you know inside
the remote firewall so Mac and fine note but it will also be done in you know the
Sdn Orchestrator of the network to some extent so it will definitely tie into to
the topic that we have today the FDX is you know is a great opportunity for
adding more upstream to our cable operators right so many of
them are seeing the challenges today dealing with the telco offering going to
high numbers on the upstream direction so a lot of our customers are definitely
looking into looking into how they increase this upstream capacity and FDX
is though you know is a great segue it’s a great tool to be doing that it’s going
to be very interesting to see how it’s materializes again because it requires
no 2+0 type of architecture right that brings so and it’s you know with it with
its some challenges on the segmentation of the network but how do you actually
get the nodes closer to the subscriber to end +0 chronic condition so yeah
that’s that’s gonna happen but it’s gonna take a little bit longer right we
don’t we see a lot of demo there’s a lot of talk about it but until it really
becomes you know a real proof a real a deployable solution it’s gonna be a
little bit more you know until we get there
awesome so so the last topic is is you know CableLabs has been busy but SpaceX
has also been busy Elon Musk this comes from science News has that just got
permission from the US communications regulation to start putting satellites
in space specifically for communication broadband camere communication for
people who live in really rural areas like John Downey who are who are stuck
without you know good communications and and so this is you know I saw this and
of course I thought of John and his poor poor video and poor audio that we all
have to do and I just I’d like this well that says the FCC Chairman a jet pipe
reposed the approval of an application by SpaceX to provide broadband services
using satellites in the United States and worldwide so nice thing that SpaceX
is doing hopefully they put a satellite right over your house John very soon did
they say it’s a real alert low-earth orbiting you know see if they say I
don’t want to take up too much time looking at it but
now III don’t know I don’t know for short but I’ll get I can send you later
we heard that Google was doing that a few years back in rural areas and kind
of developing countries right words which was really hard to get with actual
infrastructure to those remote areas they were doing that with balloons at
some point right hi orbiting balloons so you know those satellites is kind of the
next gen evolution of that one I think yeah yeah yeah I know I do remember the
balloons and also planes they were gonna do these these slow flying lyre planes
that we’re gonna suppose yeah broadband access drone yeah like drones moving
that’s moving too fast okay so so tell you have lots of
experience with cable access Sdn orchestration and I have to say this
is something that I have not been on top of a lot so I’m looking forward to this
this episode specifically to learn more about this and I would like to start off
with saying you know what does this mean to the average cable operator yeah oh my
god this could mean a lot right so I can mean there’s little or as much as they
would like it to mean but I think that’s you know if you look at enough analogies
from other industries there is already some Sdn orchestration and more
basically automation let’s make it simple right let’s not talk about the
Sdn and Vee kind of big acronyms let’s just talk about how do you automate
processes right and how do you actually orchestrate processes which means when
you want to manage two types of equipments how do you make them work
together how do you make them talk to one another or have the same
configuration so that they will not you know you know overlap each other or
interfere with each other so think about this in a kind of more simpler terms of
how do we make the network management more more efficient and we’ve been
seeing that happened already in in the telco tank telco business AT&T has been
doing some steps in that direction British Telecom has been kind of
fostering an incubator for for Sdn type of development or you know a lot of
different specs that are starting to come out for the last few years like
Meno and onap and you know lots of acronyms are not going to get into but
the point is that the whole idea of the Sdn orchestration is to make the cable
access networks simpler to manage and more automated and reduce the costs that
it takes to actually operate that cable access network so do things like
automate procedures that currently require a technician to be doing that
and create more intelligence in the network management that some when it
comes to potentially potential use of artificial intelligence machine learning
protocols that we’ll all be able to leverage you know we will all be able to
leverage those to make the access network more sophisticated more
intelligent more self-healing and more autonomous and its operation without the
needs to have a technician sitting there and clicking a mouse or typing CLI
commands all the time so it’s all about again autonomous operation automation
and orchestration ok so that’s that kind of sounds like the 50,000 foot view of
all right right so how does it how does it start to to trickle down from the
50,000 foot view actually into equipment I know last month we talked about
provisioning our fi devices and stuff like that does it actually get into that
level of detail where we’re provisioning equipment or where does it start to
actually impact let’s save maybe a technician or someone in the that’s
operating in the in the headend or like where do we actually start to feel this
yeah so what you mentioned about the discussion from last month about remote
PHY device orchestration or management that’s a very good example for the start
of this whole process right so once we have the distributed access architecture
going into the field we certainly have so many different active components
sitting in the field out there and need to be managed you need to be managed
automatically right because you don’t want to go and physically change the
configuration on each one of them and remote Y devices are going to be the
first type of distributed access architecture devices
that are going to be deployed out there so what we talked about last month I’m
gonna be kind of repeating that as a reminder that we were talking about an
application and a manager that basically on boards those remote PHY devices when
they’re in the field you know make them work out of the box just like you have a
modem that starts to work when you put it out of the box and connect that to
the network you want to make the remote PHY devices work out of the box
immediately if input into the network and we want to allow the operator the
cable operator to manage that in an efficient way such that they will
remember exactly where each node is located what software version is on it
what configuration it has it can monitor it States it can change its parameters
like power levels etc remotely and in an automated way so this has kind of a very
good example for the first step right today we manage C MTS’s or C caps using
CLI potentially sometimes we have SNMP nibs as the management tool when we go
to distributed access architecture we suddenly need something bigger than that
something that will be the overarching manager for all those smaller devices
out there and will basically be the first step into the Sdn orchestration
type of product Brady let me give you an interesting example
that I’ve heard where things you could go a different direction for capacity
planning it’s almost like load balancing is there load balancing between balance
rooms and up streams we’re load balancing between actual RPGs before you
know the kid webisodes and Q operators with se-ra my node hit 85 percent
utilization over a certain time frame so I need to do a node split I either do a
physical notes wit or I add more frequencies and do load balancing I call
it you know virtual splitting versus physical splitting
now if the RPGs are off a digital network there you don’t physically have
to combine and separate RF connectors you can have an automation process that
looks at all our PDS and say alright this our PD is underutilized because
their work Oh area tele five customers this RPD is
over utilized because it’s a couple college kids maybe I can combine those
two to kind of share resources and they’re in the same service group I see
some RPGs clustering RPGs are oversubscribed
I could either assign more license to it or more capacity somehow its physical
frequencies turned on or I could maybe separate that RPD from the other five
it’s combined with and just put it on its own service group so I can reassign
resources and do capacity planning on the fly I think that’s kind of going to
be a cool you know way to you know quickly change capacity anywhere as we
do today yeah I mean it sounds a lot like the use case that you’re talking
about John it sounds like what what is done a lot in in IP networks with like
load balancing where you do you don’t know how much traffic something’s going
to get at different times but you know that it Peaks out so you put a load
balancer in front and when a lot of traffic hits at that load balancer
naturally just distributes that the yet the traffic evenly to the different
devices behind it so is is that I mean that’s kind of what I’m understanding
you’re describing yeah I mean that I think that’s the power of going to
distributed access architecture with digital links now instead of you know
analog optical links getting rid of all the analog now every hour PDS are off
the digital links it could be over Metro Ethernet so I mean yeah you can reassign
resources on the fly I thought another cool applique on our the CTS is today
are pretty resilient for DOCSIS three one profile management you know it’s
it’s not crude but it’s it gets the job done but if we want to take that step
from other where you know you’re doing 4,000 subcarriers maybe 8,000 sub care
houses three one block to a cable modem and you want to monitor the MOR of all
those sub carriers of every single modem every 30 seconds or maybe 20 minutes or
every 30 minutes you need some type of orchestration or application external to
yes to monitor the information and assign profile many profiles that are
different the same TS can manually statically create four or five of them
and the Cape modem can hold on to what is it nvram holds on to fine I think but
what if I want to support 10 different 3-1 profiles because maybe you have a
little bit of roll-off maybe Tao has you know some problems at
the low end of the frequency maybe I have a problem a suck out right at 585
minutes maybe someone else has LTE interference at 713 makers so you know I
could have this orchestration just automatically give profiles left there’s
no right all over the place and then change them if the frequency changes
later on no you understand I’m saying yeah YouTube channel that does the owner
look balancing could also be done on between the Mac and defy right so for
example if we have a certain Mac core that has got too many remote PHY devices
Hank hanging out of off of it we can potentially use orchestration to load
balance those are pd’s to different neck cores part of the strengths of remote
PHY architecture is that the math cores are centralized potentially in data
centers or centralized location and you are reusing those resources all across
her network right no matter how far they are so the load balancing act in this
kind of context could also be moving remote PHY notes between mac cores that
are overloaded as an example and that’s the you’re providing congestion
preventing you know kind of limitations on the network that it can come from a
static configuration so to speak and in another example John I really agree on
the profile management side and I think this is potentially something dear to to
Brady’s heart the the PNM capabilities and the abilities we’re getting from the
doctors we talked to on specs are ones that we could potentially leverage with
additional level of orchestration as you said the you know the the profiles that
can be used on the network can be really you know can really change on the fly if
we want them to and if the network is seeing different types of SNR’s
different type of noise in different areas of the spectrum and different
times a day and this can really change throughout the day because the
temperature changes for example right there are certain devices that produce
more noise if they’re if it’s hotter if it’s colder so potentially changing
those profiles on the fly even throughout a day will allow the
operators to get more bandwidth out of their existing plant right getting more
bits per Hertz based on the just exactly you know existing architecture and
existing infrastructure the fibers and the coax as they have you know dug deep
they can just get more bits out of that existing Network which is you know it’s
priceless it’s an it’s an amazing development to be able to do that so if
I’m a field tech this orchestration is basically like dynamically doing node
splits for me when on-demand when I need it if I’m a DOCSIS engineer it’s
basically dynamically adding and deleting cmts is on demand as I need it
that’s managing my my my modulation profiles for OFDM as I as I need it
dynamically of course if I’m me and I’m a PNM guru it’s it’s also helping me be
much more intelligent about the way that I’m managing that okay so I’m starting
to get pretty excited about this orchestration so now so let me let me
take that a step further right you know if we have the ability to move to
virtualization as well right we talked about that last time and I’m sure it’s
gonna be discussed in the future as well if you have virtualized Mac core then
you’re orchestration network can basically spin up new Mac course on
demand if you see that you don’t you need more Mac processing capability out
of your network and your Sdn controller can basically just spin up a new virtual
machine right with the instance of that virtual Mac core connects that to a new
Naruto to existing set of RPGs potentially as part of the load
balancing mechanism and boom you’ve got a new cmts basically and you got a new
type of access architecture on the fly and completely automated isn’t that you
know magnificent as their as an option so it can it can spin it up I yeah that
okay let’s get it’s getting pretty late yeah so yeah you can just dynamically
create it and add it in if you have a licensing of course but probably later
where does okay yeah go ahead John I’m sorry my second delay on my satellite I
had to start my I said are you even say anything
the you’re excited about it but what about the field tech that looks like
he’s losing his job there’s no RF in the head end it’s all
digital they there’s no notes would argue that that might be part of the job
that the text hates because the notes but they have to do from twelve o’clock
at night so 5:00 in the morning you know so that might be a good thing it goes
away so I think when we look at you know supporting a cable network there’s
enough work to be done that even this automation stuff is just going to
alleviate or relieve some pressure so they can get the other stuff done that
they usually would do you know the optics don’t need to be done it out in
the field Ron Radek would still argue to still
sweep you know a node plus zero I would say that we could use full
bandwidth capture in lieu of actual sweeping but maybe there’s still a case
to use you know frequency domain reflectometry to find suck outs and
reflections but we could still use PNM for that stuff as well
so it’s definitely there’s going to be a lot of work that’s still there for the
field text the RF Tech’s the head end text might transition from a typical
heading RF guy to really an IP guy or a digital guy optics guy you know you
still have optical levels that need to be maintained it’s not this stuff
doesn’t this is not Skynet yet you know this is not Cyberdyne networks no and a
lot of the functionality is going to be basically decision supporting systems
right not necessarily ones that will do and execute the change in the network
for the operator but just recommend what they should do right potentially we
would create optimization systems that would say okay this note split should be
considered for splitting right not necessarily executed that immediately
and in the end it’s going to be a man there that will have – or a woman yeah
that will have to decide you know when is a good time to actually be doing this
this kind of activity and they will have to kind of harmonize that with the rest
of the things that that operator is planning on doing right so it’s going to
be a lot of you know decision support systems does orchestration also get
involved in configurations is a lot of a lot of the equipment now is has the
automated attenuators automated equalizers does orchestration get
involved in that so you know as far as like changing configuration of the
equipment itself yeah I think I think so I mean I am pretty positive that’s gonna
go in that direction the change puts figuration is something that should have
been long automated let’s put it this way right I mean the fact that we’re
still using CLI in 2018 these days it’s kind of ridiculous a lot of the other
industries that that are parallel to ours have gone to to the automation
direction then we must go there ourselves otherwise we’ll lose you know
we’re loose for our market to some extent the actual automation requires us
to change configuration for C MTS’s for amplifiers for nodes on an on-the-fly
basis we needs to be able to react fast to changing environment to noises on the
to noises on the network and to potential changes in the service that we
have to do rights for example an operator a cable operator needs to be
able to very quickly spin up a new VPN network for a new business as a customer
they need to automate that because the customer the the new business customer
will not wait for two weeks until they spin up their new VPN or MPLS network
right they need to be automated so that the operators can can offer a more
flexible and a faster reaction to the market demands sure so really you saw
our demo at the STX row with the item the inode got away from pluggable eq’s
pads they’re all internal all automated I first see that being you know I set up
my inode according to your specs your RF output levels your kilt in the magnifier
in the warehouse and then I give it to the subcontractor the contract the RF
tech field and you know install it but it’s a pre setup and then maybe they use
a QR reader or something to scan it in and tie it to the lat/long GIS location
but it would be all automated pads any cues you know we had for many years very
crude and we still have it a AGC an ALC automatic level control
based on a single pilot which you know that single pilot if it goes down you’re
screwed for all the other frequencies so it’s like we need something that could
monitor you know the RF output it maybe every single RF channel and then change
everything to be exactly the levels you need I would say that the upstream is
going to be more of a problem if we go 85 mega Hertz because the temperature
fluctuations there’s no ADC in the upstream the only AGC we have in the
upstream is the fibre or the cable modems in the cmts negotiating their
levels so I mean 85 merits you could have a big swing at 85 makers now we’re
talking two or four megahertz now we’re talking FDX up to 500 600
megahertz in the upstream you know coax is a killer and temperature alone coax
is a killer so if we limit our coax we’re much better you know if we control
the temperature and I’d be great but we go down that route we could talk about
temperature change I guess that’s where I was going with my question about does
does this orchestration have the ability to make those changes in the amplifiers
and you know normally we would say that would be a bad practice we why would we
you know once we set up and sweep the plant why would we ever change the pads
or the equalizers well I mean one one concept would be if we had something bad
happen to the cable plant where we have degraded cable and we have modems start
to go offline this this this is kind of like the concept of a pre equaliser in a
cable modem where that pre equaliser compensates for something that’s bad
happened to the coax or maybe it’s in a subscribe
home the pre equalizer in a cable modem is changing the way that cable modem is
transmitting it may cause the cable modems transmit at a higher power so it
can compensate the same thing might be happening in the amplifiers where we
with the orchestration we can see that suddenly all the cable modems have a
drop in receive power in the downstream right we don’t know why and on a certain
leg but maybe there’s damage the squirrel was chewing on the coax cables
so now we have the ability through orchestration to increase the transmit
power on the amplifiers temporarily well so we can get those motors back online
and set-top boxes until we can go out and find and repair that damage cable so
I mean that’s something that’s possible you know if you think about that Brady
is because the if we’re doing an RPD remote PHY device the RF is being
generated right there at the node we have control of individual RF channels
not like changing a pad across the spectrum or an EQ to the tilt we could
control each individual RF output so if we notice that maybe and this could be
hypothetical like once we hit 50 or 70% of the motors all reporting the same RF
channel bad maybe the other 30% art but maybe that’s enough for me to justify
turning it up at the carpeting it might be hitting hot on the other 30% but the
70% are now getting proper levels you understand like maybe yeah you’re gonna
have some criteria and this is where I think that you know the machine learning
or the intelligence that Murrow mentioned that’s in the orchestration
would come into place it’s a if if this happens then we can come in and make
some adjustments to get her subscribers back online until we can get a tech out
to find out where the problem is fix the problem yeah exactly that’s that’s what
integration of different data sets comes in right so we should have data points
from the amplifiers from the notes from the transceivers inside the head ends
from the edge qualms from the cmts if you combine all that those data points
together this is where the real synergy comes in this is where you start really
benefiting from all the data that you have collected because you can cross
those data points they know exactly what’s wrong so for
example if you see that a specific TV channel is showing some some problems
you can really track that down up until the head ends and see whether the
problem is with them look Sarah with the transmitter with the actual amplifier
right this is a squirrel let’s shoot it up and you basically can look at every
point down the chain and see the power levels potentially for the you know the
signal that is carrying that channel and potentially getting to even higher level
of accuracy on your troubleshooting and even self-healing system right you can
automate that just imagine that you get some reports from a few set-top boxes in
the same neighborhood and you know where they are right because you know where
everybody is located and you see that everybody got a sudden drop in power
because something happened to the actual cable you can create an automatic
reaction for the network a self-healing network that will automatically just
bump up the power just automatically and a couple of DBS and see if that fixes
the issue just imagine how how great would that be that is kind of the idea
of the self-healing system that is also coming with with orchestration and how
to tell anybody that it’s not that far from reality because we’re seeing some
first of all when we moved to not plus node plus zero architecture there are
less amplifiers down there right so less of less of those amplifiers to manage
but it’s gonna be a while until we reach to not plus zero all over the world
obviously and up until then we see that some you know there are products in the
markets that are actually offering amplifiers with cable modems and sound
inside them try transponders inside the amplifiers which basically make them a
communicating device for every every purpose right you can go and configure
them just like John mentions remotely you could change the settings remotely
and you can again fix problems in your network through that very cool very cool
so where does orchestration reside like where does the intelligence for
orchestration reside in the cable network so it resides in you know new
products potentially that will be the managers of that network right the
Grandmaster Orchestrator or a specific purpose type of Orchestrator
that would be sitting inside the the head end to centralized I guess
management’s network off the operator and all probability is going to be a
virtualized product right we’re going to the virtualization direction so it’s
going to be some kind of an intelligent manager sitting on a docker container
potentially a virtual machine sitting in off the shelf so you know off the shelf
type of server or Hardware potentially that can can host it and we can also
scale it you know very fast if it’s completely virtualized you can spin up
more instances of that manager if you have if it is completely completely
virtualized but you know some of those functions already exists in our cmts and
seek apps today if you think about it right some of the load balancing
mechanisms that we all implemented in our see caps and CMT SS are already
party partially as the N type of functionality because we’re load
balancing physical network elements such as modems or you know or nodes etc and
we’re changing the way they are configured we’re changing the way they
are being set up so you know if we think about it in a more philosophical way
it’s all over right it’s already existing and it’s in all of the existing
products but the ones we’re talking about now are more specialized and even
more intelligent and will be sitting higher up in the RF hierarchy of the
cable operator so John if if you so we have load balancing in the cmts right
now would you envision some day a lot of the functionalities that we you know we
would expect in the CNCs or C CAP to be just just take load balancing as tile
just mentioned with that with that functionality eventually move to work in
it orchestration and maybe some of the the intelligence that we typically have
in a cmts that just be migrated into orchestration and see MTS’s see house
would become us intelligent it’s not if it’s when I mean that has to occur honey
it has to I mean I’m not afraid of losing my job best practices so the unique situation
I’m in is I’m kind of like it liaison right
I go around all these different systems throughout the world then go back to our
own engineers and say hey this is what we’re seeing with Comcast and charter
and net in Brazil and Rogers in Canada or whoever and say you know we need
there are some default settings on the cmts
this will get you by you always keep it simple you keep it the low-hanging fruit
I use all these metaphors for no reason they don’t even apply right just throw
them out there and so we we have these defaults but then when you get in real
life real-world situations we’re like you know be better to maybe change your
parameters to be self-healing but you always have pros and cons you got to
tell the customer here’s what we could do now here’s the trade-off there’s a
little bit more overhead and less cmts there might be a little bit more
overhead on your Alex trains or your capacity drops a little bit because
you’re going more station maintenance or more monitoring more information
transmitted back and forth but it’ll give you maybe self-healing it’s kind of
like you can buy insurance for anything but you know how much insurance do you
really want so you have to have the best of both world you you’ve gotta balance
it off compromise so if I do all these best practices over the years maybe I
don’t have to do that next year when automation does it for me you know it’s
gonna monitor the health of the plant now I might have some best practices of
what I would recommend is thresholds that you put into that automation
because the automation you have to have some type of thresholds that you
configure to say when I should change my capacity when I should change my RF
levels you know we talk about quality of service but really that was that’s been
changing right we call it quality of experience qoe you know I don’t care
what the capacity is saying on a fiber node because I don’t know if that 85
percent utilization is 85% on a five-minute rolling window or an
instantaneous 85 seconds at a Laker herd 4.5 seconds are you understand I’m
saying like how can is it really can when I used to turn on power boost you
would freeze it I can have my scent you keep everything good but no one’s no
one’s complaining so there’s a case where quality of experience was great
but it would indicate I needed us a note split but that’s not true you know so I
need to test quality of experience don’t really worry about what capacity is
because everyone might be happy anyway but then I might say alright with
quality experience I’m seeing some congestion some now I’m sure suggestions
easy to track down some suggestions a little bit more difficult but you could
monitor information SNP mids and noise and then say alright when I hit a
certain threshold of cues that are backing up add more capacity I offered a
doc to three one speed I’m more concerned about peak rates and instead
of trying to offer an aggregate pipe that’s 2 X of that peak rate may be
fully one customers asking for that peak rate I can get away with 192 megahertz
of OFDM block that’ll give me one cake service but I
would never really want to offer that to for customer sharing that that block but
it’s only one customer so I can afford to only turn on one block but then with
three more customer sign up I have to now quickly turn into two blocks
so those four customers can share the two blocks and be better statistical
multiplexing better probability less you know collisions and all that I mean
that’s the case where automation which they are when can ice you talking about
spinning up no Mack buys it stuff like that I’m spinning up more capacity by
just telling the RPD to turn on more Ora frequencies RF frequencies that’s
redundant RFC hopefully Ross I gotta correct
myself dan okay so I so I don’t you actually
think that’s the app load balancing punch in itself will move when you talk
about a little balancing Kevin what it’s inside a service group I don’t think
it’s gonna move you know to the orchestrator as we talk about it right
now because it is kind of part of the function that is the stamp gas or C cap
or VC cap type of functionality and it’s
important to also keep you know though functions inside there the right place
where they can bring the value but what I think is going to be orchestrated are
the level of quality of service side for example service classes and the actual
tiers of service that can be orchestrated from the top thus providing
the input to some of that load balancing right so potentially we have a bunch of
customers who want to change their SLA is right we want to be able to change
the the quality of service they’re getting or we’re seeing that a bunch of
customers aren’t seeing a bad quality of experience right we don’t necessarily
you see the congestion but we know that the quality of experience at the
endpoint like the TV that they’re watching on Netflix is not good then we
know that we need to fix something potentially in their service tiers and
we can go in and orchestrate different parameters inside the sea caps and the
course that there are in the network to optimize you know the service and
quality of experience for those customers that’s where the actual the
actual benefit is going to come from from integrating that level of
information from from the top I agree I kind of tiptoe around there I think the
question to begin with but yeah the load balance itself to low modems single
channel tuners they got a load balance I think orchestration as the end all this
allows us to optimize the plant so less load balancing is needed and a case in
point would be also today into paying eros or Cisco what if a modem is an
8-channel capable modem and there’s 24 downstream channels so it has three
groups of eight to load balance between DOCSIS 3 or load balance what if the
modem moves to a higher set of eight frequencies and that house has roll-off
in high-end Sdn and orchestration could keep track of that modems signature and
say you know what I will allow you to load balanced but never to that last
group of 8 because I know that house has problems at the high in frequency I’ll
allow you to hello bounce between the two groups of 8 on the lower end but
exclude yourself from the last group of 8 today a lot of that
is not smart the same test might say all right you can move oh but now you have a
cm status message that said you have a problem so I’m gonna put you in partial
mode so now that modem went to seven channel bonding instead of just staying
a channel bunny on lower frequencies yeah or or it would even be it into the
failure list because it tried to move it and it would just be a failure correct
so I have to fail three to five times before it actually gets excluded from
load balance yeah and then once it gets excluded from load balance it might not
load balance routine the two other groups of me it’s stuck it’s stuck on
ask on its group because I have seems that exact scenario happen don’t
orchestration to keep track of all that and say I know it’s house MAC address
cafe whatever the MAC address is that’s a joke by the way in a beef cafe so that
when that MAC address I know his signature for his house and I could make
his own group or load-balanced group just for him yeah that would actually be
seemingly a much more intelligent way of handling it so who is driving
orchestration is this being done specifically by vendors is it being done
is it a standard being done by cable labs what who’s making these specific
decisions of Oh we’ll handle load balancing in in
orchestration or will handle load balancing in in the cmts
or oh we’re going to start handling cable or us node and amplifier
transmission increases when modems go offline like those are really really
complex and critical decisions for a network so it who’s making these
decisions so so first of all I do think that’s the driver is just emerging from
the need of cable operators to simplify their network right to automate as many
procedures as possible and to make it easier to use
they just need you know help with managing their network as I said earlier
the distributed access architecture devices out there are just becoming too
many to efficiently manage without automation and orchestration tools so in
order to really make use of the new technologies that
that were coming up with on the exercise it really must come with another you
know set of tools on the orchestration and network management level to really
be able to benefit from from from that new technology so you know customers
operators you know all over the world are asking for additional tools to do
things more or more intelligently and there are no cable apps specs per se in
terms of the orchestration of the network right when we talk about network
orchestration then it becomes a much you know kind of excess agnostic type of
question you need to manage a network whether it’s deer cell it’s excess cable
access it’s you know it’s pond you need to be able to manage a complete network
and for that there are some specs that are mentioned earlier the the own app
that man oh the oh noes type of your products out there so a lot of different
specs that exist to try to define at least the protocols and the interfaces
between the different machines in that you know and that domain so it’s
starting to come up with that you know starting to rise as aspects but in our
cable industry it’s – it’s still far from that because every device is
speaking another language right we all part of we’re John and I are part of the
vendors that are basically doing that right – creating boxes that speak their
own proprietary language to some extent some of us you know some of the box of
stuffs you know speak the DOCSIS mips according to the cable up spec which is
really good and it’s a good start but there’s still a lot of proprietary
interfaces out there so there’s still not you know not standardization not
about standardization out there that we can use to make orchestration work so
every orchestration solution that is being built by one of the CCAP vendors
or by other companies needs to be able to talk to a series of different devices
with their own proprietary interfaces which you know brings in one of the
bigger challenges of actually orchestrating the network right you need
to be able to speak to ten types of devices each one of those picking the
different type of protocol or MIPS or CLI commands so hence you know one of
the challenges for an efficient solution in that in that
area do either you see this being something that MSOs will drive to be
standardized to become like a orchestration spec i if I were an Emma
song I would do that years ago because I think there’s a lot of value for you
know standardizing the interfaces to different boxes right net conf and yank
models have been discussed in a where originally part of the C CAP spec if you
remember that the C map C CAP Caesar whatever name you call each one of them
had and had a chapter that prefers to this net comp yang type of model to talk
to a C CAP that never really happened right we vendors didn’t really implement
that because it was not needed it was not high on our priority list to do that
but eventually if we really want to be able to orchestrate the network in the
unified way there was going to have to be some kind of standardization or at
least a common ground you know for all of us to be talking to similar language
I mean I agree I mean you talked about who’s driving this pretty MSO so I mean
it’s an open market supply demand demand comes from the msot EMA so today we
would like this ends up showing up an RFQ and then we all been against each
other for the business who can support what and maybe some of us proprietary
but you know it would be nice if there was real interoperability you know the
joke is that’s a great thing about what there’s so many standards but they’re
not really standard so I guess it’s not see more on at that point yeah Nick and
I support a standard ya know kid and I thought it’s kind of
funny I heard I think one episode went to cable ad said hey can we standardize
amplifier housings and endnote’ housings I’m like late to the game and that 50
years thank you for that yeah and we’re going to note plus zero
so you know maybe they want to standardize the note itself you know fit
form factor and all that I know we’re all the standard for power is their
meaning they’re trying to push for what less than 160 watts per fiber note so
it’s trying to get a lot into a smaller that’s the big thing about you know the
remote max by if we put the Mac in the node let’s give it what’s gonna happen
with the power of Armagh yeah that’s that’s gonna be concerning so so
where do you what what is the the big lat so I can’t I can’t think of the
analogy there the crystal ball thanks John you’re you are the man with the
analogy what does the crystal ball say the future is magic 8-ball so what’s the future now I mean there’s
a lot of cool things here you guys have talked about with orchestration what
does the future say like you guys said can you can you can you kind of project
what what are even more cool things with orchestration have you yeah I think that
we’re gonna start seeing that more and more you know starting this year already
with different set of functions there’s not gonna be you know when Ben thank you
man type of solution it’s here I’m gonna put it in and it’s gonna manage my
entire network no it’s not the way it’s gonna work right it’s going to be a type
of a growing solution a type of evolution we will add more functions
that will orchestrate different types of pieces in the network right different
segments of functionality so the example for management of RPGs right remote PHY
devices out there is going to be a good start
we will add you know profile management for DOCSIS 3.1 oh those are you know
certain pieces of the puzzle that will add more and more
orchestration so we gradually are going to see more automation and more
orchestration introduced into the network and you know becoming more and
more more and more intelligent and as I said earlier the more data points we
have from the network the more collection points we have the
better synergy it will be and the better type of value that we will get from
those types of orchestrators when as as much as they can pull more points in the
network so I think it’s going to be you know an enable an evolution not as much
a revolution so it’s going to be incrementally growing in terms of the
the usages of the orchestration solutions until we get to a point where
it’s really more and more automated and almost you know self-healing
self-managing type of system which is kind of further into the future right
it’s more of an Elan mask kind of fishing I would say ela would be happy
yeah I I agree I mean you’re not gonna rip out an existing Asian of steel
plants especially if you’re a publicly traded company on Wall Street you know
it’s gonna cost way too much money and then your stocks are so you’re going to
almost cherry-pick and some people would argue well I can cherry-pick
cheap on any pond and some people have right you cherry pick a customer who
wants to do business from their home and you do any pond solution if you have the
back office for it and everything else maybe you do doctors over a pond I don’t
know but you can offer very fast doctors 3-1 speeds and compete against Jeep on
and look at the price points you already have an HST plant maybe you only upgrade
one leg of a node maybe you only upgrade one node to RPD and that gives you the
capability to do higher modulation for 3:1 higher speeds just for that section
of your cable plant and then you just keep upgrading as you go so that would
be you know evolving your network pay pay as you go okay pay to play see
metaphors just throwing about hey you grow whatever it is so I should do that
but you know talking about the crystal ball the magic 8-ball whatever the
Internet of Things Internet of Things is like everything is connected in your
home and people have talked about this years ago about you know 20 over in the
garage the car automatically connects to your network the car downloads the
latest songs the into your car maybe maps updates to the
car itself your dog gets updates with updates might be you know just the tags
and stuff and uh who he’s been around I don’t know but you get the idea is
there’s a lot more devices that are connected to the network and so we have
to make sure the network is automated enough to find provide capacity and ipv6
and everything else from from a security standpoint I can see huge opportunities
there where you could start pushing all the latest security updates to every
device that’s connected to the DOCSIS network that would be huge firmware
updates to millions all the things that were that we’re struggling with now yeah
you’re talking about security of devices like attacks on your device and
cyberattacks but a security of your past home security home automation especially
when you bring up and that could be the cool rap yeah I mean bringing up IOT but
even just down to the cable modem see where yeah if we could orchestrate that
for operators to make sure that you know when Aris or when Cisco releases the the
next security update that that’s that’s quickly pushed out and and ensured that
it’s also very well tested yep that type of orchestration and and if it’s not if
there’s a problem when it’s pushed out that it’s immediately rolled back those
are those are the types of things that they could be I see be very well
orchestrated because right now we you know we know that there’s Security’s
device there’s many devices out there that have holes in them and people are
worried about pushing firmware updates out because we have issues so right so
imagine that you do that so for cable modems today to some extent you know
you’re limited on that on that front you would be able to do that for remote PHY
devices remote magnify devices right download the software upgrade that will
prevent the security hole that you may have or run another level of validation
on those cable modems to prevent this latest attack right the ability to
configure things download software versions to those devices is a very
powerful thing we’re all benefit from yes
absolutely so any any final thoughts on on orchestration I think we used up our
analogies but just again to emphasize at the point it’s definitely not you know
it’s it’s not a revolution it’s going to be an evolution so operators and vendors
are going to be have to having to pick and choose which automation functions
and which orchestration functions will provide the most value in the short term
and apply them already you know this year next year and later on grow on
based on those on those in kind of automation functions that are already
introduced into the network all the infrastructure of those automation
functions and collect more information make it more and more intelligence up to
a point where it’s you know a really intelligent network manager John maybe
from your side you know you’re talking about analogies and metaphors and all
that and you and I both worked at you know Secor Electronics way back in the
day and there was an engineer one time that says you know John if we make
everything idiot-proof you have more idiots that work on it so my concern is
when we make things all automated and then we forget what’s really happening
in the background we lose sight of you know the basics and this is why Ron will
have a job forever teaching a basic enough stuff from soup to nuts you know
going through it all and explaining what’s really going on
because if everything is automated and you don’t understand that that
automation has the wrong parameters in it you might not question it you know
we’ve had cases where people would do an FCC proof and say oh this note came out
with a 60 DB CNR it’s perfect it passes I’m like wait away if it’s too good to
be true then you should be questioning why it’s 60 when a fiber node with a 10
DB optical link budget should only get like a 48 or 50 you know according to
the spec so how is it getting a 60 obviously you did it wrong but they
don’t question it right because they got the good numbers so
we need people to understand how things still work and not just throw people at
it once it gets automated so we still need the expertise we still any
education this is not going to be a blurb for s CTE but ICT has their
continuing credits and all the courses and all that and most of that stuff is
automated its online anyway so people can stay up to date with what’s going on
with some of the technology and backtrack to the basics yep
good point it’s everyone still needs to be shown where the power button is so behind the curtain
that’s right if operators want to learn more or find out more information on
orchestration is it just they go to your company’s website to Cisco to heiresses
website is that the best place to go right now or there or there is there any
public places that you can learn more about orchestration yeah it’s a good
question I don’t I don’t have a good recommendation
and that’s one there are a lot of I guess articles in the cable cable
industry news about the different types of approaches to that we do have some
you know some material on our website and Ericom some white papers some
presentations but it is kind of in the in the early phases of this
orchestration so it’s a lot of OCT o type of talk and kind of ideas for for
the great rate of good I’m gonna be able to provide with this network but not as
many products so I can’t really give you your daily sheets just yet as far from
this kind of product okay I would say bring it out the basket
on line one it’s not Google all right well Cal and John thank you so much this
was really really fascinating and I’m a great topic I’m gonna look forward to
see how this emerges our next episode next month is going to be all things
spectrum analysis with Larry wilcott from Comcast so tune in next month we do
our very best to bring you some good technical content every month if you
like watching this episode please click on the subscribe button or
follow us on your favorite podcaster for our audio only episode again thanks for
tuning in Tao John thanks for joining us really
appreciate your subject matter expertise on the topic and so everyone have a
great weekend we’ll see you later bye all pleasures usual

Related Posts

Rhino Rack Road-Warrior Roof Bike Racks Review – 2016 Ford Escape –

TravelSmith RFID Wallet with AntiTheft PacSafe Features

How to keep your mobile safe with Sophos

Leave a Reply

Your email address will not be published. Required fields are marked *