November 28, 2024 • 46 mins
Simon Karbowsky, CEO of StreamVX, joins us to reveal the groundbreaking technologies reshaping the video broadcasting landscape. Discover how this Polish company, born from a deep devotion to the broadcast industry, is revolutionizing traditional systems with robust and high-quality software solutions. As Simon guides us through the journey of StreamVX, we learn about the critical challenges faced by OTT services, like buffering and quality inconsistency, and how StreamVX is overcoming these with innovative solutions. Through a partnership with NetInt, they're setting new standards in power consumption efficiency, ensuring stability and reliability that meet the demands of modern broadcasting.
We also explore the dynamics between cloud deployments and on-premise solutions for broadcasters and pay-to-view operators. Local infrastructure still reigns supreme for high-throughput tasks, but Simon highlights the benefits of a hybrid approach, where peak traffic can be offloaded to the cloud, optimizing resource management. The conversation takes an exciting turn as we discuss advancements in live TV and video on demand, including the rise of time-shifted content and the transformative potential of AI in video processing. Whether it's utilizing off-the-shelf servers or embracing emerging codecs like AV1, Simon sheds light on how flexible, scalable technology is delivering a seamless end-to-end video experience, ensuring operators can meet both current and future challenges with ease.
Stay tuned for more in-depth insights on video technology, trends, and practical applications. Subscribe to Voices of Video: Inside the Tech for exclusive, hands-on knowledge from the experts. For more resources, visit Voices of Video.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:07):
Voices of Video.
Voices of Video.
Voices of Video.
Voices of Video.
Okay, it looks like we are liveWell.
Welcome to this next edition ofVoices of Video.
We are really excited about theopportunity to talk to video
(00:33):
experts and founders ofcompanies, engineers really
anybody doing something superinteresting in the area of video
.
And today I am joined by SimonKarbowsky, who is the CEO of
StreamVX.
By the way, my name is MarkDonegan, I work with NetInt and
(00:55):
you know, simon, I guess.
First of all, thank you forjoining us.
Speaker 2 (00:59):
It's great to be in conversation with you, yeah.
Speaker 1 (01:05):
Excellent.
Speaker 2 (01:05):
The pleasure is mine.
Thanks, good to be here.
Speaker 1 (01:07):
Yeah, super, super Well, I know that.
You know we're doing some realinteresting things together, our
two companies, but today we'rehere to talk about live
streaming in the box and youknow, before we get into the
product, why don't you tell uswho is StreamVX?
Get into the product?
Speaker 2 (01:25):
why don't you tell us who is streamvx?
The streamvx uh long storyshort is is a polish company
focused on building the softwareproducts towards the video
environment.
So we're delivering productswhich solves typical issues or
(01:46):
maybe sometimes not typical, theissues which broadcasters or
operators telcos happen to havewhile trying to deliver or
wanting to deliver the video.
Speaker 1 (01:54):
Yeah, absolutely, and I know that the journey of your
company.
You've been at this a littlewhile.
So what are the roots ofStreamVX?
Were you created to build thissolution or did you get here?
Um, you know what's the originstory, I guess, of the, yeah,
the company and the productthere's always.
Speaker 2 (02:15):
There's always some story and uh, of course and
probably ours quite interestingindeed.
The roots are are very, very,very long.
Yeah, once my friend said thatthe broadcast is the industry
which you either love and you'redoing this for for your life,
or you just hate it, and next,next year, you're just doing
completely different stuff.
(02:36):
So I stayed basically.
So that's what I'm doing.
My whole life is a broadcast inentertainment or pay tv market.
My whole life is a broadcastand entertainment or pay tv
market, and I always work withdifferent vendors and over my
career there was plenty ofthings I touched.
I used I had to deploy on thebig deployment, big production
(02:57):
and deployments in the multiplepay tv operations, and happened
to TV operations and happened tofind a place where I can
co-found the company whichstarted building this a little
upside down in the modern wayand solving these disadvantages
(03:18):
which I always hated.
So that's where we are, and theteam we gathered here in
SteamVX is also a lot ofpassionate people which support
this vision that we should makesophisticated processes easy but
also make sure that the basicsof broadcast are maintained
(03:44):
stability and all this, you know, assurance that the life is
life.
Speaker 1 (03:50):
Yeah, that's right, I know when you're talking.
You know broadcast or operatortype infrastructure it sometimes
those who come from, shall wesay, the more like OTT side, you
know they may say the moremodern side.
It's easy to you know, almostlook at some of the technologies
(04:15):
that are used and you know theway the systems are built.
You know there's still blackboxes in a lot of those networks
, right, black boxes in a lot ofthose networks right and you
know it's easy to look at it andthink you know that's a little
old fashioned.
And yet when you have to operatea service and you have to
guarantee five, ninesreliability and uptime and
(04:38):
there's a very specific bar forquality, that's delivered to
every household or to every TVset is actually pretty difficult
compared to the internet, whichis best efforts, ott, at the
end of the day, let's face it,it's best efforts and that's why
(04:59):
you still get buffers andbuffering events and that's why
you know you might have abeautiful 50-inch TV that's
getting 480p, because there'ssome sort of a network issue and
the ABR is you know.
Speaker 2 (05:13):
And meanwhile, it's trying to handle and help.
Speaker 1 (05:15):
Yeah, yeah, exactly.
So, yeah, why don't you give aswe were preparing?
You know, you told me thatthere's really four.
I know there's more me thatthere's really four.
I know there's more, butthere's four primary problems
that you're solving for yourcustomers and I thought they
were excellent.
So why don't you give us anoverview of what those problems
(05:37):
are?
Speaker 2 (05:39):
There is a lot more, but to pick up something which
is critical and we see ithappens all over the world with
our customers, the biggest oneconcern right now, obviously
about the geopolitics, is apower consumption, because in
Europe the power is getting moreand more expensive and
(06:02):
obviously it make a bigoperation to really reconsider
if, uh, if that usage can besomehow optimized.
So, uh, that's one, but on theother hand, of course it's, it's
much farther than that, becausethe power get generates the
heat, so the heat generates theAC to run faster and more and so
on.
So you know that, know that'sin total, a lot of things here
(06:27):
are getting to the power.
But you can optimize that,getting a very dense solution,
like we do with netting, thatyou have really like a lot of
density in really small formfactor.
So we reducing both, becausethe cards are very, very power
efficient but also they'retaking not much space, so we can
(06:50):
reduce both.
That's definitely topic numberone.
The side effect of that whichis not also very obvious or very
, very popular or very easy todo on the app is, as you
mentioned, is also, which soundsreally ridiculously simple, but
(07:12):
just the knowledge how thepower, power is used by the, by
the device, which we alsoencourage because for our
products it's it's pretty basicfunction to display that the
usage, power, usage, and you canbasically see this or attach
this to towards API to yoursystem to just monitor this
(07:35):
constantly, which is againconsidered as a really basic but
it's not obvious Usually on theappliance level.
You have to do some tricks todo that and to have this
knowledge.
Speaker 1 (07:47):
So you actually provide the power draw of the
server chassis that informationback to the user.
Speaker 2 (07:57):
Yeah, we're doing that and also yeah, we're doing
that exactly Interesting.
Speaker 1 (08:02):
Huh, wow, yeah, that's pretty unusual, I believe
.
Speaker 2 (08:08):
I mean, again, it's coming from this requirement
that the customers are askingokay, how much power does this
use there?
Because we both know that ifyou do a little more performant
tasks on the software, it canleverage a little higher
(08:31):
voltages on the servers and thecards.
But if you happen to dosomething less stressful for the
hardware, it can be lower.
So basically, okay, we can givethe be a lower.
So basically, okay, we can givethe data sheet which gives you
the range, but then it's not,it's, it's an average answer and
(08:52):
yeah, with with our metrics,you're basically getting the
answer like measurement on the,on the live stuff.
The second, going further, yeah.
The second second point is uh isdefinitely the quality.
I mean, uh, we, we're living inthe, in the world of the ott
wars or the streaming wars,however you call it, so, uh, so
(09:14):
there is really big fight tofind a differentiator, uh,
between the operations and thequality happens to be to be one
of the differentiators, quiteimportant one.
So, again, keeping the qualityas high as possible, but also
being aware of the, of themetrics, and then, like,
(09:35):
assuming that the vmas is givingyou something which is
subjective and considered by theindustry as a good metric or
consistent metric to give youthe comparison.
This is important to understand, to support this by the
products, but also to have aproduct which allows you to keep
(09:58):
these scores very, very high.
And this is again a prettyobvious metric of the quality,
but there are also latency orstability, so smoothly going
towards next.
Latency is very important forthe sports.
We all hate these cases whereyou have a screaming of your
neighbor, the goal, and then yousee the screen of the guy which
(10:21):
is just starting on the otherside.
Speaker 1 (10:25):
You already know what happened in the play right, but
you haven't seen it yet.
Speaker 2 (10:30):
Yeah, probably worse or steering scenario you can
imagine.
Yeah, so the latency comes here, also, in our opinion, a big
metric for the quality, butagain, together with NetInt,
it's being squeezed and we canreally touch really low numbers
(10:52):
with the ASICs to really pushthe latency very low.
The stability, that's somethingwhich we're speaking and I'm
encouraging all the time,pushing this towards the very
important factors, because wehave to remember that, okay,
there is a lot of products wedeploy in this bridge between
(11:15):
the typical broadcast operationand the broadcast going to OTT,
because that's also obviousdirection that a lot of
operators of the pay TV want tohave a part of the typical
broadcast operation movedtowards the devices.
So, on the OTT or OTTTechnologies, maybe not
(11:35):
necessarily out of the networkbut used ABR, uh, but they have
this as, as you mentionedearlier, they have this attitude
of the old black boxes whichare also very, very stable,
clunky and whatever, but theyare very, very stable.
So our products, moving this toour software, are trying to
(11:56):
make this nice marriage of thenew technologies which allows
you to scale fast, scaleefficiently, really leverage the
density.
But then we have to and that'sthe basic assumption we have to
be rock-solid, stable.
I mean, the life is life and itmust go on.
I mean that's the mostimportant part in here.
(12:18):
I don't want to have a break inthe game.
I mean that's something that'snot going to happen.
So stability and alsoredundancy scenarios, because
whatever the appliance is andhowever stable it is, there is
always something, like a MarfishLaw says there's always one
(12:39):
more bug, there is alwayssomething to fail, and it can be
very simple, very sophisticated.
So redundancy scenarios comesto stability.
That's also very, veryimportant.
Again, in our products,together with the NetInt, we can
do really fancy scenarios,really sophisticated scenarios,
(12:59):
which gives you this assurancethat the life runs on whatever
happens below.
Just really cool.
And then and then, of course,there is this.
This factors with the, with theusability, is really very, very
(13:22):
important Because we see that,okay, the world evolves and this
must be easy and this must bemuch easier than previously.
I mean, you don't want to makemistakes in the production or
make this more sophisticatedthan it's supposed to be.
You have to make sure that,okay, you can manage this
(13:45):
quickly and easily.
It should be accessible, andthis UX we can call it right now
and appliances was not designedin a time that somebody was
discussing UX at all, but stillyou're managing production, so
the stability is important, butyou don't want to make long
processes or a lot of decisions,which puts you in the place
(14:11):
where it's easy to make amistake.
That's the part, yeah.
Speaker 1 (14:18):
Easiness is also very , very important, just as the
reliability of the product isvery carefully engineered into
the whole solution, startingwith the software and then
extending to the hardware youchoose to support.
There know, is it easy to use?
But it actually contributes tostability, right, because a user
(14:52):
has a lower chance of maybemaking a configuration error.
You know, is that is?
Is that what you're saying, andis that?
Is that what you have found?
Speaker 2 (15:06):
And is that what you have found?
Speaker 1 (15:10):
You know that high quality UX is more than just a
beautiful screen.
It actually gives a morereliable system.
Speaker 2 (15:13):
Some smart guys said once that you know, ux is a
journey.
It's not one thing at thispoint of the time, it's a
journey and we happen to pushreally, really hard on the UI
experience in our productsbecause, as you said, it's the
processes we're handling and inthis transition from the
(15:36):
broadcast signal stores, ott,environment, transcoding,
packaging, protecting thecontent of the DRMs that all
it's very, very important.
But if you take a look on thewhole ecosystem here, the
process can be really, reallysophisticated.
At the end of the day, if itcomes to the transcoder, you
(15:57):
have hundreds of parameters youcan put towards the transcoder
to make the video, but notnecessarily being aware that
there's some combination.
We're going to blow the picturedown to the really ugly quality
or ugly effect at the end ofthe day.
So what we do and we're tryingto do, we're making sure that
(16:18):
okay most of the day.
So what we do and we're tryingto do, we're making sure that,
okay, the most of the cases youhave really easy path to
configure something quick, uhand easy.
And, of course, if you'rereally advanced and you want and
you're completely sure you wantto make a custom configuration,
you always can pick a customand type whatever you want, or
even do this through the API soyou can omit like a nice windows
(16:43):
in the clicking and you can dothis from the command line.
But that's really heavy usersand we see this really in
explicit cases when somebodyreally want to do something
extra custom.
Usually you have a really likeyou know, easy clicking up
(17:03):
several dozen times the same,the same settings because you
want to make sure also that thechosen or checked, verified
settings are always repeated theproper way, and that's that's
one of the things UX can handlefor you.
But again, the point here isthat you're not allowing to make
(17:25):
a mistake which makes the userexperience better and the
production more stable.
Speaker 1 (17:31):
That's right.
Can you explain what exactlyare you selling?
What have you developed?
Are you bringing to market justthe software?
Is it software plus hardware?
Is it software plus onlyspecific hardware?
Does it run on any x86 server?
(17:53):
I think it'd be really helpfulto explain what exactly you're
bringing to the market.
Speaker 2 (18:00):
Yeah, so the whole idea behind StreamVX is to make
software products and we reallytend to be agnostic for the
hardware, which in some cases,of course, goes towards the case
.
We are right now using net-intboards, net-int cards to
(18:21):
accelerate processing of thevideo, but that's kind of the
density which we tend to agreeon, because we leverage a
completely different level ofthe density again, again the
quality and all these things.
But, of course, from theperspective of the uh, of the,
(18:41):
of the, our product, we can dothis, the same process on the
GPU, we can do this on the CPU,and that goes just uh lines in
the different places.
When it comes to this densityor a power usage, all these
factors we mentioned in theprevious points are landing in
the different points.
(19:02):
This is always choice of thecustomer, of course, because in
some cases the customer justwants very specific hardware to
be used.
But from the perspective ofSimVx products, we really can
run on anything, basically, yeah.
So, yeah, that's what we, that,that, that's that's what the,
the dna of our products is andthat's, uh, that's how we work.
(19:23):
However, we really likepartnerships, like netting,
because, uh, if you can squeeze,uh, you know, as much as
possible from some board, fromsome card and give some extra
features or extra gains to thecustomers.
That usually lands in a lot ofinterest in our customers.
Speaker 1 (19:50):
Are most of your customers deploying in public
cloud, private cloud, just anon-prem data center.
What does it look like in termsof how they're deploying your
software?
Speaker 2 (20:08):
I would say that most of them again it's broadcast
operators or pay-to-viewoperators, so usually they own
the head ends somehow or somedata centers, however we call it
.
So usually they choose todeploy in their own
infrastructure because they ownit already.
So we can be very optimal onthe footprint together with this
(20:30):
solution we're right nowthinking about.
So we're not really rising thisfootprint heavily and adding a
few more servers to do work forthe appliances which was using
several times more.
So that's a big advantage.
(20:51):
But there are some cases wherethere is a demand for deploying
on the private or public cloud.
Of course it's just changing.
In my opinion, it's never goingto land the big scale
(21:13):
operations like we're poweringright now 300 channels for the
four biggest operators in thecountry on cable.
That's not going to land in thecloud because the traffic
throughput towards the cloud andback from the cloud is just too
expensive to sustain the modelof the cloud.
Yeah, that's right.
The one thing is thecapabilities, because the design
(21:33):
and the architecture of ourproducts allowing you to deploy
this basically almost anywherepublic, private cloud, local,
prem but most of the big caseswe deployed are local.
Speaker 1 (21:45):
Yeah, we're definitely seeing the same thing
for channels that are streaming, shall we say, 24-7.
So, true, live linear, thecloud, effectively the economics
don't work, you know it doesn'twork.
You have significant, and ofcourse everybody has a different
(22:07):
deal that they strike.
But you know you have toconsider egress cost, and you
know it's not just the cost ofgetting the bits into the cloud,
it's traversing the cloud andthen getting the bits back out,
and you know it's.
And then the other problem isis that if the business is
scaling, there's really noeconomies with the cloud.
(22:32):
So the whole, you know, idea ishey, but you can start small,
and then, you know, you don'thave to go make a very large
upfront capital investment, youknow, by going to the public
cloud.
Well, that's all true.
The problem is, as you scaleyour operations though you're,
you know, meaning, like thenumber of channels you offer, et
(22:53):
cetera Well, guess what?
Your costs scaleproportionately.
And so you never, you know,you're not able to benefit from
an investment.
That, true, maybe on day oneyou had to write a big check,
but you know you're able toamortize that over not only a
longer period of time but alsoover over a lot of users, you
(23:15):
know.
So we're definitely, we'redefinitely seeing that in
broadcast.
Well, let's get to the demo,are you?
Speaker 2 (23:24):
just well.
Yeah, just one comment towardsthat, if, if I may.
But?
But there is a really, reallyinteresting case, uh, which
which happened to be more andmore on the table in our
discussions with customers rightnow.
Then you have an operationwhich we all know is not so much
(23:49):
resistant I mean, the streamingoperations are not so much
resistant for the peak hoursthan the broadcast was,
obviously as the technologybehind it.
So what happens with the cloudand what's?
For us, it's quite easy againto leverage the architecture
we've done is that for the peakhours, the offloading of the
speaks towards cloud worksperfectly fine, because then
(24:12):
there is really small incident,like a few hours, where you have
to offload your infrastructurebecause you have too many
requests and you're handing thattowards the cloud and then,
when it's gone, you're back toyour own infrastructure.
Then you're having a bill,exactly what the cloud is for.
You're getting a bill for a fewhours done, deal, but you're
not having any outage.
That's the point where we cansay that's right for it.
Speaker 1 (24:36):
Absolutely, yeah, yeah, we can say that's right
for it.
Absolutely, yeah, yeah,definitely, that is a.
You know, this hybrid approachis um, is very powerful, and so,
yeah, well, let's get into ademo.
I think everybody you know, ofcourse we can talk more um and
maybe some questions will comein.
By the way, for those watchinglive, of course this will be
(25:01):
available as a replay.
So if some of your colleaguesweren't able to see it, we will
make the video available.
But if you're watching live,type in questions into the chat,
and I don't know that there'stoo much that we won't be
willing to cover.
So any questions you have,either about what Simon has said
(25:22):
, about live streaming, aboutthe cloud, about on-premise
appliances, asics, gpus, cpunothing's off the list, that's
true.
Yep, okay, are you ready, simon?
Speaker 2 (25:38):
Let me just yeah, I, I'm ready.
Happened to have some demoprepared for you guys.
Let me just we have justoverview of the channels which
the transcoding is running rightnow or has set up because they
can be stopped from streaming.
Adding a new channel is reallyreally simple with the platform,
Because you're just clickingthat and basically, when you
(26:02):
have to type in some nice name,you will find it later on.
And then you of course have totype some input address which in
most of the cases right now onthe market is some UDP multicast
, what happens right now.
Usually you should take a lookon some analyzer or you have to
(26:25):
make a note or whatever.
You just simply have to knowwhat's there.
But we have a nice button herewhich will make a probing, which
will basically tell you what'sin the stream which which you
just typed.
Then you have templates which Iprepared before and I want to
use that one.
(26:46):
I'll just show in a second whyand I can press the fill because
I'm really lazy.
So it's just making all theforms here and all I have to do
is just say, okay, I'm going tooutput this towards this place.
Let's say like that and that,and I'm going to save.
Speaker 1 (27:12):
Yep, I see my channel popped up.
Speaker 2 (27:16):
Yeah, exactly here.
And because by default we'renot starting it to make sure
that you really want to have iton the, on the infrastructure.
You just want to start it byyourself.
We have a nice notificationsthere and here you go which one?
Of course you always want anice notifications there and
here you go, it's right.
Of course you always want tochange something there.
(27:39):
So let's say, I want to add thisin a different place.
What happens now, and notalways happening, that's the
again UX.
We we tend to think it's savinga lot of issues on the
operation.
What happens is just warningyou that the change will take
(28:01):
effect, which you literally willrestore it.
It's just saving the changeitself, but it didn't apply.
Here you have to restart it.
Then it's going to be applied,so it's not breaking any
production.
It's just a Got, got it andthat kind of.
I'm not going to go for allthat kind of things, but believe
me, there is plenty of featureslike that which is trying to
(28:25):
just indicate okay, are you?
I really feel that you want tospoil your production just to
make sure that sometimes, if youjust be in a hurry or you want
to make something quick, you'renot really doing this, impacting
your customers.
Yeah, why I think this channelis so easy is that we
(28:46):
implemented something which isnot really rocket science.
It's the templates, whichallows you to define some
settings before, but, as you saw, adding the channel, I can
choose the template very easily.
I can, of course, create atemplate from here, but I think
you can also create it rightexactly in the channel, in the
other.
(29:06):
You don't have to go anywhereelse, you're just clicking here
and you can create it byyourself.
What you can do with thecreating the new template, of
course, using the names andchanging that typical stuff.
You also can type the completeconfiguration from the scratch.
So you're choosing the settings, you're choosing resolutions
(29:30):
predefined, what you can putyour own, you can choose all
these parameters.
But if you happen to startsetting up a dozen of channels
already, you probably have someof these profiles.
So you can load them, not goinganywhere, not losing the
context, and you can basicallybrowse the settings you like.
Let's say I like this one andthe funny part is I can do two
(29:55):
things here.
So I can share it, so I can useexactly this setting, this
profile I'm using, for example,for HBO.
I can use it for for 11 sports,whatever, or HBO two maybe
better example here or I canclone it because I like all the
parameters but I want to changebitrate, for example.
So if I clone it, I like allthe parameters but I want to
change bitrate, for example.
(30:15):
So if I clone it, I'm justmaking easy copy and I can
change one or two parameters allof them, but I like the name or
whatever, or I can share it.
So I'm using exactly the, theprofile I'm seeing.
So it's so, so easy.
The nice thing also is that ifyou have a lot of templates or a
lot of profiles, like in here,you can change it in line, so
(30:41):
you're again not losing thecontent, but you see other
profiles which we again, from aUX perspective, if you're
changing, like you know imaginechanging 300 channels you really
want to see what other profilesare, just to make sure that
you're really making the ladder,going down or up the way you
want, and not really go to theother screen and make notes and
(31:04):
make copies and make sure thatyou didn't mess up, because
you're making very often you'remaking quite a lot of repetitive
tasks, one after another, andthat kind of mistakes really.
We, we saw a, we saw a bigoperations really messy to clean
up this stuff.
So, yeah, we put, as you see, alot, of, a lot of effort into
(31:26):
this configuration and even ifyou add a new one, you can see I
can add like a next one, andit's still on top of the of the
previous.
I'm editing to just make sureyou you have the flow and you
understand what's what's theprevious task he was doing.
Speaker 1 (31:46):
Yeah, the same goes for the video profiles.
Speaker 2 (31:48):
yeah, so you, you got .
I think you got my idea.
It's really in the in mind withthe guys, which wants to make
sure that it's in order and inthe process, because when you're
adding one channel it probablydoesn't matter, but usually
operations we support reallyusing hundreds of channels I'm
(32:13):
not mentioning you know yourtypical cable network has.
Speaker 1 (32:17):
What about for your users?
Has about 200 channels, 300channels.
Is that standard?
Speaker 2 (32:24):
Yeah, typical operations right now is between
200 to 300 channels.
That's probably lineups we areon.
So still, even if you're reallyconservative with the ladders,
you still have a few pages ofthis profile set up.
Yeah, that can be tricky to gothrough that and not make a
(32:46):
mistake.
Which one you are assigningtowards what?
Speaker 1 (32:49):
Yeah, so, simon, a question came in that I think
it's a good time to answer, andthat is what codecs do you
support?
And then I'll even ask afollow-on question, you know is
what codecs do you support now?
And then you know what'sinvolved with adding additional
codecs.
(33:09):
So you know, as we well know,every, you know it feels almost
like every couple of yearsthere's, you know, there's some
new standard that's that's beingintroduced, but maybe not quite
that frequency that's beingadopted, obviously, but yeah.
Speaker 2 (33:23):
So what codecs do you ?
Speaker 1 (33:25):
support.
Speaker 2 (33:26):
So what we?
What we do right now, of courseit's H.264 and HBC or H.265.
But then we also support,together with the net imports,
AV1.
Av1, yeah, which gainspopularity more and more.
Speaker 1 (33:47):
And, by the way, that's AV1 live.
I want to make that clear.
Yeah, that's not exactly thesame.
Yeah, very, very very important.
Speaker 2 (33:54):
In general, the design we we have here, uh,
basically doesn't close us toany codecs.
So if the boards which waswhich we used to to support or
to uh, uh to make theperformance of the acceleration
better, like in this case,nothing, nothing boards when the
(34:17):
net in start to support othercodecs or a newer codecs and a
version of the codecs, thesoftware will simply follow on,
yeah, yeah.
So I would say, whatever is isbelow on the hardware.
If you have some specifics, uh,we can go there.
And the second point it soundswe have to be clear about it is,
(34:37):
of course, licensing.
Yeah, but all of the codecs arejust uh, free to use.
So, uh, yeah, but also can comewith some extra costs or extra
paperwork, at least uh, to cover, sure, but that's uh, that's to
be considered.
Speaker 1 (34:52):
So, basically, yeah, that's for the distributor.
That's independent of thesolution.
Even if you use an open sourceKodak, you're going to have to
go get the appropriate licensing.
Question that I have as well isis this all single bitrate?
(35:13):
Do you support HLS?
Speaker 2 (35:19):
Do you have a packaging function?
Yeah, we have a packaging, ofcourse, as a function.
It's a separate product in theline.
But again, an advantage ofStreamVx architecture is that
you can get it transparently, soas much as we can sell this
like Lego blocks.
So you can have a transcoderwhich is outputting already
(35:41):
packaged content, but you canhave a transcoder which you're
using just to fit the currentlyexisting package.
Yeah, that's a huge advantageof our architecture.
But yeah, we can of coursesupport HLS-DASH and CMAP a huge
advantage of our architecturebut yeah, we can.
Uh, we can of course support uh,hls, dash and cmf.
Uh, we can, of course also drmthat we can serve this as a cdn.
(36:02):
So the whole flow, startingfrom the inputs towards the
distribution, it's, it's coveredand it can be covered in the in
the smooth, uh smooth path.
We can also provide therecordings of this.
So you can make live TV offline, like Timesheet, restore TV,
that kind of services.
And you can do opposite waywith the wizard.
(36:26):
We call the product wizardbecause it's also doing the
offline content towards the livechannels and, as you see here,
because it's also it's I knowthat the topic today's live, but
I just show it quickly.
To just prove the case, with theconsistency and the UX, we have
a transcoder for the VOD herealso, and it's based on the
(36:50):
watch folders.
But the funny part, or to justagain sustain the case, uh, the
configuration is, uh is exactlythe same.
So you're choosing the sametemplates, basically from the
same list.
So, uh, it's very easy.
And then you just, with oneclick, see all the files which
(37:11):
was processed by this watchfolder, or, if they are in the
process, they are progressinghere.
So it's as easy as that.
Speaker 1 (37:21):
Now, can you do live to VOD?
So let's say that I have a livechannel, but I want to create a
VOD asset.
Can you do that?
Speaker 2 (37:31):
That's exactly what we also can do.
It's, of course, very popularbecause you know the tendency
how we consume the video.
Content changed, uh, so rightnow we prefer to watch this,
even the live content, but movedinto the time, yeah, while we
have a time really to to consumeit.
So, yeah, we're doing the, thetime shifted, let's say content,
(37:55):
uh, and it's again integrateddeeply towards the, the products
of stream vx.
So I'm not having this rightnow in the demo, but you can
imagine, with the one click.
It's just if you, if you happento have our packaging system,
it just simply allows you to toshow the channels which we're
packaging for you and justsetting up how big time frame
(38:19):
you want to record, and it'sjust press start and it's done
Again this tendency to make asimple few clicks.
Configuration is all over theplace.
Speaker 1 (38:31):
Yeah, yeah.
Well, simon, we could keeptalking for much longer.
I'm looking at the clock hereand we've already gone a few
minutes over, which is fine.
I hope everybody is gettingsome value.
So I do want to open it up.
If there's some furtherquestions you have a minute or
(38:52):
two here Feel free to just putthem in the chat and we'll
address them.
At the same time, if you addcomments to the LinkedIn event,
we'll jump on and address them.
So lots of ways to get in touchwith us, but I want to, I think
(39:12):
, end with this.
I know that we're both veryexcited about OK good, I see a
question that came in and I'llget to this to this one.
But you know, and bring theusability and the performance,
you know, to what otherwise areLego blocks that are often
bolted, together with APIs andcommand line interfaces, you
know.
But we're really excited abouta project that I think we're
(40:00):
thinking about and that isbringing to market more of a
full solution, you know so, likea server.
Do you want to comment aboutthat?
Speaker 2 (40:08):
And yeah, again we really see the big power into
the kind of delivering theend-to-end solution that's the
topic of today, called like inone box.
Again it's the very importantpart is here it's still okay,
it's a box, but it's notappliance, because the hardware
(40:30):
is a out-of-the-shelf server, sothere is no specific I don't
know around modules or the driveor whatever.
It's just out of the shelfserver, so there is no specific
I don't know around modules orthe drive or whatever.
It's just out of the shelvesequipment.
The boards from NetInt okay,they are NetInt boards, but you
can, you can buy them freely andyou can exchange them freely.
There is a huge advantage whichwe didn't mention, but I love
(40:51):
it.
Actually it's touching a littlethis codex.
Imagine having 408s which doingH.264 and 5.
And then, if you happen to wantto have AV1, you're just
swapping the card, it's hotpluggable, and then suddenly you
support the technology which isjust on the market for live
(41:14):
content.
It's really fresh stuff.
You don't have to exchange thehalf of your act.
Yeah, you're just extending areally small factor board and
you can do this, uh, basicallyinstant, uh, so yeah, yeah, that
we really believe in in goingthat, that direction, that we,
uh, basically basically combineour forces let's say that way
(41:36):
nicely that we, we make a reallyfitted solution which you can
multiply, of course, but if youjust need a like a 30 channels,
it's a one box.
If you need 100 channels, it'sprobably four boxes.
Yeah, on top of that, all thesefeatures we spoke redundancy,
stability, we mentioned thatthat's really direction which is
(42:01):
, in our opinion, allowingreally huge market, huge
audience, or huge amount ofcustomers, operators, to get
into the ISE encoding,transcoding part, because, okay,
the big guys we spoke aboutoffline, about that the big guys
(42:21):
are developing their own ASICs,because obviously that's the
way you should do this and we'retrying to build something which
will do this to the rest of themarket.
For the masses.
Speaker 1 (42:35):
Exactly, exactly, well, exactly exactly, well,
that's great, well, simon.
One last question, then we willwrap up here.
So somebody is asking have youintegrated, you know, like chat,
gpt or GPT-3 into your systemso it can create program video
(43:00):
highlights automatically?
And I guess maybe you cancomment if there's just some
functionality that can do someAI infused.
Speaker 2 (43:13):
Yeah, the trick.
Here again, try to make a shortanswer for the long story.
Indeed, we have.
Actually, we have some productwhich is designed based around
the AI, to look for the videopatterns into the video, which
is more into the advertisingmarket or the metadata
(43:37):
enrichment, that kind of placesstill the regions we discussed
today.
But because of the high texturethe transcoder is having, it's
really easy to move the partswhich we're using in there
towards here.
And we was definitely thinkingabout it, because what we, what
we did with the, with theproduct I'm mentioning it's
(43:59):
called vx finder because it'sfinding the patterns we was
working on the on the project,while, uh, big operator had a
vast library of the vod assetsand he wanted to simply
implement the skip intro and thenext step is with buttons.
But he had no metadata tosupport this function.
So we basically scanned withthe AI the library and just
(44:21):
marked them precisely to theframe.
So it wasn't really more orless, it was to the frame when
the episode starts, when it ends.
So it was easy peasy to justmake those buttons and make the
experience similar to Netflix orother big guys, yeah that's
right On the basis of theprobably two or three days of
(44:43):
processing, instead of sittingprobably few persons and making
a lot of mistakes trying to fitit manually.
The one thing is right now it'snot there in a GPT-3 manner, I
would say, but it's really closeand for us it's very easy
(45:04):
probably to experiment.
So I would be glad to follow upon that one, if all goes well
here.
Speaker 1 (45:11):
Well, simon, thank you again, Really appreciate
your time.
We're really excited about whatyou guys are doing and
hopefully this was of interestto the viewers and so for all
those who joined us.
Thank you again for watchingVoices of Video.
We have new episodes comingvery regularly, so just watch
(45:35):
the schedule and if you're onour email list, we'll let you
know as well.
So thank you again, greatThanks.
Speaker 2 (45:42):
Mark, thanks for having me.
It was a pleasure.
This episode of Voices of Videois brought to you by NetInt
Technologies.
If you are looking forcutting-edge video encoding
solutions, check out NetInt'sproducts at netintcom.
Voices of Video.
Voices of Video.
Voices of Video.
Voices of Video.
Okay, it looks like we are liveWell.
Welcome to this next edition ofVoices of Video.
We are really excited about theopportunity to talk to video
(00:33):
experts and founders ofcompanies, engineers really
anybody doing something superinteresting in the area of video
.
And today I am joined by SimonKarbowsky, who is the CEO of
StreamVX.
By the way, my name is MarkDonegan, I work with NetInt and
(00:55):
you know, simon, I guess.
First of all, thank you forjoining us.
Speaker 2 (00:59):
It's great to be in conversation with you, yeah.
Speaker 1 (01:05):
Excellent.
Speaker 2 (01:05):
The pleasure is mine.
Thanks, good to be here.
Speaker 1 (01:07):
Yeah, super, super Well, I know that.
You know we're doing some realinteresting things together, our
two companies, but today we'rehere to talk about live
streaming in the box and youknow, before we get into the
product, why don't you tell uswho is StreamVX?
Get into the product?
Speaker 2 (01:25):
why don't you tell us who is streamvx?
The streamvx uh long storyshort is is a polish company
focused on building the softwareproducts towards the video
environment.
So we're delivering productswhich solves typical issues or
(01:46):
maybe sometimes not typical, theissues which broadcasters or
operators telcos happen to havewhile trying to deliver or
wanting to deliver the video.
Speaker 1 (01:54):
Yeah, absolutely, and I know that the journey of your
company.
You've been at this a littlewhile.
So what are the roots ofStreamVX?
Were you created to build thissolution or did you get here?
Um, you know what's the originstory, I guess, of the, yeah,
the company and the productthere's always.
Speaker 2 (02:15):
There's always some story and uh, of course and
probably ours quite interestingindeed.
The roots are are very, very,very long.
Yeah, once my friend said thatthe broadcast is the industry
which you either love and you'redoing this for for your life,
or you just hate it, and next,next year, you're just doing
completely different stuff.
(02:36):
So I stayed basically.
So that's what I'm doing.
My whole life is a broadcast inentertainment or pay tv market.
My whole life is a broadcastand entertainment or pay tv
market, and I always work withdifferent vendors and over my
career there was plenty ofthings I touched.
I used I had to deploy on thebig deployment, big production
(02:57):
and deployments in the multiplepay tv operations, and happened
to TV operations and happened tofind a place where I can
co-found the company whichstarted building this a little
upside down in the modern wayand solving these disadvantages
(03:18):
which I always hated.
So that's where we are, and theteam we gathered here in
SteamVX is also a lot ofpassionate people which support
this vision that we should makesophisticated processes easy but
also make sure that the basicsof broadcast are maintained
(03:44):
stability and all this, you know, assurance that the life is
life.
Speaker 1 (03:50):
Yeah, that's right, I know when you're talking.
You know broadcast or operatortype infrastructure it sometimes
those who come from, shall wesay, the more like OTT side, you
know they may say the moremodern side.
It's easy to you know, almostlook at some of the technologies
(04:15):
that are used and you know theway the systems are built.
You know there's still blackboxes in a lot of those networks
, right, black boxes in a lot ofthose networks right and you
know it's easy to look at it andthink you know that's a little
old fashioned.
And yet when you have to operatea service and you have to
guarantee five, ninesreliability and uptime and
(04:38):
there's a very specific bar forquality, that's delivered to
every household or to every TVset is actually pretty difficult
compared to the internet, whichis best efforts, ott, at the
end of the day, let's face it,it's best efforts and that's why
(04:59):
you still get buffers andbuffering events and that's why
you know you might have abeautiful 50-inch TV that's
getting 480p, because there'ssome sort of a network issue and
the ABR is you know.
Speaker 2 (05:13):
And meanwhile, it's trying to handle and help.
Speaker 1 (05:15):
Yeah, yeah, exactly.
So, yeah, why don't you give aswe were preparing?
You know, you told me thatthere's really four.
I know there's more me thatthere's really four.
I know there's more, butthere's four primary problems
that you're solving for yourcustomers and I thought they
were excellent.
So why don't you give us anoverview of what those problems
(05:37):
are?
Speaker 2 (05:39):
There is a lot more, but to pick up something which
is critical and we see ithappens all over the world with
our customers, the biggest oneconcern right now, obviously
about the geopolitics, is apower consumption, because in
Europe the power is getting moreand more expensive and
(06:02):
obviously it make a bigoperation to really reconsider
if, uh, if that usage can besomehow optimized.
So, uh, that's one, but on theother hand, of course it's, it's
much farther than that, becausethe power get generates the
heat, so the heat generates theAC to run faster and more and so
on.
So you know that, know that'sin total, a lot of things here
(06:27):
are getting to the power.
But you can optimize that,getting a very dense solution,
like we do with netting, thatyou have really like a lot of
density in really small formfactor.
So we reducing both, becausethe cards are very, very power
efficient but also they'retaking not much space, so we can
(06:50):
reduce both.
That's definitely topic numberone.
The side effect of that whichis not also very obvious or very
, very popular or very easy todo on the app is, as you
mentioned, is also, which soundsreally ridiculously simple, but
(07:12):
just the knowledge how thepower, power is used by the, by
the device, which we alsoencourage because for our
products it's it's pretty basicfunction to display that the
usage, power, usage, and you canbasically see this or attach
this to towards API to yoursystem to just monitor this
(07:35):
constantly, which is againconsidered as a really basic but
it's not obvious Usually on theappliance level.
You have to do some tricks todo that and to have this
knowledge.
Speaker 1 (07:47):
So you actually provide the power draw of the
server chassis that informationback to the user.
Speaker 2 (07:57):
Yeah, we're doing that and also yeah, we're doing
that exactly Interesting.
Speaker 1 (08:02):
Huh, wow, yeah, that's pretty unusual, I believe
.
Speaker 2 (08:08):
I mean, again, it's coming from this requirement
that the customers are askingokay, how much power does this
use there?
Because we both know that ifyou do a little more performant
tasks on the software, it canleverage a little higher
(08:31):
voltages on the servers and thecards.
But if you happen to dosomething less stressful for the
hardware, it can be lower.
So basically, okay, we can givethe be a lower.
So basically, okay, we can givethe data sheet which gives you
the range, but then it's not,it's, it's an average answer and
(08:52):
yeah, with with our metrics,you're basically getting the
answer like measurement on the,on the live stuff.
The second, going further, yeah.
The second second point is uh isdefinitely the quality.
I mean, uh, we, we're living inthe, in the world of the ott
wars or the streaming wars,however you call it, so, uh, so
(09:14):
there is really big fight tofind a differentiator, uh,
between the operations and thequality happens to be to be one
of the differentiators, quiteimportant one.
So, again, keeping the qualityas high as possible, but also
being aware of the, of themetrics, and then, like,
(09:35):
assuming that the vmas is givingyou something which is
subjective and considered by theindustry as a good metric or
consistent metric to give youthe comparison.
This is important to understand, to support this by the
products, but also to have aproduct which allows you to keep
(09:58):
these scores very, very high.
And this is again a prettyobvious metric of the quality,
but there are also latency orstability, so smoothly going
towards next.
Latency is very important forthe sports.
We all hate these cases whereyou have a screaming of your
neighbor, the goal, and then yousee the screen of the guy which
(10:21):
is just starting on the otherside.
Speaker 1 (10:25):
You already know what happened in the play right, but
you haven't seen it yet.
Speaker 2 (10:30):
Yeah, probably worse or steering scenario you can
imagine.
Yeah, so the latency comes here, also, in our opinion, a big
metric for the quality, butagain, together with NetInt,
it's being squeezed and we canreally touch really low numbers
(10:52):
with the ASICs to really pushthe latency very low.
The stability, that's somethingwhich we're speaking and I'm
encouraging all the time,pushing this towards the very
important factors, because wehave to remember that, okay,
there is a lot of products wedeploy in this bridge between
(11:15):
the typical broadcast operationand the broadcast going to OTT,
because that's also obviousdirection that a lot of
operators of the pay TV want tohave a part of the typical
broadcast operation movedtowards the devices.
So, on the OTT or OTTTechnologies, maybe not
(11:35):
necessarily out of the networkbut used ABR, uh, but they have
this as, as you mentionedearlier, they have this attitude
of the old black boxes whichare also very, very stable,
clunky and whatever, but theyare very, very stable.
So our products, moving this toour software, are trying to
(11:56):
make this nice marriage of thenew technologies which allows
you to scale fast, scaleefficiently, really leverage the
density.
But then we have to and that'sthe basic assumption we have to
be rock-solid, stable.
I mean, the life is life and itmust go on.
I mean that's the mostimportant part in here.
(12:18):
I don't want to have a break inthe game.
I mean that's something that'snot going to happen.
So stability and alsoredundancy scenarios, because
whatever the appliance is andhowever stable it is, there is
always something, like a MarfishLaw says there's always one
(12:39):
more bug, there is alwayssomething to fail, and it can be
very simple, very sophisticated.
So redundancy scenarios comesto stability.
That's also very, veryimportant.
Again, in our products,together with the NetInt, we can
do really fancy scenarios,really sophisticated scenarios,
(12:59):
which gives you this assurancethat the life runs on whatever
happens below.
Just really cool.
And then and then, of course,there is this.
This factors with the, with theusability, is really very, very
(13:22):
important Because we see that,okay, the world evolves and this
must be easy and this must bemuch easier than previously.
I mean, you don't want to makemistakes in the production or
make this more sophisticatedthan it's supposed to be.
You have to make sure that,okay, you can manage this
(13:45):
quickly and easily.
It should be accessible, andthis UX we can call it right now
and appliances was not designedin a time that somebody was
discussing UX at all, but stillyou're managing production, so
the stability is important, butyou don't want to make long
processes or a lot of decisions,which puts you in the place
(14:11):
where it's easy to make amistake.
That's the part, yeah.
Speaker 1 (14:18):
Easiness is also very , very important, just as the
reliability of the product isvery carefully engineered into
the whole solution, startingwith the software and then
extending to the hardware youchoose to support.
There know, is it easy to use?
But it actually contributes tostability, right, because a user
(14:52):
has a lower chance of maybemaking a configuration error.
You know, is that is?
Is that what you're saying, andis that?
Is that what you have found?
Speaker 2 (15:06):
And is that what you have found?
Speaker 1 (15:10):
You know that high quality UX is more than just a
beautiful screen.
It actually gives a morereliable system.
Speaker 2 (15:13):
Some smart guys said once that you know, ux is a
journey.
It's not one thing at thispoint of the time, it's a
journey and we happen to pushreally, really hard on the UI
experience in our productsbecause, as you said, it's the
processes we're handling and inthis transition from the
(15:36):
broadcast signal stores, ott,environment, transcoding,
packaging, protecting thecontent of the DRMs that all
it's very, very important.
But if you take a look on thewhole ecosystem here, the
process can be really, reallysophisticated.
At the end of the day, if itcomes to the transcoder, you
(15:57):
have hundreds of parameters youcan put towards the transcoder
to make the video, but notnecessarily being aware that
there's some combination.
We're going to blow the picturedown to the really ugly quality
or ugly effect at the end ofthe day.
So what we do and we're tryingto do, we're making sure that
(16:18):
okay most of the day.
So what we do and we're tryingto do, we're making sure that,
okay, the most of the cases youhave really easy path to
configure something quick, uhand easy.
And, of course, if you'rereally advanced and you want and
you're completely sure you wantto make a custom configuration,
you always can pick a customand type whatever you want, or
even do this through the API soyou can omit like a nice windows
(16:43):
in the clicking and you can dothis from the command line.
But that's really heavy usersand we see this really in
explicit cases when somebodyreally want to do something
extra custom.
Usually you have a really likeyou know, easy clicking up
(17:03):
several dozen times the same,the same settings because you
want to make sure also that thechosen or checked, verified
settings are always repeated theproper way, and that's that's
one of the things UX can handlefor you.
But again, the point here isthat you're not allowing to make
(17:25):
a mistake which makes the userexperience better and the
production more stable.
Speaker 1 (17:31):
That's right.
Can you explain what exactlyare you selling?
What have you developed?
Are you bringing to market justthe software?
Is it software plus hardware?
Is it software plus onlyspecific hardware?
Does it run on any x86 server?
(17:53):
I think it'd be really helpfulto explain what exactly you're
bringing to the market.
Speaker 2 (18:00):
Yeah, so the whole idea behind StreamVX is to make
software products and we reallytend to be agnostic for the
hardware, which in some cases,of course, goes towards the case
.
We are right now using net-intboards, net-int cards to
(18:21):
accelerate processing of thevideo, but that's kind of the
density which we tend to agreeon, because we leverage a
completely different level ofthe density again, again the
quality and all these things.
But, of course, from theperspective of the uh, of the,
(18:41):
of the, our product, we can dothis, the same process on the
GPU, we can do this on the CPU,and that goes just uh lines in
the different places.
When it comes to this densityor a power usage, all these
factors we mentioned in theprevious points are landing in
the different points.
(19:02):
This is always choice of thecustomer, of course, because in
some cases the customer justwants very specific hardware to
be used.
But from the perspective ofSimVx products, we really can
run on anything, basically, yeah.
So, yeah, that's what we, that,that, that's that's what the,
the dna of our products is andthat's, uh, that's how we work.
(19:23):
However, we really likepartnerships, like netting,
because, uh, if you can squeeze,uh, you know, as much as
possible from some board, fromsome card and give some extra
features or extra gains to thecustomers.
That usually lands in a lot ofinterest in our customers.
Speaker 1 (19:50):
Are most of your customers deploying in public
cloud, private cloud, just anon-prem data center.
What does it look like in termsof how they're deploying your
software?
Speaker 2 (20:08):
I would say that most of them again it's broadcast
operators or pay-to-viewoperators, so usually they own
the head ends somehow or somedata centers, however we call it
.
So usually they choose todeploy in their own
infrastructure because they ownit already.
So we can be very optimal onthe footprint together with this
(20:30):
solution we're right nowthinking about.
So we're not really rising thisfootprint heavily and adding a
few more servers to do work forthe appliances which was using
several times more.
So that's a big advantage.
(20:51):
But there are some cases wherethere is a demand for deploying
on the private or public cloud.
Of course it's just changing.
In my opinion, it's never goingto land the big scale
(21:13):
operations like we're poweringright now 300 channels for the
four biggest operators in thecountry on cable.
That's not going to land in thecloud because the traffic
throughput towards the cloud andback from the cloud is just too
expensive to sustain the modelof the cloud.
Yeah, that's right.
The one thing is thecapabilities, because the design
(21:33):
and the architecture of ourproducts allowing you to deploy
this basically almost anywherepublic, private cloud, local,
prem but most of the big caseswe deployed are local.
Speaker 1 (21:45):
Yeah, we're definitely seeing the same thing
for channels that are streaming, shall we say, 24-7.
So, true, live linear, thecloud, effectively the economics
don't work, you know it doesn'twork.
You have significant, and ofcourse everybody has a different
(22:07):
deal that they strike.
But you know you have toconsider egress cost, and you
know it's not just the cost ofgetting the bits into the cloud,
it's traversing the cloud andthen getting the bits back out,
and you know it's.
And then the other problem isis that if the business is
scaling, there's really noeconomies with the cloud.
(22:32):
So the whole, you know, idea ishey, but you can start small,
and then, you know, you don'thave to go make a very large
upfront capital investment, youknow, by going to the public
cloud.
Well, that's all true.
The problem is, as you scaleyour operations though you're,
you know, meaning, like thenumber of channels you offer, et
(22:53):
cetera Well, guess what?
Your costs scaleproportionately.
And so you never, you know,you're not able to benefit from
an investment.
That, true, maybe on day oneyou had to write a big check,
but you know you're able toamortize that over not only a
longer period of time but alsoover over a lot of users, you
(23:15):
know.
So we're definitely, we'redefinitely seeing that in
broadcast.
Well, let's get to the demo,are you?
Speaker 2 (23:24):
just well.
Yeah, just one comment towardsthat, if, if I may.
But?
But there is a really, reallyinteresting case, uh, which
which happened to be more andmore on the table in our
discussions with customers rightnow.
Then you have an operationwhich we all know is not so much
(23:49):
resistant I mean, the streamingoperations are not so much
resistant for the peak hoursthan the broadcast was,
obviously as the technologybehind it.
So what happens with the cloudand what's?
For us, it's quite easy againto leverage the architecture
we've done is that for the peakhours, the offloading of the
speaks towards cloud worksperfectly fine, because then
(24:12):
there is really small incident,like a few hours, where you have
to offload your infrastructurebecause you have too many
requests and you're handing thattowards the cloud and then,
when it's gone, you're back toyour own infrastructure.
Then you're having a bill,exactly what the cloud is for.
You're getting a bill for a fewhours done, deal, but you're
not having any outage.
That's the point where we cansay that's right for it.
Speaker 1 (24:36):
Absolutely, yeah, yeah, we can say that's right
for it.
Absolutely, yeah, yeah,definitely, that is a.
You know, this hybrid approachis um, is very powerful, and so,
yeah, well, let's get into ademo.
I think everybody you know, ofcourse we can talk more um and
maybe some questions will comein.
By the way, for those watchinglive, of course this will be
(25:01):
available as a replay.
So if some of your colleaguesweren't able to see it, we will
make the video available.
But if you're watching live,type in questions into the chat,
and I don't know that there'stoo much that we won't be
willing to cover.
So any questions you have,either about what Simon has said
(25:22):
, about live streaming, aboutthe cloud, about on-premise
appliances, asics, gpus, cpunothing's off the list, that's
true.
Yep, okay, are you ready, simon?
Speaker 2 (25:38):
Let me just yeah, I, I'm ready.
Happened to have some demoprepared for you guys.
Let me just we have justoverview of the channels which
the transcoding is running rightnow or has set up because they
can be stopped from streaming.
Adding a new channel is reallyreally simple with the platform,
Because you're just clickingthat and basically, when you
(26:02):
have to type in some nice name,you will find it later on.
And then you of course have totype some input address which in
most of the cases right now onthe market is some UDP multicast
, what happens right now.
Usually you should take a lookon some analyzer or you have to
(26:25):
make a note or whatever.
You just simply have to knowwhat's there.
But we have a nice button herewhich will make a probing, which
will basically tell you what'sin the stream which which you
just typed.
Then you have templates which Iprepared before and I want to
use that one.
(26:46):
I'll just show in a second whyand I can press the fill because
I'm really lazy.
So it's just making all theforms here and all I have to do
is just say, okay, I'm going tooutput this towards this place.
Let's say like that and that,and I'm going to save.
Speaker 1 (27:12):
Yep, I see my channel popped up.
Speaker 2 (27:16):
Yeah, exactly here.
And because by default we'renot starting it to make sure
that you really want to have iton the, on the infrastructure.
You just want to start it byyourself.
We have a nice notificationsthere and here you go which one?
Of course you always want anice notifications there and
here you go, it's right.
Of course you always want tochange something there.
(27:39):
So let's say, I want to add thisin a different place.
What happens now, and notalways happening, that's the
again UX.
We we tend to think it's savinga lot of issues on the
operation.
What happens is just warningyou that the change will take
(28:01):
effect, which you literally willrestore it.
It's just saving the changeitself, but it didn't apply.
Here you have to restart it.
Then it's going to be applied,so it's not breaking any
production.
It's just a Got, got it andthat kind of.
I'm not going to go for allthat kind of things, but believe
me, there is plenty of featureslike that which is trying to
(28:25):
just indicate okay, are you?
I really feel that you want tospoil your production just to
make sure that sometimes, if youjust be in a hurry or you want
to make something quick, you'renot really doing this, impacting
your customers.
Yeah, why I think this channelis so easy is that we
(28:46):
implemented something which isnot really rocket science.
It's the templates, whichallows you to define some
settings before, but, as you saw, adding the channel, I can
choose the template very easily.
I can, of course, create atemplate from here, but I think
you can also create it rightexactly in the channel, in the
other.
(29:06):
You don't have to go anywhereelse, you're just clicking here
and you can create it byyourself.
What you can do with thecreating the new template, of
course, using the names andchanging that typical stuff.
You also can type the completeconfiguration from the scratch.
So you're choosing the settings, you're choosing resolutions
(29:30):
predefined, what you can putyour own, you can choose all
these parameters.
But if you happen to startsetting up a dozen of channels
already, you probably have someof these profiles.
So you can load them, not goinganywhere, not losing the
context, and you can basicallybrowse the settings you like.
Let's say I like this one andthe funny part is I can do two
(29:55):
things here.
So I can share it, so I can useexactly this setting, this
profile I'm using, for example,for HBO.
I can use it for for 11 sports,whatever, or HBO two maybe
better example here or I canclone it because I like all the
parameters but I want to changebitrate, for example.
So if I clone it, I like allthe parameters but I want to
change bitrate, for example.
(30:15):
So if I clone it, I'm justmaking easy copy and I can
change one or two parameters allof them, but I like the name or
whatever, or I can share it.
So I'm using exactly the, theprofile I'm seeing.
So it's so, so easy.
The nice thing also is that ifyou have a lot of templates or a
lot of profiles, like in here,you can change it in line, so
(30:41):
you're again not losing thecontent, but you see other
profiles which we again, from aUX perspective, if you're
changing, like you know imaginechanging 300 channels you really
want to see what other profilesare, just to make sure that
you're really making the ladder,going down or up the way you
want, and not really go to theother screen and make notes and
(31:04):
make copies and make sure thatyou didn't mess up, because
you're making very often you'remaking quite a lot of repetitive
tasks, one after another, andthat kind of mistakes really.
We, we saw a, we saw a bigoperations really messy to clean
up this stuff.
So, yeah, we put, as you see, alot, of, a lot of effort into
(31:26):
this configuration and even ifyou add a new one, you can see I
can add like a next one, andit's still on top of the of the
previous.
I'm editing to just make sureyou you have the flow and you
understand what's what's theprevious task he was doing.
Speaker 1 (31:46):
Yeah, the same goes for the video profiles.
Speaker 2 (31:48):
yeah, so you, you got .
I think you got my idea.
It's really in the in mind withthe guys, which wants to make
sure that it's in order and inthe process, because when you're
adding one channel it probablydoesn't matter, but usually
operations we support reallyusing hundreds of channels I'm
(32:13):
not mentioning you know yourtypical cable network has.
Speaker 1 (32:17):
What about for your users?
Has about 200 channels, 300channels.
Is that standard?
Speaker 2 (32:24):
Yeah, typical operations right now is between
200 to 300 channels.
That's probably lineups we areon.
So still, even if you're reallyconservative with the ladders,
you still have a few pages ofthis profile set up.
Yeah, that can be tricky to gothrough that and not make a
(32:46):
mistake.
Which one you are assigningtowards what?
Speaker 1 (32:49):
Yeah, so, simon, a question came in that I think
it's a good time to answer, andthat is what codecs do you
support?
And then I'll even ask afollow-on question, you know is
what codecs do you support now?
And then you know what'sinvolved with adding additional
codecs.
(33:09):
So you know, as we well know,every, you know it feels almost
like every couple of yearsthere's, you know, there's some
new standard that's that's beingintroduced, but maybe not quite
that frequency that's beingadopted, obviously, but yeah.
Speaker 2 (33:23):
So what codecs do you ?
Speaker 1 (33:25):
support.
Speaker 2 (33:26):
So what we?
What we do right now, of courseit's H.264 and HBC or H.265.
But then we also support,together with the net imports,
AV1.
Av1, yeah, which gainspopularity more and more.
Speaker 1 (33:47):
And, by the way, that's AV1 live.
I want to make that clear.
Yeah, that's not exactly thesame.
Yeah, very, very very important.
Speaker 2 (33:54):
In general, the design we we have here, uh,
basically doesn't close us toany codecs.
So if the boards which waswhich we used to to support or
to uh, uh to make theperformance of the acceleration
better, like in this case,nothing, nothing boards when the
(34:17):
net in start to support othercodecs or a newer codecs and a
version of the codecs, thesoftware will simply follow on,
yeah, yeah.
So I would say, whatever is isbelow on the hardware.
If you have some specifics, uh,we can go there.
And the second point it soundswe have to be clear about it is,
(34:37):
of course, licensing.
Yeah, but all of the codecs arejust uh, free to use.
So, uh, yeah, but also can comewith some extra costs or extra
paperwork, at least uh, to cover, sure, but that's uh, that's to
be considered.
Speaker 1 (34:52):
So, basically, yeah, that's for the distributor.
That's independent of thesolution.
Even if you use an open sourceKodak, you're going to have to
go get the appropriate licensing.
Question that I have as well isis this all single bitrate?
(35:13):
Do you support HLS?
Speaker 2 (35:19):
Do you have a packaging function?
Yeah, we have a packaging, ofcourse, as a function.
It's a separate product in theline.
But again, an advantage ofStreamVx architecture is that
you can get it transparently, soas much as we can sell this
like Lego blocks.
So you can have a transcoderwhich is outputting already
(35:41):
packaged content, but you canhave a transcoder which you're
using just to fit the currentlyexisting package.
Yeah, that's a huge advantageof our architecture.
But yeah, we can of coursesupport HLS-DASH and CMAP a huge
advantage of our architecturebut yeah, we can.
Uh, we can of course support uh,hls, dash and cmf.
Uh, we can, of course also drmthat we can serve this as a cdn.
(36:02):
So the whole flow, startingfrom the inputs towards the
distribution, it's, it's coveredand it can be covered in the in
the smooth, uh smooth path.
We can also provide therecordings of this.
So you can make live TV offline, like Timesheet, restore TV,
that kind of services.
And you can do opposite waywith the wizard.
(36:26):
We call the product wizardbecause it's also doing the
offline content towards the livechannels and, as you see here,
because it's also it's I knowthat the topic today's live, but
I just show it quickly.
To just prove the case, with theconsistency and the UX, we have
a transcoder for the VOD herealso, and it's based on the
(36:50):
watch folders.
But the funny part, or to justagain sustain the case, uh, the
configuration is, uh is exactlythe same.
So you're choosing the sametemplates, basically from the
same list.
So, uh, it's very easy.
And then you just, with oneclick, see all the files which
(37:11):
was processed by this watchfolder, or, if they are in the
process, they are progressinghere.
So it's as easy as that.
Speaker 1 (37:21):
Now, can you do live to VOD?
So let's say that I have a livechannel, but I want to create a
VOD asset.
Can you do that?
Speaker 2 (37:31):
That's exactly what we also can do.
It's, of course, very popularbecause you know the tendency
how we consume the video.
Content changed, uh, so rightnow we prefer to watch this,
even the live content, but movedinto the time, yeah, while we
have a time really to to consumeit.
So, yeah, we're doing the, thetime shifted, let's say content,
(37:55):
uh, and it's again integrateddeeply towards the, the products
of stream vx.
So I'm not having this rightnow in the demo, but you can
imagine, with the one click.
It's just if you, if you happento have our packaging system,
it just simply allows you to toshow the channels which we're
packaging for you and justsetting up how big time frame
(38:19):
you want to record, and it'sjust press start and it's done
Again this tendency to make asimple few clicks.
Configuration is all over theplace.
Speaker 1 (38:31):
Yeah, yeah.
Well, simon, we could keeptalking for much longer.
I'm looking at the clock hereand we've already gone a few
minutes over, which is fine.
I hope everybody is gettingsome value.
So I do want to open it up.
If there's some furtherquestions you have a minute or
(38:52):
two here Feel free to just putthem in the chat and we'll
address them.
At the same time, if you addcomments to the LinkedIn event,
we'll jump on and address them.
So lots of ways to get in touchwith us, but I want to, I think
(39:12):
, end with this.
I know that we're both veryexcited about OK good, I see a
question that came in and I'llget to this to this one.
But you know, and bring theusability and the performance,
you know, to what otherwise areLego blocks that are often
bolted, together with APIs andcommand line interfaces, you
know.
But we're really excited abouta project that I think we're
(40:00):
thinking about and that isbringing to market more of a
full solution, you know so, likea server.
Do you want to comment aboutthat?
Speaker 2 (40:08):
And yeah, again we really see the big power into
the kind of delivering theend-to-end solution that's the
topic of today, called like inone box.
Again it's the very importantpart is here it's still okay,
it's a box, but it's notappliance, because the hardware
(40:30):
is a out-of-the-shelf server, sothere is no specific I don't
know around modules or the driveor whatever.
It's just out of the shelfserver, so there is no specific
I don't know around modules orthe drive or whatever.
It's just out of the shelvesequipment.
The boards from NetInt okay,they are NetInt boards, but you
can, you can buy them freely andyou can exchange them freely.
There is a huge advantage whichwe didn't mention, but I love
(40:51):
it.
Actually it's touching a littlethis codex.
Imagine having 408s which doingH.264 and 5.
And then, if you happen to wantto have AV1, you're just
swapping the card, it's hotpluggable, and then suddenly you
support the technology which isjust on the market for live
(41:14):
content.
It's really fresh stuff.
You don't have to exchange thehalf of your act.
Yeah, you're just extending areally small factor board and
you can do this, uh, basicallyinstant, uh, so yeah, yeah, that
we really believe in in goingthat, that direction, that we,
uh, basically basically combineour forces let's say that way
(41:36):
nicely that we, we make a reallyfitted solution which you can
multiply, of course, but if youjust need a like a 30 channels,
it's a one box.
If you need 100 channels, it'sprobably four boxes.
Yeah, on top of that, all thesefeatures we spoke redundancy,
stability, we mentioned thatthat's really direction which is
(42:01):
, in our opinion, allowingreally huge market, huge
audience, or huge amount ofcustomers, operators, to get
into the ISE encoding,transcoding part, because, okay,
the big guys we spoke aboutoffline, about that the big guys
(42:21):
are developing their own ASICs,because obviously that's the
way you should do this and we'retrying to build something which
will do this to the rest of themarket.
For the masses.
Speaker 1 (42:35):
Exactly, exactly, well, exactly exactly, well,
that's great, well, simon.
One last question, then we willwrap up here.
So somebody is asking have youintegrated, you know, like chat,
gpt or GPT-3 into your systemso it can create program video
(43:00):
highlights automatically?
And I guess maybe you cancomment if there's just some
functionality that can do someAI infused.
Speaker 2 (43:13):
Yeah, the trick.
Here again, try to make a shortanswer for the long story.
Indeed, we have.
Actually, we have some productwhich is designed based around
the AI, to look for the videopatterns into the video, which
is more into the advertisingmarket or the metadata
(43:37):
enrichment, that kind of placesstill the regions we discussed
today.
But because of the high texturethe transcoder is having, it's
really easy to move the partswhich we're using in there
towards here.
And we was definitely thinkingabout it, because what we, what
we did with the, with theproduct I'm mentioning it's
(43:59):
called vx finder because it'sfinding the patterns we was
working on the on the project,while, uh, big operator had a
vast library of the vod assetsand he wanted to simply
implement the skip intro and thenext step is with buttons.
But he had no metadata tosupport this function.
So we basically scanned withthe AI the library and just
(44:21):
marked them precisely to theframe.
So it wasn't really more orless, it was to the frame when
the episode starts, when it ends.
So it was easy peasy to justmake those buttons and make the
experience similar to Netflix orother big guys, yeah that's
right On the basis of theprobably two or three days of
(44:43):
processing, instead of sittingprobably few persons and making
a lot of mistakes trying to fitit manually.
The one thing is right now it'snot there in a GPT-3 manner, I
would say, but it's really closeand for us it's very easy
(45:04):
probably to experiment.
So I would be glad to follow upon that one, if all goes well
here.
Speaker 1 (45:11):
Well, simon, thank you again, Really appreciate
your time.
We're really excited about whatyou guys are doing and
hopefully this was of interestto the viewers and so for all
those who joined us.
Thank you again for watchingVoices of Video.
We have new episodes comingvery regularly, so just watch
(45:35):
the schedule and if you're onour email list, we'll let you
know as well.
So thank you again, greatThanks.
Speaker 2 (45:42):
Mark, thanks for having me.
It was a pleasure.
This episode of Voices of Videois brought to you by NetInt
Technologies.
If you are looking forcutting-edge video encoding
solutions, check out NetInt'sproducts at netintcom.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!