`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 1 of 8
`
`
`
`
`
`
`EXHIBIT D
`EXHIBIT D
`
`
`
` 1 IN THE UNITED STATES DISTRICT
`
` 2 SOUTHERN DISTRICT OF NEW YORK
`
` 3
`
` 4 NETWORK-1 TECHNOLOGIES, INC., )
` )
` 5 Plaintiff, )
` )
` 6 vs. ) Case Nos.
` ) 14 Civ. 2396 (PGG)
` 7 GOOGLE, INC., and YOUTUBE LLC, ) 14 Civ. 9558 (PGG)
` )
` 8 Defendants. )
`
` 9
` London, United Kingdom,
`10
` Wednesday, October 30th, 2019
`11
`
`12 **Highly Confidential - Outside Counsels' Eyes Only**
`
`13 **Prosecution/Acquisition Bar Materials**
`
`14
`
`15 Videotaped deposition of MATTHIAS KONRAD, in
`
`16 the above-entitled action, by and before, AUDREY
`
`17 SHIRLEY, Qualified Real-time Reporter, Accredited,
`
`18 Court Reporter and Member of the British Institute
`
`19 of Verbatim Reporters, at the offices of Wilmer
`
`20 Hale, 49 Park Lane, London, W1K 1PS, United
`
`21 Kingdom, commencing at 9:08 a.m.
`
`22
`
`23
`
`24
`
`25
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 2 of 8
`
`1
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
` 1 that need to be able to utilize this capacity. For
`
` 2 some teams, it's important to have their services
`
` 3 run in certain locales, other teams just need the
`
` 4 resources. I would say ContentID is a team that
`
`10:53 5 just needs the resources, so we don't put too many
`
` 6 constraints.
`
` 7 There are maybe considerations to not
`
` 8 have everything in just one data center because of
`
` 9 the potential of power outages and other things and
`
`10:54 10 having higher redundancy to be able to have better
`
`11 reliability. But I don't believe we ever had
`
`12 a conversation as to it should be all in the US or
`
`13 not in the US as part of any such consideration for
`
`14 reliability.
`
`10:54 15 Q. Do you have any understanding of what
`
`16 would be required to, for example, not have any in
`
`17 the US?
`
`18 A. I believe that would be -- the first
`
`19 thing that we would be doing is we would contact
`
`10:54 20 the resource planners, Christine Moor's team, to
`
`21 make sure we have enough provisioning in other data
`
`22 centers. I guess there would be other teams that
`
`23 could be locating their services out of non-US data
`
`24 centers into US data centers and we could get those
`
`10:55 25 resources there and, once we would have that, it
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 3 of 8
`
`68
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 4 of 8
`
` to bring it up in non-US data
`
` 1 would be as simple
`
`
`
` 3 centers, yes.
`
` 4 Q. Have there been any discussions about
`
`10:55 5 whether that's feasible and under what
`
` 6 circumstances Google would be able to do that?
`
` 7 A. I have been in conversations about
`
` 8 this yesterday as part of the preparation to this
`
` 9 meeting. We have never otherwise discussed this.
`
`10:55 10 Q. And when you say you were in
`
`11 conversations yesterday, what do you mean? What
`
`12 are you referring to?
`
`13 A. As part of the preparation of this
`
`14 deposition we discussed whether it would be
`
`10:55 15 difficult to operate ContentID just in data centers
`
`16 exclusively outside of the United States.
`
`17 Q. And that was the first time any
`
`18 analysis had been done of that type to your
`
`19 knowledge?
`
`10:55 20 A. I do not recollect that otherwise we
`
`21 have ever done any analysis, like other people
`
`22 might have. And I wouldn't say we did very
`
`23 specific analysis but, like, the experts on that --
`
`24 like, this would be Oleg again on the team --
`
`10:56 25 sounded very solid that it would be a
`
`69
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 5 of 8
`
` 1
`
` 2 Q. Why hasn't Google done it already?
`
` 3 MR. HARDY: Objection to form.
`
` 4 THE WITNESS: I mean, yeah, so going
`
`10:56 5 back to my previous answer, I don't -- I could
`
` 6 neither tell you why Google hasn't moved ContentID
`
` 7 fully to the US versus fully to the non-US.
`
` 8 BY MR. LEDAHL:
`
` 9 Q. You mentioned that it would require
`
`10:56 10 adjustment or re-provisioning of some services to
`
`11 different data centers. What other services would
`
`12 have to move to make that accommodation?
`
`13 A. So the -- I mean, this would be one
`
`14 solution how -- like, if you look at, like, what
`
`10:57 15 our resource use is compared to YouTube's total
`
`16 resource usage or Google's total resource usage,
`
`17 it's a very, very small fraction. So, I would
`
`18 believe that, even without moving any services, we
`
`19 could -- we could operate all of ContentID, I'm
`
`10:57 20 sure we could operate all of ContentID with today's
`
`21 resources outside of the US.
`
`22 In general, the more flexibility you
`
`23 have, the higher utilization can be for these kinds
`
`24 of systems. So, whenever a service can run
`
`10:57 25 everywhere, there is some advantages in overall
`
`70
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 6 of 8
`
` 1 utilization. But whether you can run it everywhere
`
` 2 outside of the United States or everywhere doesn't
`
` 3 really matter in overall utilization. So, I don't
`
` 4 believe, for example, that it would increase
`
`10:58 5 resource cost for us to do so.
`
` 6 Q. So, what would be the cost associated
`
` 7 with making that switch?
`
` 8 A. So, I'm a bit like -- the thing that
`
` 9 I cannot assess is general resource cost in data
`
`10:58 10 centers outside of US versus inside of US. So if
`
`11 there's a difference in how data centers operation
`
`12 prices are, that might, of course, factor into
`
`13 that. I would be surprised if that was too big.
`
`14 There's going to be a bit of cost of people that
`
`10:58 15 have to do a bit of planning. I couldn't assess
`
`16 a dollar number, but it might
`
`
`
`18 Q. And how did you arrive at that
`
`19 figure?
`
`10:59 20 A. That is based on conversations we had
`
`21 with Oleg yesterday and the main question was is it
`
`22 -- is where a -- where a match system or an indices
`
`23 is run, is it
`
`
`
`
`
`
`
`71
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 7 of 8
`
` He has
`
`
`
`
`
` 3 Q. And do you know what information Oleg
`
` 4 looked at to make those assessments?
`
`10:59 5 A. So, Oleg would often be involved in
`
`
`
`
`
`
`
` 9 done such things multiple times, also when we had
`
`11:00 10 rollouts of new versions. So, based on that,
`
`11 I believe he has very firm knowledge as to how much
`
`12 it would take to move an instance of a match system
`
`13 from one data center to another.
`
`14 Q. Now, when you talk about moving
`
`11:01 15 instance of the match system, are you referring to
`
`16 just the portion that -- basically from the
`
`
`
`
`
`19 A. Yeah, that's the portion I'm talking
`
`11:01 20 about.
`
`21 Q. Okay. So you're not referring to the
`
`
`
`
`
`
`
`
`
`72
`
`MATTHIAS KONRAD-Highly Confidential
`
`
`
`Case 1:14-cv-02396-PGG-SN Document 269-4 Filed 09/07/22 Page 8 of 8
`
`
`
` 1 A. In that specific case I was referring
`
` 2 to the
`
` 3 Q. Do you know how many -- so we talked
`
` 4 about the number of data centers and I -- and
`
`11:01 5 I think you said historically you thought it was
`
`
`
`
`
`
`
` 9 A. Exactly.
`
`11:01 10 Q. Okay.
`
`11 A. I do not know in how many data
`
`12 centers we do transcoding. I would assume it's
`
`13 more, but I wouldn't know.
`
`14 Q. And you said you didn't know where
`
`11:02 15 the claiming portions of the system ran?
`
`16 A. That's exact -- actually, I would
`
`17 assume it
`
`
`
`
`
`11:02 20 don't know where they're running. It's overall
`
`21 a computationally much less heavy process, so ...
`
`22 Q. Is the -- is the matching portion, is
`
`23 that what you're describing as the computationally
`
`24 heavy portion?
`
`11:02 25 A. So, both the fingerprint and the
`
` but I really
`
`
`
`73
`
`MATTHIAS KONRAD-Highly Confidential
`
`