[00:00:05] Speaker 01:
The United States Court of Appeals for the Federal Circuit is now open and in session.

[00:00:11] Speaker 01:
God save the United States and this honorable court.

[00:00:15] Speaker 02:
Good morning, please be seated.

[00:00:19] Speaker 02:
We've consolidated the first three cases for argument today.

[00:00:24] Speaker 02:
They are 162063, 162067, and 162072, all titled Delphic Corporation v. Mittal.

[00:00:34] Speaker 02:
Mr. Hatton, whenever you're ready.

[00:00:37] Speaker 03:
May I please record Dave Hatton for Delphics?

[00:00:43] Speaker 03:
The PTAB made a fundamental claim construction errors in this case.

[00:00:49] Speaker 03:
The first was by equating a file system with the database.

[00:00:54] Speaker 03:
From that false premise, the p-tab rewrote the definition of database block to be a unit of data used by a file system.

[00:01:05] Speaker 02:
Well, we all agree that the broadest reasonable interpretation is what governs here.

[00:01:10] Speaker 02:
So I understand the portion of the spec that you point to that you argue sounds definitional.

[00:01:16] Speaker 02:
But otherwise, the references in the specification to the database contain really, really, really broad language, right?

[00:01:23] Speaker 02:
Column four, for example, a database comprises data stored in a computer for use in computer-implemented applications.

[00:01:32] Speaker 02:
There are really broad descriptions of the database in this specification, right?

[00:01:36] Speaker 02:
So how do you get around that?

[00:01:37] Speaker 03:
Sure.

[00:01:38] Speaker 03:
I agree, Your Honor.

[00:01:39] Speaker 03:
With respect to column four,

[00:01:41] Speaker 03:
It is true that a database comprises data used by a computer system.

[00:01:47] Speaker 03:
That's talking about what is inside a database.

[00:01:52] Speaker 03:
There's no dispute about that.

[00:01:55] Speaker 03:
But the patent goes on and describes, and every claim requires, that the database is accessed by a database server.

[00:02:03] Speaker 03:
So the patent is talking about conventional databases as provided by the companies that are identified in column five,

[00:02:11] Speaker 03:
Oracle, Sybase, Microsoft.

[00:02:14] Speaker 03:
These are all databases that are structured, organized collections of data that are optimized for searching and retrieval.

[00:02:22] Speaker 03:
And they're accessed through a database server, as the patent describes.

[00:02:27] Speaker 03:
And in that context, the notion of a database block is a well-known concept.

[00:02:34] Speaker 03:
And there's really no dispute about this in the evidence in the record.

[00:02:38] Speaker 03:
Treatises like Professor Molina's book

[00:02:41] Speaker 03:
Database systems, the complete book, describe this idea.

[00:02:45] Speaker 03:
The documentation from every database provider referenced in the patent describes a database block or a database page as a known concept.

[00:02:55] Speaker 03:
It's the logical unit of data that the database server uses to read and write to the database.

[00:03:02] Speaker 03:
And to do that, that has to be defined by the database server, and it has to include metadata that the database server uses

[00:03:11] Speaker 03:
to interpret that block of data and put it into the right place in the organized structure.

[00:03:17] Speaker 03:
And that's exactly what the 808 patent defines as a database block.

[00:03:24] Speaker 03:
And Atifio, the petitioner's expert, agreed.

[00:03:31] Speaker 03:
He cited the same documentation and said that a SQL server data page from Microsoft

[00:03:38] Speaker 03:
is the database block that's described in this patent, and that it includes metadata.

[00:03:44] Speaker 03:
It went through the same thing for IBM DB2 and Oracle, and in all cases, there was agreement.

[00:03:51] Speaker 04:
Do all databases use metadata, or can that function be performed by different methods?

[00:04:00] Speaker 03:
There's no indication in any of the database literature or in the patent

[00:04:07] Speaker 03:
that there are database blocks that don't include metadata.

[00:04:12] Speaker 03:
And if you think about it, that has to be the case.

[00:04:14] Speaker 03:
If the database server is reading some unit of data, it has to place in a logical arrangement, like a table in a relational database.

[00:04:23] Speaker 03:
It has to know how that data, those bits of user data, relate to that logical structure.

[00:04:30] Speaker 04:
And that is what the metadata in the header... Can you point to me somewhere in the patent or the record that says that's the only way that databases can work?

[00:04:39] Speaker 03:
Well, I can't point to you a negative that says all databases have database blocks.

[00:04:45] Speaker 03:
But all database blocks... I'm not asking for a negative.

[00:04:49] Speaker 04:
I'm asking for a positive that all databases have metadata.

[00:04:56] Speaker 03:
The patent defines a database block as including metadata.

[00:05:00] Speaker 03:
The patent describes the use of the metadata at each step in this process by identifying the database blocks that need to get transferred to the database storage system.

[00:05:10] Speaker 04:
Right, I get that.

[00:05:11] Speaker 04:
But I thought that the board's opinion suggested or the record suggested that databases could operate without metadata.

[00:05:19] Speaker 03:
There is no evidence in the record of a database operating without a metadata.

[00:05:24] Speaker 03:
But the board pointed to for the contention that there could be

[00:05:29] Speaker 03:
database block without metadata was this reference to a Google Bigtable data structure.

[00:05:37] Speaker 03:
It was provided for the first time in the petitioner's reply.

[00:05:42] Speaker 03:
Now there's nothing in that document that actually says that that is a database.

[00:05:46] Speaker 03:
It says it operates like a database.

[00:05:48] Speaker 03:
And there's nothing in that document.

[00:05:50] Speaker 04:
More importantly... Are we getting tied up on a specific definition of database?

[00:05:54] Speaker 04:
Because

[00:05:55] Speaker 04:
I mean, your claims don't confine them to relational databases.

[00:05:59] Speaker 04:
They say database, which seems to be a pretty broad term.

[00:06:02] Speaker 03:
That's true, but they all require a database block.

[00:06:06] Speaker 04:
But you're now arguing that this Google document is in a database, even though it operates like a database, which sounds like semantics to me, at least under a BRI.

[00:06:16] Speaker 03:
Sure.

[00:06:17] Speaker 03:
And that was just one point about the Google document.

[00:06:20] Speaker 03:
The bigger point with the Google document is that

[00:06:24] Speaker 03:
it doesn't refer to database blocks at all.

[00:06:28] Speaker 03:
So you can't define the balance of a database block based on a reference that doesn't even talk about database blocks.

[00:06:37] Speaker 03:
So the claimed term is database block.

[00:06:40] Speaker 03:
The question is, what is the ordinary meaning of that term in light of the specification?

[00:06:45] Speaker 02:
And the board relied on two embodiments and said they were distinct.

[00:06:49] Speaker 02:
So you've got the file sharing embodiment.

[00:06:51] Speaker 02:
Sure.

[00:06:53] Speaker 03:
Yeah, let me go to that.

[00:06:55] Speaker 03:
The first point on that is the patent makes very clear that the database blocks exist in the source database system from which they are transferred to this database storage system to create the virtual database files.

[00:07:11] Speaker 03:
There's nothing in the patent

[00:07:13] Speaker 03:
that would suggest that the content of those database blocks changes depending on how they are transferred.

[00:07:21] Speaker 03:
There's no reason to say that.

[00:07:22] Speaker 03:
The board talks about you could pack them with metadata for the streaming body and then somehow take that out.

[00:07:29] Speaker 03:
That is just made up.

[00:07:30] Speaker 03:
There's nothing in the patent that says that at all.

[00:07:33] Speaker 03:
So what the board points to for the non-streaming embodiment that they claim exists, they point to figure one.

[00:07:43] Speaker 03:
And they point to some language at column.

[00:07:46] Speaker 03:
I should find the right language.

[00:07:50] Speaker 03:
At column six, this is at line, and I'm in the 808 pattern.

[00:07:56] Speaker 03:
Line 11, they say figure one illustrates one embodiment of how information may be output from a picture.

[00:08:02] Speaker 02:
I'm sorry, where are you in column eight?

[00:08:03] Speaker 03:
I'm sorry, yeah.

[00:08:05] Speaker 03:
I'm in column six, line 11 of the 808 pattern, which is describing figure one.

[00:08:12] Speaker 03:
It says, figure one illustrates one embodiment for how information may be copied from a production database to a database storage system.

[00:08:20] Speaker 03:
And then the board puts in ellipses, dot, dot, dot, using file sharing system.

[00:08:27] Speaker 03:
But if you look at what the patent actually says, it says, and provisioned as virtual databases using a file sharing system.

[00:08:36] Speaker 03:
So the board left out the part that actually talks about how the file sharing system is used.

[00:08:41] Speaker 03:
which is in provisioning, that is, the database servers in these virtual database systems accessing the database storage system.

[00:08:52] Speaker 03:
If you look farther in column six, it confirms that in all cases, and this is farther down at column six around line 28, it says the production database system 110

[00:09:12] Speaker 03:
response by sending information stored in the production database as a stream of data.

[00:09:19] Speaker 03:
So the board just misread figure one in coming up with this alternative file sharing embodiment.

[00:09:27] Speaker 02:
What the board says is the database files can be copied, files to be copied have defined boundaries and known structures.

[00:09:35] Speaker 02:
This is referring to what they call the file sharing embodiment.

[00:09:38] Speaker 02:
and database blocks stored in the files can be accessed directly.

[00:09:42] Speaker 02:
And that seemed to indicate that metadata was unnecessary.

[00:09:47] Speaker 03:
Right.

[00:09:47] Speaker 03:
But none of that is in the patent.

[00:09:49] Speaker 03:
When it says that the patent is describing in all embodiments that the database blocks in the source database are accessed through what it calls the vendor interface module.

[00:10:02] Speaker 03:
And the patent explains that the vendor interface module is essentially an add-on

[00:10:07] Speaker 03:
to the database server provided by a vendor like Oracle.

[00:10:11] Speaker 03:
The example of the patent is Oracle RMAN, which stands for Recovery Manager.

[00:10:16] Speaker 03:
So that is a way to read the database blocks from the production database and output them to some backup source.

[00:10:29] Speaker 03:
None of that describes at all that there is not metadata in those blocks.

[00:10:35] Speaker 03:
or that the blocks are somehow different when they're read through this vendor interface module or through the database server.

[00:10:45] Speaker 03:
The other thing that the board kind of relies on, and I think this is a column A, and it might be what your honor was referring to, it says another embodiment, and this is a column A to the 808 pattern around line 32.

[00:11:00] Speaker 03:
It says in another embodiment, the vendor interface module

[00:11:05] Speaker 03:
may retrieve the necessary database blocks from storage level snapshots of the production database or from production or clones.

[00:11:14] Speaker 03:
Now, all that is talking about is that the database blocks can come from not the live production database that you're running to run your business, but a copy of it, right?

[00:11:27] Speaker 03:
And this is actually referring to these snapshots or database clones are really what this NetApp priority is about.

[00:11:35] Speaker 03:
Right?

[00:11:36] Speaker 03:
If you run your production database and use a NetApp storage system for your production database, you can create these snapshots and clones that the NetApp prior talks about at the storage level.

[00:11:49] Speaker 03:
But when the system, the vendor interface module reads the database blocks from those copies, those are the same database blocks that would be in the production system.

[00:11:59] Speaker 03:
There's nothing to suggest their form is different or that this process operates any differently

[00:12:05] Speaker 03:
whether you're reading the live production database or a copy.

[00:12:10] Speaker 03:
There's another thing that the board just kind of misquotes that I want to get to.

[00:12:17] Speaker 03:
And it might go to Judge Hughes' question about the breadth of databases.

[00:12:23] Speaker 03:
So the board kind of goes off repeatedly on a statement that the patent says it applies to every database.

[00:12:34] Speaker 03:
And this is what they quote from, is it column five of the 808 patent.

[00:12:42] Speaker 02:
And we're in the board opinion.

[00:12:44] Speaker 02:
What were you saying from the board?

[00:12:46] Speaker 04:
Which case number are we talking about, the 808 patent?

[00:12:49] Speaker 04:
Yes, Your Honor.

[00:12:50] Speaker 04:
I need to get the right appendix.

[00:12:52] Speaker 04:
I'm sorry, Your Honor.

[00:12:53] Speaker 04:
Which case number?

[00:12:54] Speaker 04:
We've got three different case numbers.

[00:12:55] Speaker 04:
Which appendix are we looking at?

[00:13:03] Speaker 02:
Okay, sure.

[00:13:04] Speaker 02:
Go ahead.

[00:13:05] Speaker 02:
Is it the 2063?

[00:13:06] Speaker 02:
Is that the 808 case that we're in?

[00:13:08] Speaker 02:
Yeah, let me just find one.

[00:13:20] Speaker 03:
Yeah, so I was looking at the 014 is the one I was looking at, which is the one that was attached to our opening three.

[00:13:32] Speaker 03:
the first one.

[00:13:36] Speaker 03:
But going to the point, Your Honor, which is what the PTAB relied on from column five, where they say that the claims cover all databases.

[00:13:50] Speaker 03:
If you read it in context, it doesn't really say that at all.

[00:13:53] Speaker 03:
What it says is, and this is in the 808 patent calling, starting at line 11, says the term production database

[00:14:01] Speaker 03:
is used in particular examples to illustrate a useful application of the technology.

[00:14:07] Speaker 03:
However, it can be appreciated that the techniques disclosed can be used for any database, regardless of whether the database is used as a production database.

[00:14:17] Speaker 03:
So all this is saying is that, you know, the primary example, we talked about the source being the production database, but you can use a non-production database and perform the same operation.

[00:14:29] Speaker 03:
This is not a statement that redefines the meaning of database block, which is a term that's actually used in the claims.

[00:14:44] Speaker 03:
So there was no dispute going back to kind of the main kind of jujitsu move that the board used, which is redefining

[00:14:57] Speaker 03:
database block to be a file system block.

[00:15:02] Speaker 03:
No party, no expert, no document equated a file system with a database in this case, right?

[00:15:12] Speaker 03:
The petitioner pointed to the DB2 or Microsoft SQL Server or Oracle as being the database that could be stored using a NetApp system.

[00:15:25] Speaker 03:
The NetApp prior itself, there's this Sanders reference that talks about cloning DB2 databases.

[00:15:32] Speaker 03:
That reference distinguishes the DB2 database, IBM database, from the NetApp storage system.

[00:15:40] Speaker 03:
And it says to create the database, you go to your IBM database server, and you type in create database.

[00:15:48] Speaker 03:
So this equating those two different constructs is something that the PTAB came up with entirely on its own.

[00:15:56] Speaker 03:
The whole point of a file system is to hide how data is stored on disk from applications like databases so they don't need to understand it.

[00:16:09] Speaker 03:
And the experts agree.

[00:16:11] Speaker 03:
There is a boundary line there.

[00:16:15] Speaker 03:
A database cannot figure out and does not get access to the blocks, the units of data that a file system uses on disk.

[00:16:24] Speaker 03:
And a file system doesn't get access to the database blocks.

[00:16:29] Speaker 03:
It doesn't know anything about the database blocks that are used by the database.

[00:16:34] Speaker 03:
And that's confirmed both by both experts, but also the Oracle documentation and the Molina treatise on database systems.

[00:16:45] Speaker 03:
So once that kind of technical dispute is clarified,

[00:16:52] Speaker 03:
And the words of the definition of the 808 patent, a database block is applied.

[00:16:59] Speaker 03:
There's no way it can be read on these NetApp file system prior.

[00:17:05] Speaker 03:
That's why the PTAB had to redefine databases as file systems.

[00:17:11] Speaker 03:
The other thing that the PTAB did here, which is to discard the definition of virtual database in every patent.

[00:17:22] Speaker 03:
And again, it did that so that it can read that term on a file system.

[00:17:28] Speaker 03:
The patent has an explicit definition, and it talks about decoupling the physical implementation of the files from their logical use by a database server.

[00:17:43] Speaker 03:
And as Professor Chinoy explained, that is the standard definition of virtualization in computer science.

[00:17:52] Speaker 03:
If you have a virtual server, it is decoupling the physical implementation.

[00:17:58] Speaker 03:
You're sharing a common piece of hardware through software, but the applications that run on it can't tell the difference.

[00:18:07] Speaker 03:
The logical use is the same.

[00:18:09] Speaker 03:
That's what this patent is talking about when it's talking about decoupling the implementation of the files from their logical use.

[00:18:17] Speaker 03:
And it's shown in figure 10 and figure 12.

[00:18:21] Speaker 03:
on this database storage system, it creates a collection of these shared database blocks.

[00:18:28] Speaker 03:
It is separate from any set that is ever on the production system and arranges them in a new way and allows these different virtual databases to share them.

[00:18:38] Speaker 03:
That is the decoupling of the physical implementation.

[00:18:42] Speaker 03:
But to the database server, it's looking at the database storage system, the files logically work the same.

[00:18:50] Speaker 03:
Right, that's the test for successful virtualization.

[00:18:54] Speaker 03:
The thing that is using the virtualized resources can't tell the difference, and that's what happens here.

[00:19:01] Speaker 03:
Now instead, the PTAB came up with its own construction that talks about mapping files to physical addresses.

[00:19:10] Speaker 03:
But mapping a file to a physical address is what all file systems do, right?

[00:19:16] Speaker 03:
Going back 40 years to UNIX, right,

[00:19:20] Speaker 03:
They use this same system of inodes and pointers to be able to map a file to a collection of blocks on disk.

[00:19:29] Speaker 03:
That is not what virtualizes a database in this pattern.

[00:19:33] Speaker 03:
So if you apply the PTAS kind of concocted construction of that term, it will literally cover any copy of any file.

[00:19:44] Speaker 03:
and they amended it for the first time in their file written description, but in a way that doesn't solve the problem.

[00:20:04] Speaker 03:
Let me go to kind of the last point on the

[00:20:08] Speaker 03:
808 patent, which is claim 36.

[00:20:11] Speaker 02:
And I think this kind of really should... Can I just ask you a logistical question, which is this is the issue that's unique to one of the appeals.

[00:20:19] Speaker 02:
It is.

[00:20:19] Speaker 02:
Whereas everything else... Everything else is common.

[00:20:21] Speaker 03:
Yes, Your Honor.

[00:20:22] Speaker 03:
So the 808 patent in claim 36 has these claims that talk about compressing prior to storage.

[00:20:30] Speaker 03:
And the PTAB's construction saying that that covers

[00:20:36] Speaker 03:
compressing at some point in the past, decompressing and then storing, it just turns the claim on its head, right?

[00:20:45] Speaker 03:
It can't mean that at some point in the history of the piece of data, it was at one time compressed.

[00:20:50] Speaker 03:
And it is then decompressed and stored in decompressed form.

[00:20:55] Speaker 03:
And this distinction is actually important, right?

[00:20:58] Speaker 03:
Because the difference between, or one of the differences between the NetApp prior art,

[00:21:05] Speaker 03:
And in particular, this snap mirror technology that's described in the Patterson reference and also in the Edward reference, is that snap mirror creates a physical exact mirror of the source.

[00:21:21] Speaker 03:
So it is essentially what's the prior art they're talking about, snap mirror is, is what's shown in Figure 19 is prior art in the 808 pattern.

[00:21:30] Speaker 03:
It is mirroring

[00:21:32] Speaker 03:
one source database to a second site.

[00:21:37] Speaker 03:
And that is all NetApp can do because NetApp operates at the physical file system level.

[00:21:45] Speaker 03:
So every block on disk gets replicated from the first site to the second site in SnapNR.

[00:21:52] Speaker 03:
So you can never have a point in time copy of your second site that you didn't also save at your first site.

[00:22:00] Speaker 03:
What it also means is the physical implementation of those two sets of files has to be identical, block for block identical.

[00:22:10] Speaker 03:
So if your source database is not compressed, you cannot mirror it storing compressed data on the second site in NetApp.

[00:22:19] Speaker 03:
It won't work because it operates at that physical block level, right?

[00:22:24] Speaker 03:
The difference in the Delphix pattern is

[00:22:27] Speaker 03:
Database blocks are a logical construct, right?

[00:22:30] Speaker 03:
It's a set of data that is understood by the database server.

[00:22:35] Speaker 03:
Now, when Delphix copies some of those from the source database to the database storage system, they can store them physically in any way they want, including compressed, whereas the patent also describes encrypted, or in the case of sensitive data like social security numbers,

[00:22:53] Speaker 03:
You can mask those out when you store them on your virtual database system so that when you do testing and development, you don't expose that confidential information.

[00:23:03] Speaker 03:
That's why the 808 pattern can compress data prior to storing it, store it in compressed form as the claim requires.

[00:23:11] Speaker 03:
But you can't do that in NetApp because you have to have exactly the same bits on disk in the NetApp system on the source as on the mirrored destination.

[00:23:21] Speaker 03:
which just kind of shows what the fundamental difference in these technologies is.

[00:23:26] Speaker 03:
I see my yellow light is on.

[00:23:27] Speaker 03:
I will sit down.

[00:23:28] Speaker 03:
Good morning, Your Honors.

[00:23:43] Speaker 00:
May it please the Court?

[00:23:45] Speaker 00:
I'd like to begin by addressing a few points my friend raised in the past few minutes.

[00:23:49] Speaker 00:
And starting with the question from Judge Hughes about whether there's evidence that all databases must use metadata.

[00:23:58] Speaker 00:
And to begin with, there's evidence that some databases do not, including flat files, CSV files.

[00:24:04] Speaker 02:
But isn't that the ordinary meaning?

[00:24:06] Speaker 02:
I mean, outside of this patent, was there an expert that said that wasn't their agreement that that was the plain and ordinary meaning of the language that would include metadata?

[00:24:17] Speaker 00:
If I understand correctly, there was not agreement that there was an ordinary meaning that database blocks have to contain metadata.

[00:24:24] Speaker 00:
Now, what there appears to have been agreement about is that several popular commercial relational databases have database blocks or pages that, at least in some cases, have metadata.

[00:24:39] Speaker 00:
Now, there's also evidence that many databases have metadata, but not necessarily in the database blocks.

[00:24:46] Speaker 00:
And the Google example, the Bigtable, is a great example of that.

[00:24:50] Speaker 00:
The Google reference explains that it has, and it calls them blocks, it doesn't use the exact term database block, but it has blocks of data and an enormous amount of data, same database used for the Google indexing when you do Google searches.

[00:25:05] Speaker 00:
And it explains that there is metadata that is stored external to those blocks.

[00:25:09] Speaker 00:
And that was an example that was cited by Actifio, the petitioner's expert, Dr. Zadek,

[00:25:16] Speaker 00:
And that rebuts the idea that all databases necessarily have metadata in the block.

[00:25:25] Speaker 00:
We would agree that, generally speaking, with a large database, metadata is going to make it easier to determine where these blocks go.

[00:25:31] Speaker 00:
The patent talks about taking blocks from one server to another, virtualizing them.

[00:25:36] Speaker 00:
Metadata will certainly help with that.

[00:25:38] Speaker 00:
But the question here is, do these claims require the metadata to be stored in the database block?

[00:25:46] Speaker 00:
And we would submit that if that was so important, if that's what the patent applicant intended, why does that not appear in the claims?

[00:25:54] Speaker 02:
Well, it appears in the specification, as you know from Phillips and at least when we're not in broadest reasonable interpretation land, we rely on the spec to inform the claims.

[00:26:05] Speaker 02:
It seems to me the language that's been talked about in this case in column two is quite specific about metadata being included.

[00:26:14] Speaker 02:
I mean, if you were, we're trying to do that, let's leave aside the claims now.

[00:26:17] Speaker 02:
But in the specification, if you were trying to make this kind of a definitional, a psychography thing, how else would you, would you say it other than the way it's included in the specification here?

[00:26:28] Speaker 02:
Right?

[00:26:28] Speaker 02:
I mean, look at the language in column two.

[00:26:30] Speaker 02:
A portion of the database block stores metadata associated with the database block.

[00:26:36] Speaker 02:
How could one have made the point that your friend on the other side is trying to claim here.

[00:26:44] Speaker 02:
better more clearly than by saying what it said in the spec?

[00:26:49] Speaker 00:
Well, I would be speculating, but one could certainly say that a database block must include metadata or necessarily includes metadata and explain why that is so important or what the consequence would be of excluding the metadata.

[00:27:04] Speaker 02:
Well, it says a portion of the database block stores metadata associated.

[00:27:09] Speaker 02:
You're saying if it had said the portion of the database block

[00:27:13] Speaker 02:
has to store metadata?

[00:27:15] Speaker 02:
I mean, I'm not sure how you're saying you could make that clear.

[00:27:20] Speaker 00:
So our read of that paragraph, of that sentence, is that that is describing what some embodiments can do as opposed to a definition.

[00:27:28] Speaker 00:
If you look earlier in that paragraph in column two, there's the discussion of production database and source database.

[00:27:34] Speaker 00:
And the spec is very clear that those terms are used interchangeably.

[00:27:38] Speaker 00:
It's a clear definition.

[00:27:40] Speaker 00:
And the parties agreed that when

[00:27:42] Speaker 00:
two sentences, excuse me, two sentences above, a database block is, and it goes on to describe a unit of data used by a database.

[00:27:51] Speaker 00:
But I don't want to lose sight of the fact that this language that is relied upon so heavily by Delphics appears in only one of the three patents that are at issue, and yet Delphics has argued that there should be a consistent interpretation across.

[00:28:04] Speaker 00:
It does not appear in the claims, it appears only sporadically in the spec, and some of those descriptions in the specification

[00:28:12] Speaker 00:
reference metadata stored in other places, such as in a file and not in the block.

[00:28:17] Speaker 00:
In addition, the spec consistently describes the metadata as being associated with the block, which the board correctly pointed out doesn't require it to be in the block.

[00:28:27] Speaker 00:
It could certainly be somewhere else and still be associated with the block.

[00:28:31] Speaker 05:
Are you saying that that statement in column two looks sort of casual rather than definitional?

[00:28:40] Speaker 05:
As if it says, can include metadata.

[00:28:44] Speaker 05:
Equivalent to that.

[00:28:46] Speaker 00:
If it had said, can include metadata, certainly my job would be easier here today.

[00:28:51] Speaker 05:
It says includes.

[00:28:55] Speaker 05:
Is that not definitional?

[00:28:59] Speaker 00:
So if you look at that statement in isolation, I would agree that certainly seems probative of database blocks that store metadata.

[00:29:08] Speaker 00:
But what the board did, and what is appropriate under this court's precedent, is looked at the specification as a whole.

[00:29:15] Speaker 00:
And the specification as a whole has the examples I just gave, as well as the file sharing embodiment, as well as the streaming.

[00:29:23] Speaker 02:
And also, critically- But the file sharing, so the answer is they didn't call out specifically the inclusion of metadata in the file sharing embodiment.

[00:29:31] Speaker 02:
That's what the board relies on, right?

[00:29:35] Speaker 00:
Well, so they do.

[00:29:37] Speaker 00:
dig deeper and they say, would you need to have metadata?

[00:29:40] Speaker 00:
Would this embodiment work?

[00:29:42] Speaker 00:
Because Delphix had argued this wouldn't work without the metadata.

[00:29:45] Speaker 00:
First of all, it could work if the metadata were stored somewhere else.

[00:29:48] Speaker 00:
But secondly, the board noted that if you're just moving the file blocks from one place to another, you're not analyzing, you're not interpreting the blocks.

[00:29:57] Speaker 00:
And column 13, where counsel referred us to earlier, actually talks about the file sharing example being perfect for

[00:30:04] Speaker 00:
situations when the database server does not support appropriate APIs for extracting the necessary information.

[00:30:10] Speaker 00:
And then it talks about going on and being able to move the data.

[00:30:14] Speaker 00:
And again, it talks about metadata associated with the blocks.

[00:30:19] Speaker 00:
And a different portion of the specification I think is really relevant here.

[00:30:22] Speaker 00:
In the 808 patent, if we look at column 31, column 31 has some disclaimers or language to aid one in interpreting the patent.

[00:30:34] Speaker 00:
And here the patentee says, this is column 31, starting at line 44, structures and functionality presented as a single component may be implemented as separate components.

[00:30:45] Speaker 00:
So to the extent that it's not clear whether or not, in all cases, one must include the metadata in the database block, particularly where it's not recited in the claim, the patentee seems to have told us if the patent says you could do it with one unit, you could do it with two units, which

[00:31:04] Speaker 00:
should point out is undisputed.

[00:31:05] Speaker 00:
The NetApp prior art does have metadata.

[00:31:09] Speaker 00:
The Edwards reference, which is the primary reference across these cases, is replete with references to metadata.

[00:31:15] Speaker 00:
Granted, it does not explicitly talk about storing metadata in the database blocks.

[00:31:20] Speaker 00:
But nonetheless, there is metadata, and it is used.

[00:31:24] Speaker 02:
So wait a minute.

[00:31:24] Speaker 02:
So you've got two opposite arguments here.

[00:31:27] Speaker 02:
One is that the claims don't compel the use of metadata in blocks.

[00:31:30] Speaker 02:
And then you're saying that even if it does,

[00:31:33] Speaker 02:
the prior right also includes metadata?

[00:31:35] Speaker 02:
Are those alternative arguments?

[00:31:37] Speaker 00:
Not alternative arguments.

[00:31:42] Speaker 00:
Well, you could call them alternative arguments.

[00:31:45] Speaker 00:
To the extent that those claims are interpreted to require metadata, I'm just pointing out that the NetApp references do use metadata.

[00:31:54] Speaker 00:
Now, to the extent that claims require, or this court disagrees with the board and says that the claims require metadata in the database blocks,

[00:32:02] Speaker 00:
There is an alternative argument to that as well, which is that some of the prior art did have what Delfix has called incidental metadata.

[00:32:14] Speaker 00:
And that is because these references describe using NetApp to store databases, mirror databases, virtualized databases, such as Microsoft SQL Server.

[00:32:24] Speaker 00:
And Delfix's own expert has explained that the database block sizes

[00:32:29] Speaker 00:
can vary in size and so can the NetApp file blocks.

[00:32:33] Speaker 00:
And it was argued below by Actifio that you can have even examples where you have an identical block size.

[00:32:40] Speaker 00:
I think the examples of four kilobytes where you have the file system block and the database block that's being stored there on would be the same size, in which case virtually all of the blocks would have metadata.

[00:32:51] Speaker 00:
But in any event, there is no dispute that at least in some cases there are some blocks in the prior art that contain metadata.

[00:33:00] Speaker 00:
Now, another point that was made earlier was about the database server and how you need to have a database server.

[00:33:06] Speaker 00:
And we would certainly agree that the claims do expressly recite database server, but they do it in a different limitation.

[00:33:12] Speaker 00:
In 808 claim one, it's in the last limitation about mounting the virtual databases on a database server.

[00:33:21] Speaker 00:
And there's been no dispute that that claim was taught by the prior art.

[00:33:25] Speaker 00:
All of these NetApp references talk about

[00:33:28] Speaker 00:
copying, virtualizing, mirroring, et cetera, blocks that include databases which can then be mounted on whether it's Microsoft SQL Server or Oracle or any other program that people may often think of as a database.

[00:33:44] Speaker 00:
But when we talk about database, the spec does have a definition that is very broad.

[00:33:50] Speaker 00:
It says a database is data stored in a computer for use by computer implemented applications.

[00:33:55] Speaker 00:
And although Delphix has tried to walk away from that,

[00:33:58] Speaker 00:
Their own expert, Dr. Chenoy, said that a database is a collection of data that is organized so that it can be easily accessed, managed, or updated, which is exactly what the NetApp waffle file system is.

[00:34:13] Speaker 00:
Now, turning to the decoupling argument or the virtual database term, it seems at the heart of Delphix's argument with the board on this point is that the board did not use this term decoupling.

[00:34:27] Speaker 00:
I think the easiest approach for this report would be to look at what is actually meant by decoupling and whether that would have any impact on the obviousness finding.

[00:34:37] Speaker 00:
And I'd submit that it would not.

[00:34:39] Speaker 00:
The primary reason, in my view, that the board, in my reading of the board's decision, that they rejected that particular language is that it was unclear what decoupling meant.

[00:34:49] Speaker 00:
If you look through the specification, there's not a clear explanation of what you mean by decouple.

[00:34:55] Speaker 00:
And DelFix has offered some meanings in its brief, either

[00:34:57] Speaker 00:
disassociating, or things of that nature.

[00:35:00] Speaker 00:
But at the end of the day, if we look at the expert declarations, the evidence of the prior art, and also the patent, what is really meant by virtual, which is the term in the claim, and possibly by decoupling, is that there's some level of indirection, that you're not dealing with something directly, but through a layer of abstraction.

[00:35:21] Speaker 00:
And Delphix has already admitted, and their expert admitted, that that exists in the NetApp system.

[00:35:27] Speaker 02:
Can I just ask a basic question?

[00:35:29] Speaker 02:
We have to agree with the board on both of those sort of claim construction questions, right, for you to prevail.

[00:35:36] Speaker 02:
I mean, there are two things we're talking about here.

[00:35:38] Speaker 02:
We're talking about database blocks and we're talking about virtual database.

[00:35:42] Speaker 02:
Do we have to agree with the board on one to sustain the conclusion?

[00:35:45] Speaker 02:
For both of them, I'm sorry.

[00:35:46] Speaker 02:
Do we have to agree with the board on both?

[00:35:48] Speaker 02:
That's my understanding.

[00:35:50] Speaker 02:
If you disagree with one of those, there would be a problem there, right?

[00:35:54] Speaker 00:
If you disagree with one or both of them, then you'd have to look to whether there's still an obviousness.

[00:36:02] Speaker 00:
Finding would be supported by the evidence under whatever you would find to be the appropriate construction.

[00:36:08] Speaker 04:
Wouldn't we just send it back?

[00:36:10] Speaker 04:
I mean, if we come up with a different claim construction, isn't it?

[00:36:15] Speaker 04:
It sounds like you're asking us to do new fact-finding if we come up with a new claim construction.

[00:36:23] Speaker 00:
Certainly it's not this court's role to engage in fact-finding in the first instance.

[00:36:29] Speaker 00:
But on the two main terms here, the metadata in the database blocks, whether that's required, the board did note that there is evidence of metadata

[00:36:39] Speaker 00:
being stored in the database block.

[00:36:41] Speaker 00:
So there's that.

[00:36:42] Speaker 02:
And on the virtual database... So you're saying that the board had an alternative holding that even if metadata, even if we conclude that metadata is required, it's also found in the prior art?

[00:36:54] Speaker 02:
It's also in the prior art?

[00:36:55] Speaker 02:
I'm not understanding what you're saying.

[00:36:58] Speaker 02:
That's correct.

[00:37:32] Speaker 00:
I apologize, I don't have it immediately at my fingertips.

[00:37:38] Speaker 00:
But there is, I'm glad to provide the court with references later if requested.

[00:37:46] Speaker 04:
Board did reference- Do you know what prior art they were relying on for that point?

[00:37:52] Speaker 00:
So it is, it's the NetApp prior art and it's specifically the example of where the NetApp blocks could themselves contain blocks

[00:38:02] Speaker 00:
from, for example, a Microsoft SQL server.

[00:38:06] Speaker 04:
I don't understand that sentence.

[00:38:08] Speaker 04:
You'd have to explain it better to me.

[00:38:10] Speaker 04:
So a lot of this... For example, because I've understood you've argued that some of these prior art references, even if they don't contain the metadata in a database block, have it elsewhere.

[00:38:23] Speaker 04:
But I think you're now saying that some of this prior art shows the metadata in a database block.

[00:38:30] Speaker 04:
You have to explain that to me.

[00:38:32] Speaker 04:
If that was what you were trying to do, you didn't get there.

[00:38:36] Speaker 00:
Sure.

[00:38:37] Speaker 00:
So to try to start with the big picture, Delphix has pointed out that NetApp is a file system, which it is, and it has file blocks in which it organizes the data that is ultimately stored to a physical storage device of some sort.

[00:38:53] Speaker 00:
And Delphix has pointed out that you also have blocks that can exist

[00:38:58] Speaker 00:
in what we might traditionally think of as a database, something like Microsoft SQL Server, and that those are logically distinct from the blocks that are being used by the file system.

[00:39:10] Speaker 00:
And what Actifio, the petitioner, argued below is that the references, and I think this is undisputed, that references teach that you could use NetApp to store, replicate, mirror, et cetera, database data, databases in these file blocks.

[00:39:26] Speaker 00:
And the argument was, and the board noted and agreed, that in that context, you could have NetApp blocks that themselves contain or even be coextensive with, for example, a Microsoft SQL Server block and the metadata that's in that SQL.

[00:39:44] Speaker 04:
Are you saying that a Microsoft SQL block has metadata in it?

[00:39:48] Speaker 00:
That is undisputed.

[00:39:49] Speaker 00:
OK.

[00:39:50] Speaker 04:
And so your view is NetApp can

[00:39:53] Speaker 04:
stored those Microsoft blocks in a file server, and therefore that's what covers it?

[00:40:00] Speaker 00:
Almost, Your Honor, is that the NetApp file blocks themselves, it contained the Microsoft SQL Server blocks, for example, which contain the metadata, and therefore there would be the metadata contained in the block.

[00:40:14] Speaker 02:
Can I go back to what we're talking about, which is where we started?

[00:40:19] Speaker 02:
And you noted, and it's fair observation, that you're looking to the claims, and so the claims don't explicitly call out metadata.

[00:40:27] Speaker 02:
But what about claims 32 and 33, which do?

[00:40:30] Speaker 02:
And I understand they're dependent claims, but there's nowhere in.

[00:40:34] Speaker 02:
It seems like they're talking about

[00:40:37] Speaker 02:
when they speak to analyzing the metadata of database blocks, that seems to be almost another definitional point that's consistent with the language and the spec that we've been quibbling about this morning in column two.

[00:40:51] Speaker 02:
Right?

[00:40:52] Speaker 00:
So claims 32 and 33 in my reading, I would disagree that they're definitional.

[00:40:59] Speaker 02:
I think they are a- But they say, identify database blocks and analyzing the metadata of database blocks.

[00:41:08] Speaker 02:
So why is that not almost definitional or assuming there's metadata?

[00:41:16] Speaker 00:
So three points, Your Honor.

[00:41:17] Speaker 00:
The first one would be that they may define what's covered or attempted to be covered by claims 32 and 33.

[00:41:24] Speaker 00:
But secondly, there seems to be an antecedent basis problem because the claims 32 and 33 talk about the metadata.

[00:41:31] Speaker 00:
But there's no clear antecedent basis in the prior claims or the independent claims from which it depends.

[00:41:37] Speaker 00:
for that metadata, which seems like there's a clarity problem.

[00:41:41] Speaker 02:
But thirdly... Well, it seems to suggest that inherently the other claims did limit themselves to metadata.

[00:41:48] Speaker 02:
That's my point.

[00:41:50] Speaker 00:
And if that's the case, my third point would be that these claims don't track back directly to, for example, claim one, but rather through claim 29 and claim 29

[00:42:05] Speaker 00:
is particularly talking about the data stream embodiment.

[00:42:10] Speaker 00:
And as we've discussed already, there are these alternate embodiments where there's the file sharing embodiment as well as the data stream.

[00:42:16] Speaker 02:
And you can see that the data stream embodiment includes metadata.

[00:42:20] Speaker 00:
Yes, we agree that the patent spec describes it that way, although the board did interpret the spec to read that there would be possible ways of

[00:42:31] Speaker 00:
implementing the stream environment without using metadata, but it would agree that the patent spec does talk about the use of metadata in the database blocks for the streaming environment, embodiment.

[00:42:43] Speaker 00:
So going back to the question of what happens if this court disagrees with the claim constructions, we talked about the metadata.

[00:42:52] Speaker 00:
I want to go back to the virtual database term because the virtual database term, if the board were to

[00:43:01] Speaker 00:
adopt this decoupling language, it seems as though, if we look at the evidence, that there would be no difference in what that actually means.

[00:43:12] Speaker 00:
Because the claim, as interpreted by the board, already requires a set of readable and writable database files capable of being mapped to physical addresses for stored database blocks associated with another database.

[00:43:25] Speaker 00:
And this mapping, which is using the pointers,

[00:43:28] Speaker 00:
to point back to where the files are actually stored is what everyone seems to agree is meant by virtualization.

[00:43:35] Speaker 00:
And specifically, Dr. Chenoy, Delphic's expert at Appendix 15967, specifically talks about the NetApp system, the PriorArt system.

[00:43:50] Speaker 00:
And he says, the NetApp waffle file system described below and relied upon in the petition virtualizes aspects of the storage layer

[00:43:58] Speaker 00:
by decoupling storage structures such as volumes and files from the specific implementation of the underlying physical storage media.

[00:44:06] Speaker 00:
That sounds an awful lot like the construction that's being argued by Delphics.

[00:44:11] Speaker 00:
And then if we turn from there to the Edwards reference, which again is really one of the primary references that was relied upon across the seven cases, the seven IPRs, the first page, and this is Appendix 7511 in the 2063 case,

[00:44:30] Speaker 00:
I apologize, 7512, the second page, says by implementing a level of indirection between physical storage containers called aggregates and logical volumes, flexible volumes, we virtualize the allocation of volumes on physical storage, allowing multiple independently managed file volumes along with their snapshot copies to share the same storage.

[00:44:53] Speaker 00:
This virtual database term is

[00:44:57] Speaker 00:
not something that was new, and I don't believe Delphix has ever claimed to have invented virtualization or virtual databases.

[00:45:06] Speaker 00:
The claims require a lot more, and generally what they seem to require is that you have a source database or production database, you copy that to a new location, multiple times, reflecting multiple points in time as that database changed, and then you set up one or more virtual databases that all share that same copied data.

[00:45:27] Speaker 00:
But that's not the virtual database.

[00:45:28] Speaker 00:
The virtual database term viewed in the claim as a whole is just talking about this mapping.

[00:45:35] Speaker 00:
We have the pointers that allow multiple databases to all reference the same underlying storage, which has a lot of efficiencies given the needs of computing resources, storage, limitations, et cetera.

[00:45:49] Speaker 00:
And it seems that all agree that that is here.

[00:45:52] Speaker 00:
So the decoupling term, in our view, is simply

[00:45:56] Speaker 00:
an unclear term, but to the extent it means what Delphix's expert said it means, it seems that that is clearly taught in the art.

[00:46:05] Speaker 00:
Delphix has also argued that there have to be various actions by the database server that are incorporated into the definition of virtual database.

[00:46:17] Speaker 00:
And the board disagreed with that and they explained that the database, that's the files themselves, that's the data.

[00:46:24] Speaker 00:
You have a database system

[00:46:25] Speaker 00:
And that incorporates a server, which is software that operates on that data.

[00:46:30] Speaker 00:
But that's something different.

[00:46:32] Speaker 00:
And it's also recited separately in the claims.

[00:46:35] Speaker 00:
And again, for example, in claim one of the 808, the last limitation talks about mounting the database on a database server.

[00:46:43] Speaker 00:
And that allows the database server to read and write to the files.

[00:46:47] Speaker 00:
And again, there's no dispute that that was met.

[00:46:50] Speaker 02:
Can I bring you back to the metadata issue we were talking about?

[00:46:55] Speaker 02:
Okay, I just want to clarify sort of before your time runs out where we are on that.

[00:47:00] Speaker 02:
Is it, were you saying that even if we were to conclude that the board's construction of data blocks is not requiring metadata was wrong and that we think that Delphic's claim construction is the correct one, that the board made alternative findings that even under Delphic's construction the prior art would read on it?

[00:47:24] Speaker 02:
Is that what you were saying, or did it?

[00:47:27] Speaker 00:
That is correct, Your Honor.

[00:47:29] Speaker 02:
And you were going to try to find us where in the board opinion those alternative findings are.

[00:47:37] Speaker 02:
Because I'm not finding it.

[00:47:40] Speaker 02:
And the board opinion is very wrong.

[00:47:44] Speaker 00:
The opinions are thorough, yes, Your Honor.

[00:47:47] Speaker ?:
Yes.

[00:47:51] Speaker 02:
And I mean, maybe if your friend agrees, then we don't have to quibble about this.

[00:47:56] Speaker 00:
Well, I apologize to delay your honor.

[00:48:04] Speaker 00:
I'll attempt to find that for you now.

[00:48:07] Speaker 02:
I just think it would be very helpful so we could have a good time.

[00:48:20] Speaker 05:
I didn't see that either, counsel.

[00:48:23] Speaker 05:
I'm looking at a portion of the board's opinion that says, Dr. Zadok explains that at least Oracle and IBM databases also include database blocks without metadata, and concludes that the ordinary meaning of the term database block does not include metadata.

[00:48:45] Speaker 05:
Isn't that supportive of your position?

[00:48:54] Speaker 00:
Those were definitely facts found by the board That I think there's a separate related issue I don't want to confuse them, but there's a separate question about whether all databases at the time Necessarily had metadata in the database blocks which went to the point about whether there was some common ordinary understood meaning separate and apart from the patents

[00:49:22] Speaker 00:
And the board did make factual findings that I think merit deference on appeal, that no, that's not the case.

[00:49:29] Speaker 00:
And that's the case not only because of Google, but also because of the evidence, as shown by Dr. Zadek, that there are examples even within, for example, Microsoft SQL Server of database blocks that do not contain metadata.

[00:49:45] Speaker 02:
But that's a completely different point.

[00:49:48] Speaker 00:
That is a different point.

[00:49:50] Speaker 04:
I mean, you didn't make this alternative argument in your brief.

[00:49:53] Speaker 04:
I didn't see anything in the headings that suggested we can affirm on a different basis.

[00:49:59] Speaker 00:
So in our brief, for example, in the 2063 case, the green brief on page 47, this is the brief for 2063, 64, 65, and 69.

[00:50:19] Speaker 00:
And on page 47 of that brief, we do make the argument.

[00:50:26] Speaker 00:
We say, further, Delfix concedes that some of the NetApp database blocks contain some database metadata.

[00:50:33] Speaker 00:
And that some database metadata is a quote from the blue brief on page 43.

[00:50:38] Speaker 00:
And we also quote blue brief 42 through 43, talking about no record evidence shows that the waffle data blocks necessarily include database metadata.

[00:50:48] Speaker 00:
and in the reply brief.

[00:50:50] Speaker 04:
Can you give me your page number again?

[00:50:52] Speaker 04:
Sorry.

[00:50:53] Speaker 00:
Page 47.

[00:50:53] Speaker 00:
Okay.

[00:50:54] Speaker 02:
You say it's sufficient for obvious purposes that some of the NetApp database blocks include some metadata, even under their construction.

[00:51:06] Speaker 02:
I'm not seeing that you cited to the boards.

[00:51:09] Speaker 00:
I don't.

[00:51:10] Speaker 00:
And if I have misstated that, then I sincerely apologize.

[00:51:14] Speaker 00:
I do not have reference here.

[00:51:16] Speaker 00:
I believe

[00:51:18] Speaker 00:
that in at least one of the board decisions that there is reference to that.

[00:51:24] Speaker 00:
But in any event, there is concession from Delfix that it is the case that, at least in some circumstances, there will be blocks that contain metadata.

[00:51:36] Speaker 00:
And in the reply brief, there's also reference to this on page 24 in the reply brief in the 2063 case talking about

[00:51:44] Speaker 00:
the fact that NetApp's file system blocks may incidentally contain some database metadata.

[00:51:49] Speaker 00:
And that goes on to make their arguments why they don't think that metadata corresponds with the file blocks.

[00:51:56] Speaker 05:
Why is the patent office making an appearance here?

[00:51:59] Speaker 05:
Why are you here when the petitioner didn't bother to show up?

[00:52:08] Speaker 05:
And you're not arguing any matter of procedure

[00:52:12] Speaker 05:
the board ought to be especially interested in.

[00:52:15] Speaker 05:
This is a nuts and bolts argument, so why are you here?

[00:52:20] Speaker 00:
So we are here as we are statutorily authorized to do to defend the determination of the board.

[00:52:28] Speaker 05:
Entitled to, but not required to.

[00:52:32] Speaker 00:
Not required to, but in the interest of defending what we believe to have been a correct determination by the board regarding the patentability

[00:52:41] Speaker 00:
of these three particular patents, we are here given the fact that Actifio has stepped aside and was not participating in the appeal.

[00:52:50] Speaker 02:
But you don't do it in all cases.

[00:52:52] Speaker 02:
Somebody just decides which cases you should come in for.

[00:52:55] Speaker 02:
I mean, I had an appeal last month on the calendar, and it was one sided because the petitioner dropped out, and the PTO decided not to come in and defend it.

[00:53:08] Speaker 00:
Well, I'm aware that we have not intervened in every appeal.

[00:53:12] Speaker 00:
And as Judge Loria used to point out, the statute doesn't require us to intervene in every appeal.

[00:53:18] Speaker 00:
And that particular decision is certainly above my pay grade.

[00:53:23] Speaker 04:
Well, just for myself, I appreciate the fact that we have briefing on both sides here.

[00:53:28] Speaker 04:
So I mean, don't take, and I'll leave it at that.

[00:53:35] Speaker 04:
You've run out of time, and I know you had a lot.

[00:53:38] Speaker 04:
Can you just give me two sentences on why that claim 36 construction is right?

[00:53:42] Speaker 04:
Because it doesn't seem like that's right to me.

[00:53:45] Speaker 00:
So on claim 36, the reason it's right, and I admit that on first blush it seems counterintuitive.

[00:53:56] Speaker 00:
What the board did is say, regardless of what the patent owner may have intended to claim here, if we're reading the claims, it doesn't say storing

[00:54:05] Speaker 00:
compressed data blocks.

[00:54:06] Speaker 00:
It says compressing prior to storing.

[00:54:09] Speaker 00:
But if I just may make a very brief point, and that is that the board, before doing that, observed that it would have been obvious in any event to combine compression with the other methods and systems disclosed in the NetApp references.

[00:54:23] Speaker 00:
And then it went on to cite the Singh reference and that example as one example of compression that did meet the literal terms of the claim.

[00:54:33] Speaker 00:
Thank you.

[00:54:34] Speaker 00:
Thank you.

[00:54:43] Speaker 03:
Just to respond to Judge Post's question about whether if the PTAB is wrong on either the database block instruction or the virtual database instruction, does the court have to overturn the decision?

[00:55:00] Speaker 03:
And the answer is yes.

[00:55:02] Speaker 03:
And there is no alternative finding by the PTAB that these patents would be invalid under either of the constructions proposed by Delphi.

[00:55:13] Speaker 02:
But the Patent Office did suggest in their brief, as your friend pointed out, on page four.

[00:55:17] Speaker 02:
There is a suggestion in the brief.

[00:55:19] Speaker 02:
That's not something that... So that would be a remand, not a reversal, right?

[00:55:22] Speaker 02:
I mean, if we disagreed with the claim construction,

[00:55:25] Speaker 02:
It wouldn't reverse necessarily.

[00:55:29] Speaker 02:
But if there's might be an alternative grounds to hold it to affirming, we could either do it here.

[00:55:34] Speaker 02:
But you might have to.

[00:55:35] Speaker 03:
Yeah, though I don't believe there is an alternative grounds.

[00:55:37] Speaker 03:
Certainly there was not before the PTAB.

[00:55:39] Speaker 03:
So the argument that the director is making today, they make two arguments.

[00:55:46] Speaker 03:
On the database block argument, that somehow a file system block could happen to have some of the metadata from a database block.

[00:55:55] Speaker 03:
That doesn't work.

[00:55:57] Speaker 03:
And it doesn't work for two reasons.

[00:55:59] Speaker 03:
One, what's stored in the file system block is just bits, right?

[00:56:04] Speaker 03:
To be metadata, it has to be understood as being something different than data.

[00:56:09] Speaker 03:
The point of the file system in the NetApp block, it just stores bits.

[00:56:13] Speaker 03:
It doesn't know what's metadata, it doesn't know what's data.

[00:56:16] Speaker 03:
So it's not metadata when it's existing in a file system.

[00:56:20] Speaker 03:
And no one has ever suggested that it is.

[00:56:23] Speaker 03:
The other requirement is that even under the PTAB's kind of partial construction, the database block has to be used by the database.

[00:56:32] Speaker 03:
And once you undo the false equivalence of file system with the database, that fails.

[00:56:40] Speaker 03:
The undisputed evidence is that NetApp blocks are not used by databases.

[00:56:45] Speaker 03:
They're not exposed to the database.

[00:56:48] Speaker 03:
Database doesn't know they exist, doesn't know what structure they have,

[00:56:51] Speaker 03:
doesn't know what size they are, knows nothing about them.

[00:56:54] Speaker 03:
So they cannot be the required database blocks.

[00:56:58] Speaker 03:
On the virtual database construction, the decoupling matters.

[00:57:04] Speaker 03:
And it matters because, as I pointed out, NetApp operates at the disk physical layer.

[00:57:10] Speaker 03:
It cannot decouple that representation of the file when it uses a snap mirror process.

[00:57:17] Speaker 03:
That's also why it can't mirror the data and store it in a compressed form.

[00:57:21] Speaker 03:
is just copying block from block from disk to disk.

[00:57:26] Speaker 03:
And this notion that he uses pointers and somehow that's the same virtualization that creates a virtual database in the patent is just wrong, right?

[00:57:35] Speaker 03:
What the director quotes in its brief about the NetApp, this is from their brief at 10, they said, the NetApp system uses inodes containing pointers which point to the physical location of the associated data blocks

[00:57:51] Speaker 03:
thereby virtualizing the file system.

[00:57:53] Speaker 03:
That is just technically wrong, right?

[00:57:56] Speaker 03:
High nodes pointing to data blocks is, and this is in the NetApp references, that is what they copied from Unix.

[00:58:04] Speaker 03:
That's been around for 40 years.

[00:58:06] Speaker 03:
That's how all file systems find data that's stored on disk.

[00:58:10] Speaker 03:
It has nothing to do with virtualizing the databases in this case.

[00:58:15] Speaker 03:
So there's no,

[00:58:18] Speaker 03:
alternative support for a finding of invalidity under either of the claim constructions proposed by Delfix here.

[00:58:28] Speaker 03:
And again, claim 36, I don't want to beat a dead horse, but they can't make an obvious argument because the priority itself would prohibit you from storing the data in compressed form.

[00:58:43] Speaker 03:
That snap mirror won't do it.

[00:58:45] Speaker 03:
It has to be exactly the data that was on the source.

[00:58:48] Speaker 03:
And so that's why the obviousness argument doesn't even hold up.

[00:58:53] Speaker 03:
One other point I just want to go back to.

[00:58:58] Speaker 03:
Council said that there was evidence in this Google big table thing that there were data blocks and there was separate metadata.

[00:59:07] Speaker 03:
There's no evidence of any data blocks in the big table.

[00:59:11] Speaker 03:
Dr. Zadok, TIFIA's expert, raised this prior for the first

[00:59:16] Speaker 03:
I don't think it's not even a priority.

[00:59:17] Speaker 03:
It raises document for the first time in its reply.

[00:59:21] Speaker 03:
It didn't map anything to database blocks or data blocks in that system.

[00:59:25] Speaker 03:
And so that's your wrong.

[00:59:29] Speaker 03:
And as far as there being other metadata in files, certainly there's metadata in files within a database that describe the file itself.

[00:59:40] Speaker 03:
But there is separate metadata in the database block as described in the patent.

[00:59:45] Speaker 03:
and all of the documentation.

[00:59:47] Speaker 03:
And as the court has noted, required by claims 32 and 33 explicitly.

[00:59:53] Speaker 03:
Talk about the metadata in the block.

[00:59:56] Speaker 03:
If there's nothing, no further questions, I will leave it at that.

[01:00:01] Speaker 03:
Thank you very much.