Mangosoft v . Oracle Corp. CV-02-545-SM 09/21/04 UNITED STATES DISTRICT COURT

DISTRICT OF NEW HAMPSHIRE

Mangosoft, Inc. and Mangosoft Corporation, Plaintiffs

v. Civil N o . 02-545-SM Opinion N o .

Oracle Corporation, Defendant

O R D E R

This is a suit for patent infringement. Plaintiffs,

Mangosoft, Inc. and Mangosoft Corporation (collectively,

“Mangosoft”), say defendant, Oracle Corporation,

is making, selling, and/or offering for sale computer software

that infringes two of Mangosoft’s patents: United States Patent

N o . 6,148,377 (“the ‘377 patent”) and United States Patent N o .

5,918,229 (“the ‘229 patent”). The court held a Markman hearing,

at which the parties presented evidence and argument in support

of their respective constructions of various terms used in the

patents’ claims. See generally Markman v . Westview Instruments,

Inc.,

(1996). Subsequently, the parties filed post-

hearing memoranda. Legal Standard Governing Claim Construction

Patent infringement analysis involves two steps: first,

properly construing the asserted claim; and second, determining

whether the accused method or device infringes the asserted claim

as properly construed. See Vitronics Corp. v . Conceptronic,

Inc.,

(Fed. Cir. 1996) (citing Markman v .

Westview Instruments, Inc.,

6 7 , 979 (Fed. Cir. 1995),

aff’d,

(1996)). Step one of that process - claim

construction - presents a question of law to be resolved by the

court. See Pitney Bowes, Inc. v . Hewlett-Packard Co.,

(Fed. Cir. 1999). The second step - determining

whether the accused process or device infringes the patent -

presents a question of fact.

At this stage of the

litigation, the court is focused exclusively on the first step:

properly construing the meaning and scope of various claim terms

used in the ‘377 and ‘229 patents.

Construing patent claim terms generally means ascertaining

the meaning of those terms in light of the intrinsic evidence of

record, which includes: the claims, the specification, and the

prosecution history. See Vitronics,

. But, the

2 court may consider extrinsic evidence as well. See Apex Inc. v .

Raritan Computer, Inc.,

(Fed. Cir. 2003)

(“Courts may also review extrinsic evidence to assist them in

comprehending the technology in accordance with the understanding

of skilled artisans and as necessary for actual claim

construction.”). Extrinsic evidence is external to the patent,

“such as expert testimony, inventor testimony, dictionaries, and

technical treatises and articles.” Pitney-Bowes,

(citing Vitronics,

). See generally Ferguson

Beauregard/Logic Controls v . Mega Systems LLC,

(Fed. Cir. 2003) (“The ordinary and customary meaning of a

claim term may be determined by reviewing a variety of sources.

Some of these sources include the claims themselves, dictionaries

and treatises, and the written description, the drawings, and the

prosecution history.”) (citations omitted).

The court observed in Vitronics that, “In most situations,

an analysis of the intrinsic evidence alone will resolve any

ambiguity in a disputed claim term. In such circumstances, it is

improper to rely on extrinsic evidence.” Id., at 1583.

3 Nevertheless, even when patent language is unambiguous, a court

may still consider extrinsic evidence for limited purposes.

Vitronics does not prohibit courts from examining extrinsic evidence, even when the patent document is itself clear. Moreover, Vitronics does not set forth any rules regarding the admissibility of expert testimony into evidence. Certainly, there are no prohibitions in Vitronics on courts hearing evidence from experts. Rather, Vitronics merely warned courts not to rely on extrinsic evidence in claim construction to contradict the meaning of claims discernible from thoughtful examination of the claims, the written description, and the prosecution history - the intrinsic evidence.

Pitney Bowes,

(citations omitted) (emphasis

supplied). Consequently, the court concluded:

Thus, under Vitronics, it is entirely appropriate, perhaps even preferable, for a court to consult trustworthy extrinsic evidence to ensure that the claim construction it is tending to from the patent file is not inconsistent with clearly expressed, plainly apposite, and widely held understandings in the pertinent technical field. This is especially the case with respect to technical terms, . . . . Indeed, a patent is both a technical and a legal document. While a judge is well-equipped to interpret the legal aspects of the document, he or she must also interpret the technical aspects of the document, and indeed its overall meaning, from the vantage point of one skilled in the art. Although the patent file may often be sufficient to permit the judge to interpret the technical aspects of the patent properly, consultation of extrinsic evidence is particularly appropriate to ensure that his or her understanding of the technical

4 aspects of the patent is not entirely at variance with the understanding of one skilled in the art.

(citation omitted) (emphasis supplied). See also Key

Pharms. v . Hercon Lab. Corp.,

(Fed. Cir. 1998).

Giving proper effect to disputed technical terms in a patent

requires a court to construe them as they would be construed by

those skilled in the relevant art. See Hoechst Celanese Corp. v .

BP Chems. Ltd.,

(Fed. Cir. 1996) (“A technical

term used in a patent document is interpreted as having the

meaning that it would be given by persons experienced in the

field of the invention, unless it is apparent from the patent and

the prosecution history that the inventor used the term with a

different meaning.”). See also Ferguson,

(“In

the absence of an express intent to impart a novel meaning to the

claim terms, the words take on the full breadth of the ordinary

and customary meanings attributed to them by those of ordinary

skill in the art.”). Here, nothing suggests that the terms in

dispute are used in the patents in any way other than as they

would be commonly understood by those skilled in the relevant

art. With respect to the patents at issue here, a person skilled

5 in the relevant art would hold a bachelor’s degree in computer

science or electrical engineering and have two to three years

experience in the field of distributed computing.

In this case, the court has, with appreciation, relied upon

the capable expert testimony presented by the parties to obtain a

general understanding of the computer system described in the

‘377 patent, as well as the method for providing “distributed

control over a structured store of data” described in the ‘229

patent.

Discussion

I. The Patents at Issue.

Generally speaking, the patents, taken together, teach a

“distributed shared memory system.” They describe computer

systems consisting of groups of computers linked by a network

connection, also known as a “cluster” or “computer cluster.”

Each computer, or “node,” in the cluster manages its own memory

(both volatile and non-volatile) and, employing the inventions,

makes that memory available to other nodes in the cluster. And,

says Mangosoft, unlike earlier systems, which only provided a

6 means for sharing data stored in non-volatile memory (e.g., hard

disks), these inventions allow nodes to also share volatile

memory (e.g., random access memory or “RAM”) with other nodes in

the cluster. Thus, the inventions teach a means by which nodes

may share both non-volatile and volatile memory space, by

creating a “pool” of shared memory space which is accessible by

all nodes participating in the system.

The ‘377 patent teaches a computer system that allows

multiple computers to share volatile and non-volatile memory

resources. Claim 1 , the sole independent claim of the patent,

teaches:

A computer system having a shared addressable memory space, comprising

a data network for carrying signals representative of computer readable information, and

a plurality of computers, each of said plurality of computers sharing the shared addressable memory space and including

an interface, coupled to said data network, for accessing said data network to exchange data signals therewith,

a local volatile memory device coupled to said computer and having volatile storage for data signals,

7 a local persistent memory device coupled to said computer and having persistent storage for data signals, and

a shared memory subsystem for mapping a portion of said shared addressable memory space to a portion or the whole of said persistent storage and said volatile storage to provide thereby addressable persistent and volatile storage for data signals accessible by each of the plural computers, said shared memory subsystem including

a distributor for mapping portions of said addressable memory space across said plurality of local persistent memory devices, to distribute said addressable memory space across said plurality of local persistent memory devices, and

a disk directory manager for tracking said mapped portions of said addressable memory space to provide information representative of which of said local persistent memory devices has which of said portions of said addressable memory space mapped thereon.

The ‘377 patent, 15:56-67 - 16:1-23 (emphasis supplied).

The ‘229 patent is a continuation-in-part of the application

that issued as the ‘377 patent, and it incorporates the ‘377

patent by reference. See ‘229 patent at 1:4-10, 14:58-65.

Drawing on the ‘377 patent’s shared memory system for providing

8 an addressable shared memory space across the physical memory

devices of multiple computers on a network, the ‘229 patent

teaches methods for providing distributed control and persistent

storage for a “structured store of data,” including database

records and web pages.

The ‘229 patent contains 5 independent claims (claims 1 , 3 0 ,

3 1 , 3 4 , and 35) and 32 dependent claims. All terms disputed by

the parties are included in independent claim 1 , which teaches:

A method for providing distributed control over a structured store of data, comprising:

providing a plurality of nodes inter-connected by a network, each of the plurality of nodes sharing a shared addressable memory space of a shared memory system and including (i) an interface for accessing the network, (ii) a local volatile memory device coupled to the node and providing volatile storage, (iii) a local persistent memory device coupled to the node and providing persistent storage, and (iv) a shared memory subsystem for mapping a portion of the shared addressable memory space to at least a portion of the persistent and volatile storage to provide thereby addressable persistent and volatile storage accessible by each of the plurality of nodes, the shared memory subsystem including (a) a distributor for mapping portions of the addressable memory space across the plurality of local persistent and volatile memory devices to distribute the addressable memory space across the plurality of local persistent and volatile memory

9 devices, and (b) a disk directory manager for tracking the mapped portions of the addressable memory space to provide information representative of which of the local persistent and volatile memory devices has which of the portions of the addressable memory space mapped thereon;

storing on each node an instance of a data control program for manipulating the structured store of data to provide multiple, distributed instances of the data control program;

interfacing each instance of the data control program to the shared memory subsystem; and

operating each instance of the data control program to employ the shared memory system as a memory device having the structured store of data contained therein, whereby the shared memory system coordinates access to the structured store of data to provide distributed control over the structured store of data.

‘229 patent, 28:21-59 (emphasis supplied).

The parties disagree as to the meaning of the following

phrases, as used in the ‘377 and ‘229 patents: (1) “shared

addressable memory space”; (2) “local memory device”; (3) “shared

memory subsystem”; (4) “structured store of data”; and (5) “each

of said plurality of computers.”

10 II. Claim Construction.

A. “Each of Said Plurality of Computers”

The parties agree that the word “plural” means two or more.

They disagree, however, as to what is meant by the word “each,”

when used to modify the phrase “of said plurality of computers.”

As noted above, the ‘377 patent describes a “computer system

having a shared addressable memory space, comprising . . . a

plurality of computers, each of said plurality of computers

sharing the shared addressable memory space.” ‘377 patent,

15:58-63 (emphasis supplied).

Oracle asserts that the quoted language requires all of the

computers participating in the shared memory system to share the

shared addressable memory space. Oracle’s position rests upon

its interpretation of the words which describe a “plurality of

computers” participating in a shared memory system, and require

that “each of said plurality” share the shared addressable memory

space. In other words, according to Oracle, if the “plurality”

of computers participating in the system is five, then all five

of those computers (i.e., each computer making up said plurality)

must share the shared addressable memory space.

11 Mangosoft, on the other hand, says the phrase “each of said

plurality” requires only that two or more of the computers in the

overall system actually share the shared addressable space. That

i s , Mangosoft asserts that the phrase “each of said plurality”

does not modify the “plurality of computers” that form the

system, but instead refers to a smaller subset of computers

(i.e., a new plurality). S o , according to Mangosoft, if the

plurality of computers participating in the shared memory system

is five, only two of those computers need actually share the

shared addressable memory space. The court disagrees.

Mangosoft’s interpretation of the disputed claim language is

not supported by the precedent upon which it relies, nor is it

consistent with customary uses of the words “each” and “said.”

Had Mangosoft intended the interpretation it advances here, it

likely would have used language such as “comprising . . . a

plurality of computers, some of which share the shared

addressable memory space.” Alternatively, it might have said “a

plurality of which” or “two or more of which” or “a subset of

which” share the memory space. It did not. Instead, the ‘377

patent uses very specific language, which must be given meaning.

12 Mangosoft’s interpretation of the disputed language does not give

meaning to the words “said” or “each” and, instead, “render[s]

the contested terms surplusage.” Wright Med. Tech., Inc. v .

Osteonics Corp.,

(Fed. Cir. 1997). See also

Lantech, Inc. v . Keip Mach. Co.,

4 2 , 546 (Fed. Cir.

1994) (“All limitations in a claim must be considered

meaningful.”).

As used in the ‘377 patent, the phrase “said plurality”

refers to the plurality of computers that form the system. And,

the word “each” refers to that same plurality of computers; it

does not refer to a new, smaller subset of the original plurality

of computers. Accordingly, the court construes the disputed

language - “A computer system having a shared addressable memory

space, comprising . . . a plurality of computers, each of said

plurality of computers sharing the shared addressable memory

space” - to mean: a computer system having a shared addressable

memory space, comprising two or more computers, every one of

which of those two or more computers participating in the system

has access t o , and may contribute t o , the shared addressable

memory space.

13 The court, however, agrees with Mangosoft to the extent it

says that not all computers on a particular network must

necessarily participate in the system described in the ‘377

patent. In other words, the ‘377 patent teaches a system in

which fewer than all computers on a network may participate in

the described shared addressable memory system. See, e.g.,

“Summary of the Invention,” ‘377 patent at 2:22-29 (“The

invention provides systems that can create and manage a virtual

memory space that can be shared by each computer on a network and

can span the storage space of each memory device connected to the

network.”) (emphasis supplied). See also “Abstract,” ‘377 patent

at page 1 (“Distributed shared memory systems and processes . . .

that optionally spans across each memory device connected to the

computer network. Accordingly, each node on the network having

the distributed shared memory system of the invention can access

the shared memory.”) (emphasis supplied).

B. “Shared Addressable Memory Space”

Here, the parties’ dispute appears to focus on whether the

“shared addressable memory space” must be addressed by a common

14 addressing scheme. Oracle asserts that it must, while Mangosoft

disagrees. Specifically, Mangosoft says:

[T]he shared addressable memory space is shared by the plural computers. That method of sharing and addressing, however, need not be through the use of common addresses. Indeed, it is the shared memory subsystem, . . . that allows the different computers to address the shared addressable memory space without the requirement of common addresses. As [Mangosoft’s expert] explained, the shared memory subsystem provides the necessary translation to allow the various nodes to address the shared addressable memory space, even if they don’t utilize common addresses, or by the analogy, a common language (e.g., some speak English, some French, some Norwegian). While Oracle argued that the nodes all must speak English (or use common addresses), that argument ignores entirely the role of the shared memory subsystem in providing the necessary translation. Moreover, it ignores the claim language that does not contain the “common addresses” limitation, the Federal Circuit law that prohibits reading into the claim limitations from the preferred embodiment, and the doctrine of claim differentiation.

Mangosoft’s Supplemental Brief (document n o . 46) at 1 1 .

Claim 1 of the ‘377 patent does not specifically require

that the nodes participating in the shared memory system utilize

a “common” addressing scheme. Instead, claim 1 simply provides

that the memory subsystem component of each participating node

tracks the data and available memory space in the system. See,

15 e.g., ‘377 patent 7:27-37 (“The memory subsystems further track

the data stored in the local memory of each node and further

operate network connections with network 38 for transferring data

between the nodes 12a-12c. In this way, the memory subsystems

32a-32c access and control each memory element on the network 38

to perform memory access operations that are transparent to the

operating system 1 6 . Accordingly, the operating system 16

interfaces with the memory subsystem 32 as an interface to a

global memory space that spans each node 12a-12c on the network

38.”). It would, then, appear that the “memory subsystem”

functions in a manner similar to a traditional network server,

translating the various addressing schemes utilized by each node

on the system.

In other words, the memory subsystem is capable of

translating the various “languages” spoken by each participating

node; they need not all speak a common language. And, contrary

to Oracle’s suggestion, claim 1 of the ‘377 patent does not

require the use of either “common memory addresses” or “global

memory addresses;” those requirements appear only in particular

embodiments of the patented system. The court will not read

16 those embodiments into the patent as limitations. See, e.g.,

Electro Med. Sys., S.A. v . Cooper Life Sciences, Inc.,

(Fed. Cir. 1994) (“[A]lthough the specifications may

well indicate that certain embodiments are preferred, particular

embodiments appearing in a specification will not be read into

the claims when the claim language is broader than such

embodiments.”).

Accordingly, the court construes the phrase “shared

addressable memory space” to mean memory space distributed across

the volatile and non-volatile memory of all nodes participating

in the patents’ shared memory system (though not necessarily all

nodes on the network), which shared memory space can be accessed

by the various participating nodes using one or more addresses.

The participating nodes need not, however, utilize a common or

global addressing scheme. As described in the embodiment

depicted in Figure 3 of the ‘377 patent, the use of a “global

address generator” is optional. ‘377 patent at 8:41-47.

17 C. “Local Memory Device”

The parties disagree as to whether a local memory device

must be “attached” to only a single computer (Oracle’s view), or

whether it need only be “accessible” by a computer without having

to go through another node or computer controlling access to that

device (Mangosoft’s view). Neither party has, however, explained

precisely how this dispute i s , for purposes of Mangosoft’s

infringement claims, a meaningful one. Nevertheless, because the

parties seek an interpretation of that term, the court will

provide one.

The court concludes that, at the time the ‘377 patent

issued, one of ordinary skill in the relevant art would

understand that the word “local” is used to describe computer

devices that are directly attached to a single computer’s

processor (by, for example, the computer’s b u s ) , without the need

for an intervening communication channel. Thus, “local” devices

are distinct from “shared” or “networked” or “remote” devices

which, by virtue of some intervening communication channel, might

be accessed by more than one computer (though a “local” device

18 might also be “shared” with other nodes, through such an

intervening communication channel).

Nevertheless, Mangosoft asserts that the use of the word

“coupled” in the claim language (i.e., “local . . . memory device

coupled to said computer”) suggests that the device need not be

“directly attached” to the computer but may, instead, communicate

with the computer’s processor in a more “indirect” manner. See

‘377 patent at 16:1-4. See also ‘229 patent at 28:29-32. The

court is not persuaded. The use of the word “coupled” simply

makes clear that the local memory device in question must be

“local” to the particular computer being described. It does not

serve to modify or redefine the commonly understood meaning of

the word “local,” as used at the time the patents issued.

Were the court to construe the disputed claim language as

Mangosoft suggests, that construction would ignore the word

“local.” In other words, by simply requiring a computer memory

device that is somehow “linked” to a computer (whether directly

or indirectly), Mangosoft’s construction would recast the claim

so that it merely requires “a memory device coupled to a

19 computer.” Importantly, however, the claim language requires a

local memory device that is coupled to a computer. To avoid

rendering the word “local” entirely superfluous (or, at best,

redundant), it must be given a meaning other than “coupled.”

Accordingly, as used in the ‘377 patent (as well as the ‘229

patent), the word “local” when used to modify a computer device

means a computer device (e.g., a hard drive) that is directly

attached to a single computer’s processor by, for example, the

computer’s bus (though it may, of course, be accessed by other

computers through any number of the interconnection technologies

discussed in the exhibits to the declaration of David Klausner

(submitted with Mangosoft’s Opposition Brief (document n o . 4 3 ) ) .

That is to say, a computer memory device that is “local” to one

computer may also be shared with, or accessed by, other computers

on the network (or, of course, other computers participating in

the shared memory system).

D. “Shared Memory Subsystem”

Both the ‘377 and ‘229 patents describe the use of a “shared

memory subsystem for mapping a portion of said shared addressable

20 memory space to a portion or the whole of said persistent storage

and said volatile storage . . ..” ‘377 patent at 16:6-9. See

also ‘229 patent at 28:31-34 (“a shared memory subsystem for

mapping a portion of the shared addressable memory space to at

least a portion of the persistent and volatile storage . . . . ” ) .

Oracle asserts that the “shared memory subsystem” described in

the patents is a means-plus-function claim. The court disagrees.

First, the absence of the word “means” undermines Oracle’s

claim. See, e.g., Al-Site Corp. v . VSI Int’l, Inc.,

(Fed. Cir. 1999) (“[W]hen an element of a claim does

not use the term ‘means,’ treatment as a means-plus-function

claim element is generally not appropriate.”). Second, the claim

element “shared memory subsystem” recites sufficiently definite

structure to dispel the argument that it is a means-plus-function

claim. Third, the phrase “shared memory subsystem” appeared in

both patents and technical literature at the time the ‘377 patent

issued. See, e.g., Exhibit J to Mangosoft’s Opposition Brief,

United States Patent n o . 5,341,475 (describing a communication

system for exchanging data, which employs a shared memory

subsystem). See also Mangosoft’s Opposition Brief at 15-16

21 (representing that the phrase “shared memory subsystem” appeared

in at least nine U.S. Patents filed before the ‘377 patent, as

well as in academic literature).

With regard to the word “mapping” - a function performed by

the shared memory subsystem - Oracle asserts that it means

“assigning a correspondence between the addresses of portions of

virtual memory to the addresses of portions of physical memory.”

Oracle’s Claim Construction Brief (document n o . 36) at 2 3 .

Mangosoft, on the other hand, says that Oracle’s construction of

the word is too narrow in two respects. First, it says the word

“mapping” means simply “defining or establishing a relationship,”

rather than assigning a direct correspondence. Mangosoft’s Claim

Construction Brief at 1 6 . Next, it says that the plain language

of the claim provides only that a portion of the shared

addressable memory space must be mapped to a portion or the whole

of the system’s persistent and/or volatile storage. It does not,

says Mangosoft, require the mapping of portions of the system’s

virtual memory to portions of the system’s physical memory.

22 The word “mapping,” as used in the patents at issue, would

be understood by one of ordinary skill in the art to mean

“creating an association between.” See, e.g., Oracle’s Claim

Construction Brief at 23 (citing IEEE Standard Dict. of Elec. and

Electronics Terms (6th ed. 1996) at 627 for the proposition that

to “map a range of addresses” means “to create an association

between a range of a process’s address space and a range of

physical memory or some memory object, such that a reference to

an address in that range of the address space results in a

reference to the associated physical memory or memory object.”).

Accordingly, the memory subsystem described in the ‘377 patent

and the ‘229 patent creates an association or relationship

between the shared addressable memory space and some or all of

the local persistent and local volatile memory space of the

participating nodes.

E. “Structured Store of Data”

Finally, the parties disagree as to the meaning of the

phrase “structured store of data,” as used in the ‘229 patent.

See, e.g., ‘229 patent at 28:24. Specifically, they disagree as

to whether the referenced data must reside exclusively in

23 persistent storage. Oracle asserts that it must, while Mangosoft

says storage of the structured data need not be limited to

persistent (i.e., non-volatile) devices.

The claims of the ‘229 patent do not make clear whether the

referenced structured store of data may be (or must be) located

in any particular form of memory. In support of its view that

such data must reside exclusively in non-volatile storage, Oracle

points out that the specification repeatedly describes the

structured store of data as being “persistent data.” See, e.g.,

‘229 patent at 3:19-22 (“The invention can be understood as

structured storage systems, and related methods, that employ a

globally addressable unstructured memory system to maintain a

structured store of persistent data within a shared memory

space.”). Consequently, says Oracle, if the data themselves are

“persistent,” they must necessarily reside on a persistent (non-

volatile) storage medium. And, according to Oracle, by

repeatedly stating throughout the specification that the data are

“persistent,” the patentees have (albeit implicitly) defined the

phrase “structured store of data” to mean data which are stored

on a persistent memory device. See, e.g., Bell Atl. Network

24 Servs., Inc. v . Covad Communs. Group, Inc.,

, 1277

(Fed. Cir. 2001) (“[T]he written description provides guidance as

to the meaning of the claims, thereby dictating the manner in

which the claims are to be construed, even if the guidance is not

provided in explicit definitional format. Because the patentees

used the [disputed] term throughout the entire patent

specification, consistent with a single meaning, they defined

that term by implication.”) (citations and internal punctuation

omitted). See also SciMed Life Sys., Inc. v . Advanced

Cardiovascular Sys., Inc.,

(Fed. Cir. 2001)

(“While it is true, of course, that the claims define the scope

of the right to exclude and that the claim construction inquiry,

therefore, begins and ends in all cases with the actual words of

the claim, the written description can provide guidance as to the

meaning of the claims, thereby dictating the manner in which the

claims are to be construed, even if the guidance is not provided

in explicit definitional format.”) (citations and internal

punctuation omitted).

The court is not persuaded that the “structured store of

data” as referenced in the ‘229 patent’s claims must, as a matter

25 of construction or of necessity, reside in persistent storage.

Claim 1 of the ‘229 patent teaches a “method for providing

distributed control over a structured store of data” among all

nodes of a cluster participating in the addressable shared memory

space system. That system, as described above, includes the use

of (and storage of data on) both persistent and volatile memory

devices, which are shared and accessible by all participating

nodes. See ‘229 patent at 28:32-37. Nothing in the claim

language suggests that, unlike other forms of data (which may be

stored in persistent memory, volatile memory, or both), the

“structured store of data” is somehow unique and must, instead,

reside exclusively in persistent storage. In fact, the entire

purpose of the ‘229 patent would be defeated if such data were

restricted exclusively to persistent storage.

The references in the specification to “persistent data”

(rather than “persistent storage” or “persistent memory”) simply

suggest that the system is designed, and the data organized, in

such a way that should one or even all of the participating nodes

lose data stored in volatile memory, the “structured store of

data” is still “persistent” and can be recovered (by, presumably,

26 collecting and compiling the various redundant portions of it

that were distributed to the non-volatile memory of the

participating nodes). In other words, the “persistent data” are

stored in the various volatile and non-volatile memory devices

attached to participating nodes in such a way that even if all

data are lost from volatile memory, they can be reconstructed

from the various portions that were stored on non-volatile memory

devices (there i s , of course, no requirement that the data not be

duplicated - in fact, it is possible that a particular piece of

datum could be stored in multiple locations: in the RAM of one

node, on the hard drive of another node, and again in the RAM of

a third node). In short, the court is not persuaded by Oracle’s

assertion that by referencing the structured store of data as

“persistent data,” the patentees have limited the scope of the

patent’s claims to require that such data reside exclusively on

non-volatile forms of storage media.

Accordingly, the court construes the phrase “structured

store of data” to mean data that are organized in some recognized

fashion (e.g., database files, word processing document files, or

27 Web pages) and stored in the volatile and/or non-volatile memory

of the various nodes participating in the shared memory system.

F. Summary

Having carefully reviewed both the ‘377 and ‘229 patents, as

well as the expert opinion and argument presented by the parties,

the court construes the disputed terms and phrases in the patents

as follows:

1. “A computer system having a shared addressable memory space, comprising . . . a plurality of computers, each of said plurality of computers sharing the shared addressable memory space” means a computer system having a shared addressable memory space, comprising two or more computers, every one of which of those two or more computers participating in the system (though not necessarily every node on the network) has access t o , and may contribute t o , the shared addressable memory space.

2. “Shared addressable memory space” means a memory space distributed across the volatile and non-volatile memory of all nodes participating in the patents’ shared memory system (though not necessarily all nodes on the network), which shared memory space can be accessed by the various participating nodes using one or more addresses. The participating nodes need not, however, utilize a common or global addressing scheme.

28 3. “Local,” when used to modify a computer device, means a computer device (e.g., a hard drive) that is directly attached to a single computer’s processor by, for example, the computer’s bus. Such a “local” device may, however, be shared with and accessible by other nodes on the network (and, of course, other nodes participating in the shared memory system).

4. “Mapping” means creating an association or relationship between the shared addressable memory space and some or all of the local persistent and local volatile memory space of the participating nodes.

5. “Structured store of data” means data that are organized in some recognized fashion (e.g., database files, word processing document files, or Web pages) and stored in the volatile and/or non-volatile memory of the various nodes participating in the shared memory system.

Conclusion

For the purposes of this litigation, the disputed terms

contained in the ‘377 patent and the ‘229 patent shall have the

meanings ascribed to them in this order.

29 SO ORDERED.

Steven J. McAuliffe United States District Judge

September 2 1 , 2004

cc: Dorian Daley, Esq. Paul T . Ehrlich, Esq. Paul J. Hayes, Esq. Matthew D. Powers, Esq. Martha Van Oot, Esq. Alexander J. Walker, Esq.

30