Ex Parte Kumar

Board of Patent Appeals and Interferences

Aug 19, 2011

11237865 (B.P.A.I. Aug. 19, 2011)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE BOARD OF PATENT APPEALS
AND INTERFERENCES
____________

Ex parte MUPPIRALA KISHORE KUMAR
____________

Appeal 2009-014552
Application 11/237,865
Technology Center 2100
____________

Before JOHN A. JEFFERY, JAMES R. HUGHES, and
ANDREW J. DILLON, Administrative Patent Judges.

DILLON, Administrative Patent Judge.

Opinion Dissenting-in-part filed by Administrative Patent Judge JEFFERY.

DECISION ON APPEAL
Appellant appeals under 35 U.S.C. § 134(a) from the Examiner’s
rejection of claims 1-16, 19-41, and 43-53. We have jurisdiction under
35 U.S.C. § 6(b).
We affirm-in part.

Appeal 2009-014552
Application 11/237,865

2
STATEMENT OF THE CASE
Appellant’s invention manages memory across a plurality of partitions
by establishing a pool of free memory, allocating some of the free memory
to a target partition when required and the identifying memory within one or
more source partitions to replace the allocated free memory. Identified
memory is cleaned and added to the pool of free memory. A second method
identifies memory within a lesser loaded source partition for transfer to the
pool of free memory. See Abstract.
Claim 1 is illustrative with key disputed limitations emphasized:
1. A method of allocating memory across a plurality of
partitions,
including the steps of:

i) establishing a pool of free memory;

ii) allocating some of the free memory to a target
partition when required;

iii) identifying memory within one or more source
partitions to replace the allocated free memory;

iv) cleaning the identified memory; and

v) adding the cleaned memory to the pool.

The Examiner relies on the following as evidence of unpatentability:
Lee US 2003/0212873 A1 Nov. 13, 2003
Goloshubin WO 03/009144 A1 Jan. 30, 2003
(hereinafter “Wind River”)

Applicant’s Admitted Prior Art (AAPA) See Spec. 1-3

Appeal 2009-014552
Application 11/237,865

3
THE REJECTIONS
1. The Examiner rejected claims 1-2, 34-35, and 50-53 under 35
U.S.C. § 103(a) as unpatentable over Wind River and Lee.
Ans. 3-7.1
2. The Examiner has rejected claims 3-16, 19-33, 36-41, and 43-49
under 35 U.S.C. § 103(a) as unpatentable over Wind River, Lee
and AAPA. Id. at 7-23.
CONTENTIONS
Regarding representative claim 1, the Examiner finds that Wind River
discloses establishing a pool of free memory, allocating some of the free
memory to a target partition when required and adding cleaned memory to
the pool. The Examiner notes Wind River does not explicitly teach
identifying memory within one or more source partitions to replace allocated
memory and then cleaning the identified memory, but cites Lee for that
teaching. Ans. 4.
Appellant argues that neither Wind River or Lee teaches that after
memory is allocated from the pool of free memory, the allocated memory is
replaced, thereby replenishing the pool of free memory. Br. 11-14.
With regard to claim 13, the Examiner finds that Lee discloses
monitoring the workloads of each partition and identifying memory from a
lesser loaded source partition to be selected to replace allocated memory.
Ans. 12.

1 Throughout this opinion, we refer to the Appeal Brief filed December 1,
2008 and the Examiner’s Answer mailed March 6, 2009. We also note that
the rejection of claims 52 and 53 under 35 U.S.C. §112, first paragraph was
withdrawn by the Examiner. Ans. 23.
Appeal 2009-014552
Application 11/237,865

4
Appellant argues that Lee merely looks for unused memory within
one or more source partitions to deallocate, clean and utilize to replace
allocated memory from the pool of free memory, noting a belief that there is
no direct correlation between the unused status of a block of memory within
a partition and the workload of the partition. Br. 15-16.
The issues before us, then, are as follows:

ISSUES
1. Under § 103, has the Examiner erred in rejecting claims 1, 2,
34, 35, and 50-53 by finding that Wind River and Lee collectively would
have taught or suggested (1) establishing a pool of free memory, allocating
some of the free memory to a target partition when required and adding
cleaned memory to the pool, and (2) identifying memory within one or more
source partitions to replace allocated memory and then cleaning the
identified memory?
2. Under § 103, has the Examiner erred in rejecting claims 3-16, 19-
33, 36-41, and 43-49 by finding that Wind River, Lee and AAPA
collectively would have taught or suggested monitoring the workloads of
each partition and identifying memory from a lesser loaded source partition
to be selected to replace allocated memory?

Appeal 2009-014552
Application 11/237,865

5
FINDINGS OF FACT
We find that the following enumerated findings of fact (FF) are
supported by at least a preponderance of the evidence. Ethicon, Inc. v.
Quigg, 849 F.2d 1422, 1427 (Fed. Cir. 1988) (explaining the general
evidentiary standard for proceedings before the Office).
1. Wind River teaches a method for managing memory by responding
to a request for a memory allocation from a first application by
freeing a sufficient amount of memory from a “memory heap”
which contains all memory not currently allocated. If insufficient
memory is available in the “heap,” memory is freed by
deallocating memory from a process having a lower priority.
¶¶ [0004-0005].
2. Lee teaches a method for managing memory blocks by selecting a
memory block which is unused, isolating that memory block from
the partition to prevent use of that memory block and then
deallocating the memory block for assignment to another partition.
¶¶ [0062-0063].
3. Appellant’s Specification teaches that “the lesser-loaded partition
refers to the partition with the lesser memory load. The lesser
memory load of a partition may be determined by using one or a
combination of the following metrics: free memory within the
partition, average free memory within the partition over a period of
time, and frequency of use of memory by the partition.” Spec. 5:1-
5.

Appeal 2009-014552
Application 11/237,865

6
ANALYSIS
Appellant argues that neither Wind River nor Lee taught or suggested
the provision of a pool of free memory, portions of which are allocated in
response to memory requests, wherein the allocated memory is then replaced
within the pool of free memory as set forth within claims 1 and 34.
Br. 11-14.
We concur. Claims 1 and 34 require that memory from a pool of free
memory is allocated and then memory within one or more source partitions
is identified to replace the allocated free memory. The identified memory is
then cleaned, and the cleaned memory is added to the pool of free memory.
Both Wind River and Lee disclose the allocation and deallocation of
memory blocks. (FF-2, 3). However, neither reference discloses the
provision of a pool of free memory that is replenished after some portion of
that pool has been allocated in response to a request. Wind River discloses
maintaining a “memory heap” which contains all unallocated memory, but
even if that “memory heap” constitutes a “pool of free memory” Wind River
fails to disclose replacing memory within the “memory heap” which has
been allocated to a requesting process. Lee discloses the allocation and
deallocation of memory blocks within partitions but is similarly silent with
regard to replacing allocated memory within a pool of free memory.
Consequently, on this record, we are persuaded the Examiner erred in
rejecting claims 1, 34, and those claims which depend therefrom.
Appellant argues that Wind River, Lee and AAPA, considered
individually or collectively, fail to teach or suggest “monitoring the
workloads of each partition” and “identifying memory from a lesser-loaded
Appeal 2009-014552
Application 11/237,865

7
source partition to transfer to a free memory pool” as set forth in claims 13
and 39. Br. 14-17. Appellant admits that Lee discloses selecting a memory
block for deallocation by making a determination that the memory block is
unused, but argues that Lee does not teach or suggest checking each of a
plurality of partitions and that the status of a memory block as unused
provides no direct correlation with the workload of a partition. Id. At 16.
We are unpersuaded by this argument in view of Appellant’s express
teaching that free memory within a partition is one of the metrics by which
partition workload may be measured (FF 3), and the express teaching within
Lee that the search for unused memory is repeated until unused memory has
been found. Lee ¶[0062].
We find that the Examiner did not err in rejecting claims 13, 39 and
those claims which depend therefrom.

CONCLUSION
We affirm the Examiner’s rejection of claims 13-16, 19-33, 39-41 and
43-49 under § 103. However, we reverse the Examiner’s rejection of claims
1-12, 34-38, and 50-53 under § 103.

ORDER
The Examiner’s decision rejecting claims 13-16, 19-33, 39-41 and 43-
49 is affirmed. The Examiner’s rejection of claims 1-12, 34-38, and 50-53 is
reversed.

Appeal 2009-014552
Application 11/237,865

8
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv).

AFFIRMED-IN-PART

Appeal 2009-014552
Application 11/237,865

9
JEFFERY, Administrative Patent Judge, DISSENTING-IN-PART:

I join the majority in affirming the Examiner’s obviousness rejection
of claims 13-16, 19-33, 39-41, and 43-49. I also join the majority’s
reversing the rejection of claims 52 and 53. But I respectfully dissent from
the majority’s reversing the obviousness rejection of claims 1-12, 34-38, 50
and 51.
The majority finds that the Wind River reference “fails to disclose
replacing memory,” and Lee “is similarly silent with regard to replacing
allocated memory within a pool of free memory.” Maj. Op. at 6 (emphases
added). But as the Examiner correctly indicates (Ans. 24-26), the recited
method of claim 1 does not actually replace allocated free memory, but
rather identifies memory within source partitions to replace that memory—a
desired replacement result that simply need not occur. In short, identifying
memory for replacement does not actually replace that memory.
This fact is evidenced by the language of claim 1 itself which recites
five affirmative steps: (1) establishing a free memory pool; (2) allocating
some of that memory to a target partition; (3) identifying memory within one
or more source partitions to replace the allocated free memory; (4) cleaning
the identified memory; and (5) adding the cleaned memory to the pool.
Noticeably absent from these affirmative steps is actually replacing the
allocated memory. Rather, the only reference to replacement in claim 1 is
regarding a desired result of memory identification, namely to replace the

Appeal 2009-014552
Application 11/237,865

10
allocated free memory at some unrecited future time. Claim differentiation
principles are telling here: If replacing memory were intended in claim 1, it
would have surely been positively recited as it was in claim 52—a claim that
directly depends from claim 1.
Thus, Appellants’ contention that Lee does not replace allocated free
memory after it is allocated from the free memory pool (Br. 12-13; emphasis
added) is simply not commensurate with the scope of the claim which, as
noted above, does not replace anything. Nor does this argument show error
in the Examiner’s position, namely that Lee’s (1) selecting a memory block
from a partition at least identifies the memory (as does its later deallocation),
and (2) placing the block in the global pool at least serves to replace
memory allocated to a target partition. Ans. 23.
I see no reason why this identification would not at least be capable of
replacing previously-allocated free memory—a capability that is all the
claim requires in this regard. Although Lee’s process ends after (1)
deallocating memory associated with a first partition; (2) placing this
memory in a pool; and (3) allocating this memory to a second partition as
Appellants indicate (Br. 13), I nonetheless see no reason why this process
could not be repeatable such that identifications within the source partition
(Lee’s first partition) would not be at least capable of replacing free memory
previously allocated to the target partition (i.e., Lee’s second partition). Nor
have Appellants shown such incapability, let alone addressed the Examiner’s
broader construction of claim 1—a crucial deficiency in establishing
Examiner error on this record in my view.
Appeal 2009-014552
Application 11/237,865

11
Ultimately, the Examiner’s rejection is predicated on the fact that
claim 1 does not actually replace anything—a position that has a rational
basis on this record. That said, the Examiner’s position would be untenable
if the claim actually replaced the allocated free memory. But it does not.
I therefore respectfully dissent from the majority’s reversing the
obviousness rejection of claims 1-12, 34-38, 50, and 51. I join the majority,
however, in reversing the rejection of dependent claims 52 and 53 since
these claims—unlike the claims from which they depend— actually replace
allocated free memory.

pgc