Ex Parte Bhattacharjee et al

Patent Trial and Appeal Board

Feb 17, 2015

12423247 (P.T.A.B. Feb. 17, 2015)

UNITED STATES PATENT AND TRADEMARK OFFICE
____________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
____________

Ex parte BISHWARANJAN BHATTACHARJEE, MUSTAFA CANIM,
and GEORGE ANDREI MIHAILA
____________

Appeal 2012-009100
Application 12/423,247
Technology Center 2100
____________

Before JEAN R. HOMERE, CARLA M. KRIVAK, and
CHARLES J. BOUDREAU, Administrative Patent Judges.

BOUDREAU, Administrative Patent Judge.

DECISION ON APPEAL
Appellants1 appeal under 35 U.S.C. § 134(a) from the Examiner’s
Final Rejection of claims 8–14. We have jurisdiction under 35 U.S.C.
§ 6(b).
We reverse.

1 Appellants identify International Business Machines as the real party
in interest. App. Br. 2.
Appeal 2012-009100
Application 12/423,247

2
STATEMENT OF THE CASE
Appellants’ Claimed Invention
The claimed invention relates to a method, information processing
system, and computer program storage product for optimizing the placement
of database objects on a multiplicity of storage devices. Abstract.
Representative Claim
Independent claim 8, reproduced below, is representative of the
subject matter on appeal.
8. An information processing system for optimizing placement of
database objects on a plurality of storage devices, the information
processing system comprising:
a memory;
a processor communicatively coupled to the memory; and
a database object placement advisor communicatively coupled
to the memory and the processor, wherein the database object
placement advisor is configured to:
place a set of database objects on a first storage device in a
plurality of storage devices, wherein each storage device in the
plurality of storage device comprises differing characteristics;
run a query workload on the set of database objects that have
been placed on the first storage device;
collect profiling information associated with the query
workload that is running;
determine, based on collecting the profiling information, a
number of pages read and written for each database object in the set of
database objects, and a total time spent through sequential and random
access for each data base object in the set of database objects;
select, based on the number pages read and written and the total
time spent through sequential and random access for each data base
object, a subset of database objects from the set of the database
objects to be stored on a second storage device in the plurality of
storage devices, wherein the second storage device is a separate
Appeal 2012-009100
Application 12/423,247

3
physical device from, and performs faster than, the first storage
device; and
store the subset of database objects on the second storage
device and all remaining database objects in the set of database
objects on the first storage device, where each database object in the
set of database objects is a randomly accessed database object, and
where each remaining database object is a sequentially accessed
database object.
Rejections on Appeal
Claims 8–12 and 14 stand rejected under 35 U.S.C. § 102(a) as being
anticipated by Ioannis Koltsidas & Stratis D. Viglas, Flashing Up the
Storage Layer, 1 PROCEEDINGS VERY LARGE DATABASE ENDOWMENT 514
(2008) (“Koltsidas”). Ans. 5–8.
Claim 13 stands rejected under 35 U.S.C. § 103(a) as being
unpatentable over the combination of Koltsidas and Bullen (US 7,707,351
B2; issued Apr. 27, 2010 (filed Oct. 31, 2002)). Ans. 9.

ISSUE ON APPEAL
The dispositive issue before us is whether the Examiner erred in
finding Koltsidas discloses “determin[ing] . . . a number of pages read and
written for each database object in the set of database objects, and a total
time spent through sequential and random access for each data base object in
the set of database objects,” as recited in claim 8.

ANALYSIS
We have reviewed Appellants’ arguments in the Briefs, the
Examiner’s rejection, and the Examiner’s response to Appellants’
Appeal 2012-009100
Application 12/423,247

4
arguments. We concur with Appellants’ conclusions that the Examiner erred
in finding Koltsidas discloses the disputed limitations set forth above.
Koltsidas describes using flash memory and magnetic disks at the
same level of memory hierarchy in commodity hardware systems and
placing a data page on one or the other storage medium according to the
workload of the page. See Koltsidas, p. 514, col. 1, para. 1. Pages with a
read-intensive workload are placed on the flash disk, for example, whereas
pages with a write-intensive workload are placed on the magnetic disk. Id.
Koltsidas presents a family of on-line algorithms “to decide the optimal
placement of a page and study their theoretical properties.” Id.
With respect to the disputed claim limitation, “determin[ing] . . . a
number of pages read and written for each database object in the set of
database objects,” the Examiner finds Koltsidas discloses “page count and
read/write cost data is used in an algorithm to determine what pages should
be migrated to the second, faster, storage device, which in this case is a flash
disk,” Ans. 6 (citing Koltsidas Fig. 1; p. 516, col. 2, para. 2; p. 517, col. 1,
para. 4 – col. 2, para. 2). According to the Examiner,
[A] cost counter is maintained for both reads and writes of
pages. This can be seen as cited on p. 517 col 1 section 4.1. As
seen in section 4.1 "Conservative Algorithm", the variable r
represents the cost of reading a page and w represents the cost
of writing a page.
Further support can be found in fig. 4, note on lines 5 and
7 that for each page the pg. reads variable is incremented for
each read, and that [pg. writes] is incremented for each write.
Also note on p. 518, col. 1. para. 2 the explanation of the figure
disclosing that for each page a read and write counter is
maintained.
Examiner has interpreted the disclosures of both a
counter and the ability to determine both when pages are read
Appeal 2012-009100
Application 12/423,247

5
and written as sufficient to anticipate “determining a number of
pages read and written.”
Ans. 12.
Appellants argue, and we agree, that although Koltsidas maintains a
counter that is incremented by a cost difference representing the cost units
that would have been saved if the page would have been read or written on
the other storage medium, that counter does not maintain the number of
pages read and written for each database object in the set of database
objects. App. Br. 10.
With respect to the disputed claim limitation, “determin[ing] . . . a
total time spent through sequential and random access for each data base
object in the set of database objects,” the Examiner finds that “B+ trees used
in the presentation use both random and sequential access patterns” and that
“a buffer manager can employ sequential access costs as opposed to random
access costs.” Ans. 6 (citing Koltsidas, p. 520, col. 2, para. 2; p. 524, col. 2,
para. 4). According to the Examiner,
The Office Action cites page 520, column 2 of the
Koltsidas reference, section 6, titled Experimental Study.
Paragraph one of the Experimental Study discloses evaluating
algorithms using a B+ tree. The B+ tree is disclosed to have
been chosen for the performance study because it exhibits “both
random and sequential access patterns.” As can be seen in table
1 and figure 9, total access times are being measured in
seconds.
Further evidence is shown on page 524, column two.
The last sentence of the first paragraph in column two explicitly
states that “Our benchmarks show that writing pages
sequentially to the flash disk is over 10 times faster than writing
them in a random fashion”, indicating that both sequential and
random access times are measured.
Appeal 2012-009100
Application 12/423,247

6
Further support for the limitation is also found on page
524, column two, paragraph 4, titled “Sequential Access
Patterns.” This paragraph discloses that sequential access costs
can be measured for page replacement and also for page
placement. Applicant suggests that this paragraph teaches away
from measuring both random and sequential access. However,
the context of the paragraph when read as a whole indicates
only that it is suggesting an alternative embodiment to utilizing
the random access costs in the page replacement algorithm.
The standard embodiment appears from this paragraph to use
random costs for page placement and sequential costs for page
replacement, which together would result in a total time for
sequential and random access for a query that consisted of both
a page placement and a page replacement.
Ans. 10–11.
Appellants argue, and we agree, that “Koltsidas . . . explicitly teaches
that either random access (read/writes) costs or sequential access
(read/writes) costs are used in the placement algorithms, not both.” App. Br.
11. As Appellants point out, the passage from page 524 of Koltsidas relied
upon by the Examiner states that a buffer manager “can employ sequential
access costs in the page placement algorithm as opposed to random access
ones” (emphasis added). We agree with Appellants that the use of the
phrase “as opposed to” shows that Koltsidas “only uses costs associated with
either random access read/writes or sequential reads/write, but not both.”
App. Br. 6.
Appellants further argue, and we again agree, that:
[T]he information shown in Table 1 and FIG. 9 of the Koltsidas
reference is with respect to experiments performed by the
authors of the paper, as compared to measurements that are
utilized by the algorithms of Koltsidas reference. For example,
Table 1 of the Koltsidas reference merely shows the average
Appeal 2012-009100
Application 12/423,247

7
time for read and writes on magnetic and flash disks for pages
placed on these disks. . . .
. . . .
. . . [E]ven assuming arguendo that the information in
Table 1 and FIG. 9 of the Koltsidas reference is used in the
placement/replacement algorithms (which it is not), the average
time . . . of all read/writes on the disk and the total execution
time of a query set (i.e., 50,000 queries) is clearly not the same
as “. . . a total time spent through sequential and random access
for each data base object in the set of database objects”.
Reply Br. 2–3.
Because we agree with Appellants that the Examiner erred in finding
Koltsidas discloses “determin[ing] . . . a number of pages read and written
for each database object in the set of database objects, and a total time spent
through sequential and random access for each data base object in the set of
database objects,” as recited in independent claim 8, we will not sustain the
Examiner’s rejection of claim 8 or its dependent claims 9–14.

DECISION
We reverse the Examiner’s rejection of claims 8–12 and 14 under
35 U.S.C. § 102(a) as being anticipated by Koltsidas.2
We reverse the Examiner’s rejection of claim 13 under 35 U.S.C.
§ 103(a) as being unpatentable over the combination of Koltsidas and
Bullen.

2 In the event of further prosecution, we leave it to Appellants and the
Examiner to determine whether the word “set” in the clause “where each
database object in the set of database objects is a randomly accessed
database object” in the last paragraph of claim 8 should more properly be
replaced with the word “subset.”
Appeal 2012-009100
Application 12/423,247

8
REVERSED

tj