Ex Parte Wong et al

Patent Trial and Appeal Board

Mar 29, 2016

11860914 (P.T.A.B. Mar. 29, 2016)

UNITED STA TES p A TENT AND TRADEMARK OFFICE
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR
11/860,914 0912512007 Michael K. Wong
51344 7590 03/31/2016
BROOKS KUSHMAN P,C /Oracle America/ SUN I STK
1000 TOWN CENTER, TWENTY-SECOND FLOOR
SOUTHFIELD, MI 48075-1238
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www .uspto.gov
ATTORNEY DOCKET NO. CONFIRMATION NO.
SUNM070362PUS 7446
EXAMINER
PREV AL, LIONEL
ART UNIT PAPER NUMBER
2475
NOTIFICATION DATE DELIVERY MODE
03/31/2016 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address( es):
docketing@brookskushman.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
BEFORE THE PATENT TRIAL AND APPEAL BOARD
Ex parte MICHAEL K. WONG, RABIN A. SUGUMAR, STEPHEN E.
PHILLIPS, HUGH KURTH, SURAJ SUDHIR, and JOCHEN BEHRENS
Appeal2014-004012
Application 11/860,914
Technology Center 2400
Before DEBRA K. STEPHENS, JOSEPH P. LENTIVECH, and
SHARON PENICK, Administrative Patent Judges.
LENTIVECH, Administrative Patent Judge.
DECISION ON APPEAL
Appellants 1 seek our review under 35 U.S.C. § 134(a) of the
Examiner's final rejection of claims 1-22, the only claims pending in the
application on appeal. We have jurisdiction over the pending claims under
35 U.S.C. § 6(b).
We AFFIRM-IN-PART.
1 According to Appellants, the real party in interest is Oracle America, Inc.
Br. 1.
Appeal2014-004012
Application 11/860,914
STATEMENT OF THE CASE
Appellants' Invention
Appellants' invention generally relates to a server interconnect system
that provides communication within a cluster of nodes. Spec. 2:23-24. The
server interconnect system is connectionless in the sense that hardware treats
each transfer of data independently, with specified data moved between the
nodes and queue/memory addresses specified for the transfer. Spec. 3 :9-11.
The invention does not require performance of a handshake before
communication starts or to maintain status information between pairs of
communicating entities. Spec. 3: 11-13. Claims 1 and 13, which are
illustrative, read as follows:
1. A server interconnect system for sending a message, the
system comprising:
a first server node operable to send and receive messages;
a second server node operable to send and receive
messages;
a first interface unit in communication with the first
server node, the first interface unit having a first set of message
send registers and a first set of message receive registers, the
first send and receive registers being mapped to user processes
executing on the first server node;
a second interface unit in communication with the second
server node, the second interface unit having a second set of
message send registers and a second set of message receive
registers, the second send and receive registers being mapped to
user processes executing on the second server node; and
a communication switch operable to receive and route a
message from the first or second server nodes when either of
the first or second send registers indicates that a valid message
is ready to be sent, wherein each message specifies a target
node and a receive queue identification via the first and second
sets of send registers without handshaking between the first and
2
Appeal2014-004012
Application 11/860,914
second server nodes before messaging begins to provide
connectionless messaging between the first and second server
nodes.
13. A server interconnect system for sending a message, the
system comprising:
a first server node operable to send and receive messages,
the first server node having a first receive queue;
a second server node operable to send and receive
messages, the second server node having a second receive
queue;
a first interface unit in communication with the first
server node, the first interface unit having a first set of message
registers, the first set of message registers including:
a first send doorbell register, the first send doorbell
register receiving information describing a message sent
by the first server node including a target node and first
server node send memory address;
a first send status register, the first send status
register receiving data regarding the status of a message
sent by the first server node;
a first receive queue configuration register, the first
receive queue configuration register providing
information regarding the first server node receive queue;
a first receive queue control register, the first
receive queue control register holding information
regarding availability of the first receive queue; and
a first receive queue status register, the first
receive queue status register holding information
regarding the current status of the first receive queue;
a second interface unit in communication with the second
server node, the second interface unit having a second set of
message registers, the second set of message registers
including:
3
Appeal2014-004012
Application 11/860,914
a second send doorbell register, the second send
doorbell register receiving information describing a
message sent by the second server node including a target
node and second server node send memory address;
a second send status register, the second send
status register receiving data regarding the status of a
message sent by the second server node;
a second receive queue configuration register, the
second receive queue configuration register providing
information regarding the second server node receive
queue;
a second receive queue control register, the second
receive queue control register holding information
regarding availability of the second receive queue; and
a second receive queue status register, the second
receive queue status register holding information
regarding the current status of the second receive queue;
a communication switch, the communication switch
being operable to receive and route a message from the first
server node to the second receive queue and the second server
node to the first receive queue when either the first or second
send status registers indicate that a valid message has been sent,
wherein each message specifies a target node and a receive
queue identification via the first and second send doorbell
registers without handshaking between the first and second
server nodes before messaging begins to provide connectionless
messaging between the first and second server nodes.
References
The Examiner relies on the following prior art in rejecting the claims:
Pandya
Cayton et al.
Ishizawa et al.
US 2004/0037319 Al
US 7,103,888 Bl
US 2006/0218336 Al
4
Feb.26,2004
Sept. 5, 2006
Sept. 28, 2006
Appeal2014-004012
Application 11/860,914
Rejections
Claims 1-3 and 5-12 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Pandya and Cayton. Final Act. 7-15.
Claim 4 stands rejected under 35 U.S.C. § 103(a) as being
unpatentable over Pandya, Cayton, and Ishizawa. Final Act. 15-16.
Claims 13-16 stand rejected under 35 U.S.C. § 102(b) as being
anticipated by Cayton. Final Act. 17-22.
Claims 17-22 stand rejected under 35 U.S.C. § 103(a) as being
unpatentable over Cayton and Pandya. Final Act. 22-27.
ANALYSIS
Claims 1-12
Issue 1: Did the Examiner err by finding the combination of Pandya
and Cayton teaches or suggests "a communication switch operable to receive
and route a message from the first or second server nodes ... without
handshaking between the first and second server nodes before messaging
begins to provide connectionless messaging between the first and second
server nodes," as recited in claim 1?
Appellants contend the combination of Pandya and Cayton fail to
teach or suggest the disputed limitations because the combination of Pandya
and Cayton "fails to disclose a connectionless RDMA system/method," as
required by claim 1. Br. 6. According to Appellants, "Pandya is directed to
a system for exchanging messages between servers using memory to
memory data transfers without substantial host intervention, it is based on a
TCP/IP hardware implementation, which is a connection-based system, to
5
Appeal2014-004012
Application 11/860,914
accomplish its functionality." Br. 6. Appellants contend Pandya does not
teach or suggest the use of registers to signal message availability but,
instead, teaches establishing "a TCP /IP connection between hosts and to
exchange RDMA type information and initiate the transfer." Br. 6.
We do not find Appellants' contentions persuasive. As found by the
Examiner (Ans. 2), Pandya teaches "[i]n addition to the TCP protocol, other
protocols such as SCTP and UDP protocols can be used, as well as other
protocols appropriate for transporting data streams." Pandya i-f 4. As also
found by the Examiner (Ans. 3), Cayton teaches sending and receiving
packets between work queue pairs via a connectionless data transfer or
connectionless channel. Cayton 6:48-52. As such, the combination of
Pandya and Cayton teaches or suggests the disputed limitation.
Issue 2: Did the Examiner err by finding the combination of Pandya
and Cayton teaches or suggests:
a communication switch operable to receive and route a
message from the first or second server nodes when either of
the first or second send registers indicates that a valid message
is ready to be sent, wherein each message specifies a target
node and a receive queue identification via the first and second
sets of send registers,
as recited in claim 1?
Appellants contend the combination of Pandya and Cayton fail to
teach or suggest the disputed limitation because "Pandya does not disclose a
communication switch operable to receive and route a message when send
registers indicate that a valid message is ready to be sent." Br. 7.
Appellants argue Pandya, instead, teaches "communicating to the target that
6
Appeal2014-004012
Application 11/860,914
it wants to read or write specific buffers using a TCP/IP connection between
the source and target." Br. 7 (citing Pandya i-f 98). Appellants contend
Cayton fails to teach or suggest the disputed limitation because:
Cayton discloses work queues having descriptors which are
defined as a data structure that describes a data movement
request that include segments that identify the type of data
movement (send or receive), completion status and an address
of an application data buffer for a data movement operation
(i.e., the address/description of where the data is to be retrieved
from in the application). There is no mention of the message
including the target node and receive queue id (used to identify
the targeted server receive queue).
Br. 7-8.
We do not find Appellants' contentions persuasive. One cannot show
non-obviousness by attacking the references individually when the rejection
is predicated upon a combination of prior art disclosures. See In re Merck &
Co., Inc., 800 F.2d 1091, 1097 (Fed. Cir. 1986). Here, the Examiner relies
on the combined teachings of Pandya and Cayton for teaching or suggesting
the disputed limitations. Ans. 4-5Appellants' contentions fail to address the
Examiner's findings regarding the combined teachings of the references and,
therefore, are unpersuasive of error.
Accordingly, we are unpersuaded the Examiner erred in finding the
combination of Pandya and Cayton teach or suggest the disputed limitation.
Issue 3: Did the Examiner err by finding the combination of Pandya
and Cayton teaches or suggests:
a first interface unit in communication with the first
server node, the first interface unit having a first set of message
send registers and a first set of message receive registers, the
7
Appeal2014-004012
Application 11/860,914
first send and receive registers being mapped to user processes
executing on the first server node;
a second interface unit in communication with the second
server node, the second interface unit having a second set of
message send registers and a second set of message receive
registers, the second send and receive registers being mapped to
user processes executing on the second server node,
as recited in claim 1?
Appellants contend the Examiner erred in finding the combination of
Pandya and Cayton teach or suggest the disputed limitations because:
Pandya discloses data transfer between client and server
networking stacks by copying data from application buffers (on
the source) to OS kernel space buffers and then to network
driver buffers. The data is then transferred to the NIC/HBA and
sent to the destination system. Pandya does not mention using
interface units in communication with the server node having
message send registers and message receive registers.
Br. 6-7.
We do not find Appellants' contention persuasive. The Examiner
relies on the combination of Pandya and Cayton as teaching the limitation.
Ans. 3--4. Specifically, Appellants acknowledge "Pandya is directed to a
system for exchanging messages between servers using memory to memory
data transfers without substantial host intervention." Br. 6. Cayton teaches
that a virtual interface consumer is a software process that communicates
using a virtual interface. Cayton 3:23-25. Cayton further teaches that a
virtual interface is an interface between a virtual interface network controller
(VI NIC) and the virtual interface consumer that allows the VI NIC to
directly access the process' (or application's) memory for data transfer
8
Appeal2014-004012
Application 11/860,914
operations between the application and the network and that the virtual
interface "includes a pair of work queues, one for send operations (a send
queue 21) and one for receive operations (receive queue 19)." Cayton 3:31-
40. As found by the Examiner (Ans. 2-3), Cayton teaches "a pair of work
queues (i.e., a send queue and a receive queue) are preferably provided at
each node of the channel based network 200." Cayton 5:3---6. Based on
these findings, the Examiner concludes it would be obvious to use the pair of
work queues included in each of the virtual interfaces of Cayton for
providing connectionless messaging for exchanging messages between the
servers of Pandya. Final Act. 9; Ans. 2-3. Appellants' contentions fail to
address the Examiner's findings regarding the combined teachings of
Pandya and Cayton and, therefore, are unpersuasive of error. Accordingly,
we are not persuaded the Examiner erred in finding the combination of
Pandya and Cayton teaches or suggests the disputed limitations.
For the foregoing reasons, we are not persuaded the Examiner erred in
rejecting claim 1 and claims 4--12, which depend from claim 1 and are not
argued separately with particularity. See Br. 6-8.
Claims 13-17
Issue 4: Did the Examiner err by finding Cayton discloses
a communication switch, the communication switch being
operable to receive and route a message from the first server
node to the second receive queue and the second server node to
the first receive queue when either the first or second send
status registers indicate that a valid message has been sent,
as recited in claim 13?
9
Appeal2014-004012
Application 11/860,914
Appellants contend Cayton does not disclose the disputed limitations
because Cayton does not disclose a server interconnect system for
messaging between server nodes, as required by claim 13. Br. 8. Appellants
contend:
Br. 8.
Cayton does not disclose a communications switch operable to
receive and route a message from the first server node to the
second receive queue and the second server node to the first
receive queue. The switched fabric referenced in Cayton does
not receive and route messages between the hosts. Rather, the
switched fabric disclosed by Cayton allows the hosts to
communicate with the shared I/O unit.
We are not persuaded. Cayton is directed to "[a] channel based
network ... that allows one or more hosts to communicate with one or more
remote fabric attached I/O units." Cayton, Abstract. Cayton further
discloses that the hosts and I/O units communicate over a switched fabric
using a push-push messaging protocol. Id. The Examiner finds Cayton's
hosts disclose the first and second server nodes (Final Act. 17 (citing Cayton
2:50-53)) and finds the pair of work queues included on each host disclose
the first and second receive queues (id. (citing Cayton 3:38--40)). As
indicated above, Cayton discloses methods of communication between the
hosts and the I/O units. The Examiner fails to make any finding, or
otherwise explain, how Cayton' s disclosure of a host and an I/O unit
communicating over a switched fabric discloses a communication switch
operable to receive and route a message from the first server node to the
second receive queue of the second server node and the second server node
10
Appeal2014-004012
Application 11/860,914
to the first receive queue of the first server node, as required by claim 13. 2
As such, we are constrained by the record to reverse the Examiner's
rejection of claim 13 and claims 14--16, which depend therefrom.
Claim 17 stands rejected under 35 U.S.C. § 103(a) based on Cayton
and Pandya. In rejecting claim 17, which depends from claim 13, the
Examiner fails to make any findings regarding how the teachings of Pandya
cure the above-noted deficiencies in the disclosure of Cayton. As such, we
are constrained by the record to reverse the Examiner's rejection of claim 17
for the reasons supra.
We do not reach Appellants' further allegations of error regarding the
rejection of claims 13-1 7 because we find the issue discussed above to be
dispositive.
Claims 18-22
Issue 5: Did the Examiner err by finding the combination of Cayton
and Pandya teach or suggest "routing the message to the second server node
without requiring handshaking between the source server node and the
second server node before routing the message to the second server node to
provide connectionless communication between the source server node and
the second server node," as recited in claim 18?
2 We express no opinion as to whether independent claim 13 would be
obvious over Cayton and Pandya. We leave any such further consideration
to the Examiner. Although the Board is authorized to reject claims under
37 C.F.R. § 41.50(b), no inference should be drawn when the Board elects
not to do so. See Manual of Patent Examining Procedure (MPEP)
§ 1213.02.
11
Appeal2014-004012
Application 11/860,914
Appellants contend the combination of Cayton and Pandya fail to
teach or suggest the disputed limitation because "Cayton discloses
communication between host nodes and I/O units and not messaging
between the host nodes" and "[l]ikewise, Pandya fails to disclose these
features as well" for the reasons discussed supra with respect to claims 1 and
13. Br. 9. As discussed above with respect to claim 1, Appellants
acknowledge (id. at 6) Pandya teaches exchanging messages between
servers using memory to memory data transfers without substantial host
intervention. Accordingly, we are unpersuaded the Examiner erred in
finding the combination of Cayton and Pandya teach or suggest the disputed
limitation for the reasons discussed supra with respect to claim 1.
Accordingly, we are not persuaded the Examiner erred in rejecting
claim 18 and claims 19-22, which depend from claim 18 and are not argued
separately with particularity.
DECISION
We affirm the Examiner's rejections of claims 1-12 and 18-22.
We reverse the Examiner's rejection of claims 13-17.
No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(l )(iv).
AFFIRMED-IN-PART
12