U.S. patent application number 11/720728 was filed with the patent office on 2010-04-15 for modular computing environments.
Invention is credited to Gerald Aigner, Donald L. Beaty, Jimmy Clidaras, William Hamburgen, Winnie Leung, Montgomery Sykora, William Whitted.
Application Number | 20100091449 11/720728 |
Document ID | / |
Family ID | 38778954 |
Filed Date | 2010-04-15 |
United States Patent
Application |
20100091449 |
Kind Code |
A1 |
Clidaras; Jimmy ; et
al. |
April 15, 2010 |
Modular Computing Environments
Abstract
A computer system may include a connecting hub having a
plurality of docking regions and be configured to provide to each
docking region electrical power, a data network interface, a
cooling fluid supply and a cooling fluid return; and a plurality of
shipping containers that each enclose a modular computing
environment that incrementally adds computing power to the system.
Each shipping container may include a) a plurality of processing
units coupled to the data network interface, each of which include
a microprocessor; b) a heat exchanger configured to remove heat
generated by the plurality of processing units by circulating
cooling fluid from the supply through the heat exchanger and
discharging it into the return; and c) docking members configured
to releasably couple to the connecting hub at one of the docking
regions to receive electrical power, connect to the data network
interface, and receive and discharge cooling fluid.
Inventors: |
Clidaras; Jimmy; (Los Altos,
CA) ; Whitted; William; (Redwood City, CA) ;
Hamburgen; William; (Palo Alto, CA) ; Sykora;
Montgomery; (Denver, CO) ; Leung; Winnie;
(Mountain View, CA) ; Aigner; Gerald; (Pfaeffikon,
CH) ; Beaty; Donald L.; (Red Bank, NJ) |
Correspondence
Address: |
FISH & RICHARDSON P.C.
PO BOX 1022
MINNEAPOLIS
MN
55440-1022
US
|
Family ID: |
38778954 |
Appl. No.: |
11/720728 |
Filed: |
June 27, 2006 |
PCT Filed: |
June 27, 2006 |
PCT NO: |
PCT/US06/25163 |
371 Date: |
December 8, 2009 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60810451 |
Jun 1, 2006 |
|
|
|
Current U.S.
Class: |
361/679.49 ;
361/679.46 |
Current CPC
Class: |
H05K 7/2079 20130101;
H05K 7/1497 20130101; G06F 2200/201 20130101; G06F 1/20 20130101;
H05K 7/202 20130101; H05K 7/20754 20130101 |
Class at
Publication: |
361/679.49 ;
361/679.46 |
International
Class: |
G06F 1/20 20060101
G06F001/20 |
Claims
1. A computer system comprising: a connecting hub having a
plurality of docking regions and configured to provide to each
docking region electrical power, a data network interface, a
cooling fluid supply and a cooling fluid return; and a plurality of
shipping containers, each shipping container enclosing a modular
computing environment that adds a predetermined amount of computing
power to the computer system, each shipping container having a) a
plurality of processing units that are coupled to the data network
interface; b) a heat exchanger configured to remove heat generated
by the plurality of processing units from the shipping container by
circulating cooling fluid from the cooling fluid supply through the
heat exchanger and discharging heated cooling fluid into the
cooling fluid return; and c) docking members configured to
releasably connect to the connecting hub at one of the plurality of
docking regions to receive electrical power, connect to the data
network interface, receive cooling fluid from the cooling fluid
supply, and discharge return cooling fluid to the cooling fluid
return.
2. The computer system of claim 1, where each processing unit
comprises at least one storage device.
3. The computer system of claim 1, where the docking members
comprise quick-connect docking members.
4. The computer system of claim 1, where the connecting hub
comprises a spine.
5. The computer system of claim 1, where the connecting hub
comprises a ring.
6. A modular data center comprising: a connecting hub having a
plurality of docking regions and configured to provide to each
docking region electrical power; a data network interface, a
cooling fluid supply and a cooling fluid return; and a plurality of
containers, each container enclosing a modular computing
environment having a plurality of computing devices that
collectively add a predetermined amount of computing power to the
modular data center, each container having a) a first heat exchange
circuit configured to transfer heat from the computing devices to a
heat exchanger; b) a second heat exchange circuit comprising the
heat exchanger, the cooling fluid supply and cooling fluid return,
the second heat exchange circuit being configured to transfer heat
from the heat exchanger inside of the container to a system outside
of the container via cooling fluid in the cooling fluid supply and
cooling fluid return; and c) docking members that releasably
connect to the connecting hub at one of the plurality of docking
regions to receive electrical power, connect to the data network
interface, receive cooling fluid from the cooling fluid supply, and
discharge return cooling fluid to the cooling fluid return.
7. The modular data center of claim 6, where each modular computing
environment comprises a) a plurality of microprocessor boards that
receive power from the connecting hub and receive data from a
network coupled to the data network interface, process the received
data and transmit the processed data to the network via the data
network interface; b) a ventilation system that draws air from a
cold air plenum adjacent to the plurality of racks, across the
plurality of microprocessor boards, and into a warm air plenum;
where the heat exchanger cools air from the warm air plenum and
discharges the cooled air into the cold air plenum.
8. The modular data center of claim 7, where the modular computing
environment is configured to accommodate human occupancy.
9. The modular data center of claim 8, where the modular computing
environment includes a fire and smoke detection system and a fire
suppression system.
10. The modular data center of claim 6, where the ventilation
system comprises a first set of fans and a second set of fans, each
fan in the first set being disposed on one of the plurality of
microprocessor boards, each fan in the second set being disposed
adjacent to the heat exchanger.
11. The modular data center of claim 10, where adjacent fans in the
second set are supplied by electrical power from different
electrical circuits.
12. The modular data center of claim 10, further comprising a duct
disposed between the heat exchanger and at least a first fan and a
second fan in the second set, the duct fluidly coupling a region
adjacent to the first fan and a region adjacent to the second
fan.
13. The modular data center of claim 6, where the plurality of
microprocessor boards are disposed in racks that are suspended from
a ceiling of a corresponding container.
14. The modular data center of claim 6, where at least one of the
plurality of containers is stacked on top of another one of the
plurality of containers.
15. The modular data center of claim 6, further comprising a
facility-level cooling system that is disposed in the second heat
exchange circuit, coupled to the cooling fluid supply and cooling
fluid return, and configured to remove heat from cooling fluid in
the cooling fluid return.
16. The modular data center of claim 6, where each container is
sealed against environmental elements of wind and moisture.
17. The modular data center of claim 6, further comprising a
normally sealed drain configured to open when in contact with an
appreciable amount of liquid.
18. The modular data center of claim 6, where at least two of the
plurality of containers are standard size shipping containers.
19. The modular data center of claim 18, where the standard size is
a 1AAA shipping container.
20. A method of deploying a modular data center comprising:
providing a cooling system configured to provide cooling fluid to
various cooling devices, receive return cooling fluid from the
various cooling devices and remove heat from the received return
cooling fluid; providing an electrical power source; providing a
data communication interface; routing electrical power from the
electrical power source, a connection to the data communication
interface, and a cooling fluid supply and cooling fluid return from
the cooling system to a plurality of docking regions that each have
stubs that are configured to couple the electrical power, cooling
fluid supply, cooling fluid return and data communication interface
connection to a modular computing environment; disposing modular
computing environments at least two of the plurality of docking
regions, each modular computing environment configured to
incrementally add computing power to a data center and including a
plurality of computing resources and integrated cooling devices;
and coupling each self contained computing environment at a
corresponding docking region to the stubs to provide electrical
power, cooling fluid supply and cooling fluid return, and
connection to the data communication interface to the modular
computing environment.
21. A method of processing data comprising: transmitting a request
for data to a system comprising a) a connecting hub having a
plurality of docking regions and configured to provide to each
docking region electrical power, a data network interface, a
cooling fluid supply and a cooling fluid return; and b) a plurality
of shipping containers, each shipping container enclosing a modular
computing environment that adds a replacement amount of computing
power to the computer system, each shipping container having 1) a
plurality of processing units that are coupled to the data network
interface; 2) a heat exchanger configured to remove heat generated
by the plurality of processing units from the shipping container by
circulating cooling fluid from the cooling fluid supply through the
heat exchanger and discharging heated cooling fluid into the
cooling fluid return; and 3) docking members configured to
releasably connect to the connecting hub at one of the plurality of
docking regions to receive electrical power, connect to the data
network interface, receive cooling fluid from the cooling fluid
supply, and discharge return cooling fluid to the cooling fluid
return; and receiving data in response to the request.
Description
TECHNICAL FIELD
[0001] This description relates to modular computing
environments.
BACKGROUND
[0002] Computers have become widely adopted for various
applications, for both personal and corporate uses. Some computers
operate as stand-alone data-processing equipment, with perhaps a
peripheral such as a printer and a display device. Although useful
for many purposes, additional features and benefits may be achieved
if multiple computers are networked together to share information
and resources.
[0003] A computer network may be formed by connecting two or more
computing devices with an information channel. One type of network
is a local area network (LAN). A typical residential LAN, for
example, may connect two computers to a printer. A typical
corporate LAN, for example, may allow many users to share resources
and large amounts of information, including databases and
application software.
[0004] A second type of network is a wide area network (WAN). An
example of a WAN is the Internet. WANs such as the Internet allow
many computer devices to communicate messages and share
information. When multiple LANs are interoperable with one or more
WANs, opportunities for computing devices to communicate and share
information greatly expand.
[0005] From the perspective of an individual computing device that
is connected to a network, users may direct the communication of
information over a network with a user interface generated by a web
browser application. A web browser is typically configured to
enable the user to access web sites on the Internet or the World
Wide Web. Web browsers allow users to easily send and receive
messages over a network in packets of information. Such packets of
information may include the address of a search engine website,
such as www.dogpile.com, for example.
[0006] The popularity and simplicity of sharing information over
networks, such as the Internet, has resulted in demand for data
processing and/or storage capacity to support high network traffic
volume. One mechanism to address this need may be referred to as a
data center. In the context of the Internet, a data center may
provide one or more of processing, storage, and support functions
that improve performance or enhance the utility of the Internet.
Data centers may also be deployed in other contexts. Financial
institutions, for example, may employ one or more data centers to
store financial account and transaction information.
[0007] A data center may provide data processing and/or storage
capacity. In operation, a data center may be connected to a
network, and may receive and respond to various requests from the
network to retrieve, process, and/or store data. In addition to
extensive data processing and data storage capabilities, data
centers typically support high speed data transfer and routing
capabilities. To meet future network demands, data center capacity
may continue to expand.
SUMMARY
[0008] A computer system may include a connecting hub having a
plurality of docking regions and be configured to provide to each
docking region electrical power, a data network interface, a
cooling fluid supply and a cooling fluid return; and a plurality of
shipping containers that each encloses a modular computing
environment that adds a predetermined amount of computing power to
the system. Each shipping container may include a) a plurality of
processing units coupled to the data network interface, each of
which include a microprocessor; b) a heat exchanger configured to
remove heat generated by the plurality of processing units by
circulating cooling fluid from the supply through the heat
exchanger and discharging it into the return; and c) docking
members configured to releasably couple to the connecting hub at
one of the docking regions to receive electrical power, connect to
the data network interface, and receive and discharge cooling
fluid.
[0009] In some implementations, each processing unit may further
include at least one storage device. The docking members may
include quick-connect docking members. The connecting hub may
include a spine; the connecting hub may include a ring.
[0010] In some implementations, a modular data center includes a
connecting hub and a plurality of containers. The connecting hub
may have a plurality of docking regions and may be configured to
provide to each docking region electrical power, a data network
interface, a cooling fluid supply and a cooling fluid return. Each
container may enclose a modular computing environment having a
plurality of computing devices that collectively add a large,
predetermined amount of computing power to the modular data center.
Each container may further have a) a first heat exchange circuit
configured to transfer heat from the computing devices to a heat
exchanger; b) a second heat exchange circuit comprising the heat
exchanger, the cooling fluid supply and cooling fluid return, the
second heat exchange circuit being configured to transfer heat from
the heat exchanger inside of the container to a system outside of
the container via cooling fluid in the cooling fluid supply and
cooling fluid return; and c) docking members that releasably
connect to the spine at one of the plurality of docking regions to
receive electrical power, connect to the data network interface,
receive cooling fluid from the cooling fluid supply, and discharge
return cooling fluid to the cooling fluid return.
[0011] In some implementations, each modular computing environment
includes a) a plurality of microprocessor boards that receive power
from the connecting hub and receive data from a network coupled to
the data network interface, process the received data and transmit
the processed data to the network via the data network interface;
b) a ventilation system that draws air from a cold air plenum
adjacent to the plurality of racks, across the plurality of
microprocessor boards, and into a warm air plenum. The heat
exchanger may cool air from the warm air plenum and discharge the
cooled air into the cold air plenum.
[0012] In some implementations, the modular computing environment
may be configured to accommodate human occupancy. The modular
computing environment may include a fire and smoke detection system
and a fire suppression system. The modular computing environment
may include an internal walkway and a means for ingress and egress
at one end of the walkway. The internal walkway may be disposed in
the cold air plenum.
[0013] In some implementations, the ventilation system may include
a first set of fans and a second set of fans. Each fan in the first
set may be disposed on one of the plurality of microprocessor
boards, and each fan in the second set may be disposed adjacent to
the heat exchanger. Adjacent fans in the second set may be supplied
by electrical power from different electrical circuits.
[0014] In some implementations, the modular data center further
includes a duct disposed between the heat exchanger and at least a
first fan and a second fan in the second set. The duct may fluidly
couple a region adjacent to the first fan and a region adjacent to
the second fan. The plurality of microprocessor boards may be
disposed in racks that are suspended from a ceiling of a
corresponding container.
[0015] In some implementations, at least one of the plurality of
containers may be stacked on top of another one of the plurality of
containers. The modular data center may further include a
facility-level cooling system that is disposed in the second heat
exchange circuit, coupled to the cooling fluid supply and cooling
fluid return. The facility-level cooling system may be configured
to remove heat from cooling fluid in the cooling fluid return.
[0016] In some implementations, each container is sealed against
environmental elements of wind and moisture. The modular data
center may further include a normally sealed drain configured to
open when in contact with an appreciable amount of liquid. In some
implementations, at least two of the plurality of containers may be
standard size shipping containers. The standard size may be a 1AAA
shipping container.
[0017] In some implementations, a method of deploying a modular
data center includes a) providing a cooling system configured to
provide cooling fluid to various cooling devices, receive return
cooling fluid from the various cooling devices and remove heat from
the received return cooling fluid; b) providing an electrical power
source; c) providing a data communication interface; d) routing
electrical power from the electrical power source, a connection to
the data communication interface, and a cooling fluid supply and
cooling fluid return from the cooling system to a plurality of
docking regions that each have stubs that are configured to couple
the electrical power, cooling fluid supply, cooling fluid return
and data communication interface connection to a modular computing
environment; e) providing modular computing environments at least
two of the plurality of docking regions, each modular computing
environment configured to incrementally add computing power to a
data center and including a plurality of computing resources and
integrated cooling devices; and f) coupling each self contained
computing environment at a corresponding docking region to the
stubs to provide electrical power, cooling fluid supply and cooling
fluid return, and connection to the data communication interface to
the modular computing environment.
[0018] In some implementations, a method of processing data
includes transmitting a request for data to a system and receiving
data in response to the request. The system may include a) a
connecting hub having a plurality of docking regions and configured
to provide to each docking region electrical power, a data network
interface, a cooling fluid supply and a cooling fluid return; and
b) a plurality of shipping containers. Each shipping container may
enclose a modular computing environment that adds a replacement
amount of computing power to the computer system. Each shipping
container may also have 1) a plurality of processing units that are
coupled to the data network interface; 2) a heat exchanger
configured to remove heat generated by the plurality of processing
units from the shipping container by circulating cooling fluid from
the cooling fluid supply through the heat exchanger and discharging
heated cooling fluid into the cooling fluid return; and 3) docking
members configured to releasably connect to the connecting hub at
one of the plurality of docking regions to receive electrical
power, connect to the data network interface, receive cooling fluid
from the cooling fluid supply, and discharge return cooling fluid
to the cooling fluid return.
[0019] The details of one or more implementations are set forth in
the accompanying drawings and the description below. Other features
and advantages will be apparent from the description and drawings,
and from the claims.
DESCRIPTION OF DRAWINGS
[0020] FIG. 1 is a perspective diagram of an example modular
computing environment.
[0021] FIG. 2 is perspective diagram of an example cooling system
that may be included in the modular computing environment that is
shown in FIG. 1.
[0022] FIG. 3 is a cross-section of the example modular computing
environment that is shown in FIG. 1.
[0023] FIG. 4 is a block diagram showing example thermal circuits
that may exist in a modular data center.
[0024] FIG. 5 is a another block diagram showing example thermal
circuits that may exist in a modular computing environment.
[0025] FIG. 6 provides a front view of an example rack.
[0026] FIG. 7 provides a rear view of an example rack.
[0027] FIG. 8 is a block diagram of an example processing board
that may be included in a modular computing environment.
[0028] FIG. 9 is perspective diagram of an example modular data
center constructed with several modular computing environments.
[0029] FIG. 10 is a perspective diagram of another example modular
data center constructed with several modular computing
environments.
[0030] FIG. 11 is a block diagram of an example facility-level
electrical distribution system for a data center.
[0031] FIG. 12 is a block diagram of an example information
provider, according to some implementations.
[0032] FIG. 13 is a block diagram illustrating how portions of an
information provider may be implemented using a modular data
center.
[0033] Like reference symbols in the various drawings indicate like
elements.
DETAILED DESCRIPTION
[0034] FIG. 1 is a perspective diagram of an example modular
computing environment that can be configured to add computing power
to a data center or other large computing system. The modular
computing environment may include a large number of processing
devices for executing computer program instructions and processing
data and a large data storage capacity for storing and retrieving
data. Each modular computing environment may provide a large,
predetermined amount of computing power (e.g., 2000 microprocessors
and 50 terabytes of storage). In operation, the modular computing
environment may be connected to a network, and may receive and
respond to various requests from the network to retrieve data,
process data or store data. In some implementations, more than one
modular computing environment may be combined to form a massively
parallel data center, and the massively parallel data center may be
replicated at various physical locations. As used in this
description, "massively parallel" refers to a distributed computer
system having many individual nodes, each of which may be
essentially an individual computer having, in some implementations,
a processor, memory, and a storage device.
[0035] The many individual nodes of the computer system may be
linked together by a network to allow nodes to communicate with
each other and to communicate with networks and devices external to
the computer system. In some implementations, each node of the
massively parallel system may be employed to execute a small part
of a larger program, task or operation.
[0036] As one example application of a data center or other
massively parallel computer system, a financial institution may
employ a number of data centers at different physical locations to
store customers' financial account and transaction information.
When a customer of the financial institution makes a withdrawal or
deposit from a teller station, a computer programming running at
the teller station may receive input from the teller identifying
the customer's account and specifying the type and amount of the
transaction; the computer program may then send information to one
or more data centers to cause the customer's account to be adjusted
to reflect the withdrawal or deposit. Because the data
corresponding to the customer's account may be replicated in
multiple physical locations, multiple data centers may be involved
in the transaction.
[0037] At substantially the same time the first customer is making
a withdrawal or deposit from a teller station, a second customer
may be initiating online a transfer of funds or a payment of a bill
from his or her home computer. A computer program running on a web
server that is accessible from the second customer's home computer
may also cause the second customer's account information to be
updated at multiple data centers; these data centers may be the
same or different data centers than those involved in the
transaction of the customer making the withdrawal or deposit from
the teller station.
[0038] Many other financial transactions involving other customers
of the financial institution may take place substantially
simultaneously, from various computer terminals within different
physical offices of the financial institution and from various
other computer terminals that access the financial institution's
network from the outside (e.g., home or office computer terminals
of customers accessing the financial institution's services via the
internet). Accordingly, the data centers may be required to process
in parallel a large number of transactions in a short amount of
time. Such a requirement may translate into a need for a massively
parallel computing environment having a large number of processors
to execute code, a large storage capacity, and a high-bandwidth
network interface.
[0039] As another example application, an online information
provider that supplies content and services (such as a search
engine function) to many users substantially simultaneously may
employ a number of data centers at different physical locations to
store the supplied content and data needed to provide the search
engine service. Some of the supplied content may be physically
stored by the information provider's computing system; other data
that may be identified in response to a user's invocation of the
search engine function may be indexed by and in the information
provider's computing system. In particular, various computer
programs running on the information provider's computer system may
automatically navigate or "crawl" networks such as the Internet to
identify and index information that has not yet been indexed or
that has been updated since last being indexed. The indexing
process may involve temporarily retrieving the information, parsing
and analyzing it to identify a theme or topic, and storing a
corresponding source identifier of the information based on the
theme or topic in an information retrieval tool such as a database
or hash table.
[0040] An online information provider may be required to handle a
very large volume of simultaneous data traffic. For example, the
information provider's computing system may receive numerous
requests at approximately the same time from computer terminals of
users attempting to employ the information to locate sources of
information corresponding to particular topics. Computer programs
running on the information provider's computing system may receive,
parse and analyze the various requests; convert them into an
appropriate format; submit them to interfaces corresponding to the
information retrieval tool; receive responses from the information
retrieval tool identifying content sources that store information
corresponding to the requests; and transmit data associated with
the content sources to the computer terminals from which the
requests were originally received.
[0041] In some implementations, the transmitted data may include a
pointer (e.g., a hyperlink) to the corresponding information. In
some implementations, the transmitted data may include a "snippet,"
or a portion, of the corresponding information (e.g., a sentence
fragment from a document that includes a search term submitted with
a request). In the latter case, in which a snippet is included with
data transmitted to a computer terminal in response to a request
received from that search terminal, snippets corresponding to
various indexed information may be stored in the index or in other
data storage facilities associated with the index or with the
information provider.
[0042] In order to simultaneously receive and process a large
number of requests from users for search engine services, the
information provider may need a massively parallel computing
environment having a large number of processors to execute code, a
large storage capacity, and a high-bandwidth network interface.
[0043] As another example, one or more modular computing
environments may be used as a high-volume "portable" data storage
device. Large quantities (e.g., tens of terabytes) of frequently
accessed data may be copied to one or more modular computing
environments, and the modular computing environments may be
transported (e.g., by cargo ship, truck or train) to a point of
use. As a more concrete example, an insurance company may employ a
modular computing environment to create a temporary claims office
at the site of a major natural disaster. In particular, following,
for example, a severe hurricane, an insurance company may store
information about its policyholders in a modular computing
environment and transport the modular computing environment to the
site of the hurricane to create a temporary claims office.
[0044] As another example, one or more modular computing
environments may be used as a laboratory network to simulate, in a
test environment, a large network such as an enterprise intranet or
a portion of the Internet. The modular computing environment may
provide enough processing power and storage capacity to mimic a
live network, while still remaining disconnected from live systems
or live data; such an application may provide network equipment
designers or manufacturers an opportunity to test or validate new
hardware designs in an environment that approximates the intended
operating environment without disrupting actual live data. Various
components and aspects of massively parallel computing environments
that may be used in the example applications outlined above are
described in detail with reference to the figures.
[0045] FIG. 1 illustrates a modular computing environment 100 that
may be used to add computing power to a massively parallel
computing system, such as, for example, a data center. In some
implementations, different modular computing environments
("modules") may provide a replacement amount of processing power or
storage capacity (e.g., 2000 microprocessors and 50 terabytes of
storage). That is, different modules may be substantially
interchangeable. In some implementations, different modules may
provide different ratios of processing power to storage capacity,
and the target application may determine the appropriate ratio. For
example, storage-intensive applications may call for modules that
have a large number of storage devices, but the applications may
not require a large number of processors; "lab network"
applications designed to simulate large networks offline may call
for modules with a large number of processors, but the applications
may not require a large number of storage devices.
[0046] As shown, the modular computing environment 100 includes
various computing resources, an integrated cooling system and an
integrated power distribution system. In some implementations, the
modular computing environment 100 is easily transported, protected
from environmental elements and thus suitable for use both inside
or outside a building, configured to be easily and quickly
connected with other modular computing environments to form larger
computing systems, and easily serviced by human operators. For
example, in some implementations, as shown, the modular computing
environment is protected by an enclosure 101, such as a shipping
container. The enclosure is sealed on the outside from
environmental elements and equipped with access to the interior for
human operators to service the various computing resources.
[0047] The computing resources of the modular computing environment
100 include a large number of processor boards. As used in this
description, "processor board" refers generally to a board that
includes computing resources. In some implementations, as described
in greater detail with reference to FIG. 8, each processor board
may include one or more microprocessors for executing computer
program instructions, dynamic memory for use while executing the
computer program instructions, storage devices for storing
persistent data, and an interface to link the processor board with
other processor boards and with one or more networks. As described
above, a target application may determine an appropriate ratio of
processors to storage devices. In some implementations, a processor
board may include only storage devices and corresponding
controller(s) and may omit general-purpose microprocessors or
general-purpose dynamic memory.
[0048] The processor boards may be physically disposed in racks,
and racks carrying processor boards may substantially line the
modular computing environment along the length of each side of the
enclosure. For example, in some implementations, the modular
computing environment may include well over 1,000 processor boards.
As shown in FIG. 1, the shaded region 104 represents a single
processor board. The processor board 104 may be designed to be
easily removed from the corresponding rack in order to be
serviced.
[0049] Each processor board may be coupled to a network internal to
the modular computing environment, and each processor board may
further be accessible from a network external to the enclosure. In
some implementations, the network internal to the modular computing
environment includes a patch panel 107. The patch panel 107 may
provide access connections (e.g., an RJ-45 Ethernet jack or optical
interface) to network interfaces on each processor board, and the
access connections may be coupled to the processor board network
interfaces by, for example, category 5 or category 6 cable or fiber
optic cable. The network may further include a router 110 to
digitally route data from a high-bandwidth data connection external
to the enclosure to processor boards within the enclosure. The
network may also include one or more switches (not shown) to
digitally route data from one processor board in the enclosure to
another processor board in the enclosure.
[0050] A high-bandwidth connection 113 may couple the patch panel
to one or more network interface connection 158 configured to
connect to a network external to the modular computing environment
100. The high-bandwidth connection 113 may include one or more
high-speed copper interfaces or one or more optical interfaces. For
example, the high-bandwidth connection may include a gigabit
Ethernet access point for balance twisted pair copper wire and an
optical interface for connection to single- or multi-mode optical
fiber. In particular, the high-bandwidth network connection may
include a 10-gigabit Ethernet interface for coupling single-mode
optical fiber carrying SONET (synchronous optical network signals)
at carrier level 192 (OC-192) to an optical router 110.
[0051] In some implementations, a monitor/control terminal 111 may
be provided to allow a human operator to monitor the various
connections on the patch panel 107, switches or router 110, or to
monitor actual data traffic. In some implementations, the
monitor/control terminal 111 may enable a human operator to
dynamically switch interconnections between various processor
boards within the modular computing environment 100, or connections
between an external network and various processor boards.
[0052] During operation, the computing resources may generate a
significant amount of heat within the enclosure 101, and as such,
the modular computing environment 100 includes an integrated
cooling system to remove the generated heat from the enclosure 101.
In some implementations, the integrated cooling system includes a
number of heat exchangers 116 and a number of fans 119 or pumps to
circulate cooling fluid (e.g., water or refrigerant) and fluid to
be treated (e.g., air) through the heat exchangers 116. In
particular, the heat exchangers 116 may be liquid-air heat
exchangers that transfer heat in air that is cycled through them to
a cooling fluid that is also cycled through the heat exchangers
116. The cooling fluid may be cycled through a system external to
the enclosure 101, where the heat absorbed in the heat exchangers
116 may be removed. Each heat exchanger 116 may be supplied with
cooling fluid from a cooling fluid supply line 122; cooling fluid
that has been warmed by the heat exchanger 116 may be discharged
into a cooling fluid return line 125. Additional details of cooling
systems are provided with reference to other figures.
[0053] The modular computing environment 100 also includes a power
distribution system that receives power from outside the enclosure
and distributes it to various loads within the enclosure. Power may
be received through a connection 128 to an external power supply.
Once inside the enclosure, the power may be routed to a power
distribution box 131 and distributed through a number of circuits
to power the computing resources, cooling system, network switching
equipment and other electrical loads within the enclosure 101. The
power distribution box 131 may include various circuit breakers or
other over-current protection for each circuit, and each circuit
may also include surge suppression devices or filters to smooth out
the power or to prevent high frequency noise generated within the
enclosure from being coupled into the power supply outside the
enclosure. Additional details of an example power distribution
system are provided with reference to other figures.
[0054] In order for the modular computing environment 100 to be
serviced, the interior of the enclosure 101 may need to accommodate
human occupancy; this may require additional features that satisfy
both physical human occupancy requirements and any legal or safety
requirements that may exist (e.g., municipal building or occupancy
requirements). For example, the modular computing environment 100
may include interior lights, a source of fresh air, fire detection
and suppression systems; and temperature, humidity and noise may be
controlled to be within certain ranges.
[0055] Additional structural details of the example modular
computing environment 100 are now described. As mentioned above,
the modular computing environment 100 is protected by an enclosure
101 that is sealed from environmental elements such as, wind, rain
or snow. In some implementations, the enclosure 101 is a standard
shipping container. For example, the enclosure may be an ISO 1AAA
container, sometimes referred to as a "high cube" 40' container,
having an approximate inside width of 7'8'', an approximate inside
height of 8'10'', and an approximate inside length of 39'5''. Other
container dimensions are possible. In particular, the modular
computing environment 100 may be enclosed by a standard height 20'
container (having an inside height of approximately 7'10'' and an
inside length of 19'4''); or the container may be longer (e.g.,
45', 48' or 53'). Moreover, enclosures other than shipping
containers may be used. For example, in some implementations, the
enclosure may be a metal- or wood-framed enclosure with protective
siding material. Racks within the enclosure 101 may be suspended,
and the walkway 142 may be segmented into small, removable sections
to facilitate access to the components under the racks (e.g., heat
exchangers 116 or fans 119).
[0056] The enclosure 101 may be configured to be easily
transportable. Shipping containers are particularly well-suited for
this purpose and are designed to be easily transported, for
example, by a cargo ship, train or truck. The enclosure may also be
stackable and may include locking mechanisms (not shown) to secure
multiple enclosures in a stacked arrangement. The enclosure may
include fork-lift pockets 143 and corner fittings 146 to facilitate
handling by cargo transport devices (e.g., fork-trucks, cranes,
overhead booms, conveyors, etc.).
[0057] In some implementations, a modular computing environment 100
may be transported to a desired location and immediately deployed
by being connected to an electrical power supply, a cooling system
and a network interface, as described in greater detail below. In
some implementations, a modular computing environment 100 may be
deployed while it is still disposed on a transport device. For
example, a modular computing environment 100 may be transported to
an intended deployment site by cargo ship, train or truck, and the
modular computing environment may be connected to electrical power
a cooling system and a network while it is still on the cargo ship,
train or truck. In general, the modular computing environment 100
may be configured to be rapidly deployed.
[0058] The modular computing environment 100 includes ports that
facilitate quick connection to utilities and other services
required for operation. For example, as shown in FIG. 1 and
described above, a power port 149 is provided to supply power to
the modular computing environment 100 via the power supply
connection 128, cooling fluid supply port(s) 152 are provided to
connect to the cooling fluid supply line(s) 122, cooling fluid
return port(s) 155 are provided to connect to the cooling fluid
return line(s) 125, and one or more network interface connections
158 are provided to couple an external network to the modular
computing environment 100 via the high-bandwidth connection 113. In
some implementations, openings 161 may be disposed in exterior
doors 137 (left door 137 shown) of the enclosure 101 to allow
access to the various ports 149, 152, 155 and 158 when the doors
137 are either opened or closed. In some implementations, the
various ports 149, 152, 155 and 158 are provided in a location that
is unaffected by operation of any exterior doors 137 (e.g., at the
top or side of the enclosure 101). In some implementations, the
enclosure 101 includes a second, smaller door 164 cut into an
exterior doors 137 such that the exterior door 137 may remain
closed and sealed at all times. Seals (not shown) may protect the
ports to keep moisture and particulates out of the interior of the
enclosure. For example, the ports may include bladder seals,
grommets or flexible dust covers.
[0059] Modular computing environments 100 with protective
enclosures 101 as described above may have several benefits. For
example, such modular computing environments 100 may facilitate
quick assembly of large data centers. Large portions of a data
center may be prefabricated and quickly deployed; in particular,
portions of data centers may be constructed in parallel, rather
than in sequence. Critical portions of data centers may be mobile
and easily transported from one site to another. Portions of the
data center may be constructed by manufacturing laborers, rather
than trade laborers, possibly resulting in reduced construction
costs.
[0060] FIG. 2 is a perspective diagram further illustrating details
of a portion 201 of an example cooling system ("cooling system
201") that may be included in the modular computing environment 100
that is shown in FIG. 1. As shown, the cooling system 201 includes
a liquid-air heat exchanger 116 that is coupled to a cooling fluid
supply line 122 and a cooling fluid return line 125. In operation,
a cooling fluid is circulated from the cooling fluid supply line
122, through the heat exchanger 116, and discharged into the
cooling fluid return line 125. Heat and air passing through the
heat exchanger (e.g., via paths A and B) may be removed and
transferred to the cooling fluid. For example, cooling fluid may
have temperature T.sub.CF in the cooling fluid supply line 122,
before being circulated through heat exchanger 116. After the
cooling fluid has been circulated through heat exchanger 116, the
cooling fluid may have a temperature of T.sub.WF, increased from
T.sub.CF by the transfer of heat from the air. Air passing through
the heat exchanger may have a warm input temperature T.sub.WA, and
a cooled output temperature T.sub.CA.
[0061] To move air through the heat exchanger 116 from an input
region (corresponding to path A) to an output region (corresponding
to path C), the cooling system 201 includes multiple fans 119. The
fans 119 may be mounted to a duct 202 that fluidly couples a space
in front of the heat exchanger 116. The duct 202 may help to
balance variations in pressure along the duct's length; in
addition, the duct 202 may improve the overall reliability of the
cooling system 201 by pulling a large volume of air through the
heat exchanger 116, even if one of the fans 119 stops functioning.
For example, even if a middle fan 119 stops functioning, the duct
202 may even out pressure variations relative to the outside fans
119 that may otherwise exist and pull air through the middle of the
heat exchanger 116. As shown, the fans 119 pull air through the
heat exchanger 116, but in other implementations, fans may be
disposed such that they push air through the heat exchanger
116.
[0062] An overall cooling system may include other portions, in
addition to the portion 201 that is illustrated in FIG. 2. For
example, a cooling system may include a line of heat exchangers 116
disposed along the length of each side of the enclosure 101;
similarly, the cooling system may include a duct 202 and a set of
fans 119 to correspond to each heat exchanger 116. The duct 202 may
be segmented and may correspond to the length of heat exchanger
116, or the duct 202 may be continuous along a larger portion of
the length of the enclosure 101. More or fewer fans 119 may
correspond to each heat exchanger 116, and the length of individual
heat exchangers 116 may vary. In some implementations, each heat
exchanger 116 is about 8' in length and corresponds to 3 motorized
impeller fans. Other arrangements are contemplated. For example,
air movers other than the fans 119 may be employed, and heat
exchangers having designs other than the design depicted by the
heat exchanger 116 may be used.
[0063] FIG. 3 is a cross-section of the example modular computing
environment 100 that is shown in FIG. 1. FIG. 3 illustrates an
example flow of air through the modular computing environment 100.
As shown in FIGS. 1 and 3, racks 303 of processor boards are
disposed along much of the length of the enclosure 101. The racks
303 are spaced apart from exterior sidewalls 306 of the enclosure,
forming an open space, or warm air plenum 309. A "cold air plenum"
312 may be formed by the interior space between the racks 303. In
some implementations, air flows between the cold air plenum 312 and
the warm air plenum 309 in one of two ways: cold air flows from the
cold air plenum 312 to the warm air plenums 309 by flowing through
the racks 303, across the processor boards (e.g., via paths C, D,
A); warm air from the warm air plenums 309 moves to the cold air
plenum 312 by flowing through the heat exchangers 116 (via path A,
B, C). As the air flows over the processor boards (via path D), it
absorbs heat generated by the processing boards, and its
temperature may rise from T.sub.CA in the cold air plenum 312 to
T.sub.WA in the warm air plenum 309. Conversely, as the air flow
through the heat exchanger 116, heat is removed and transferred to
the cooling fluid; the temperature of the air may drop from
T.sub.WA to T.sub.CA, as a result, while the temperature of the
cooling fluid circulating through the heat exchanger may rise from
temperature T.sub.CF to T.sub.WF. In some implementations,
processor boards may include a fan (e.g., fan 315) to cool
components on the processor board and/or to minimize thermal
inconsistencies within the warm air plenum 309 or cold air plenum
312.
[0064] FIG. 4 is a block diagram illustrating example thermal
circuits 401 and 402 that may exist in a modular data center 400.
As used in this context, "thermal circuit" refers to a path through
which heat flows. The thermal circuits 401 and 402 may operate to
remove heat that is generated inside enclosures 101A and 101B that
protect and enclose modules 100A and 100B of the data center 400
(e.g., modular computing environments). A first thermal circuit 401
may transfer heat from a first medium (e.g., air) to a second
medium (e.g., water) within the enclosure 101A; a second thermal
circuit 402 may remove heat from the second medium external to the
enclosure 101A.
[0065] As shown, the modular data center 400 includes two modular
computing environments 100A and 100B. (The modular data center 400
may include many other modular computing environments, but only two
are shown for purposes of illustration.) The modular computing
environments 100A and 100B include a number of computing devices
405 that generate heat within the enclosure 100A.
[0066] Air may be circulated past the computing devices 405 to
absorb the generated heat (e.g., via path D). The modular computing
environment also includes one or more integrated heat exchangers
116, through which the air may be circulated to remove the absorbed
heat (e.g., via path B). In some implementations, as described
above, the heat exchanger 116 is a liquid-air heat exchanger in
which heat removed from the air is transferred to a cooling fluid
circulated through the heat exchanger 116 (e.g., via supply and
return lines 122 and 125).
[0067] As shown, the transfer of heat from the computing devices
405 to air and from air to cooling fluid comprises a first thermal
circuit 401. In some implementations, the amount of heat generated
by the computing devices 405 is substantially equivalent to the
amount of heat transferred to the cooling fluid (ignoring heat
received or dissipated through walls of the enclosure 101A). In
these implementations, substantially all of the generated heat may
be removed by a second thermal circuit 402.
[0068] As shown, the second thermal circuit 402 includes the
cooling fluid supply line 122, a cooling fluid supply main 412 that
provides cooling fluid to multiple modular computing environments
100A and 100B, the cooling fluid return line 125, a cooling fluid
return main 415 that receives cooling fluid from multiple modular
computing environments 100A and 100B, and a cooling system 420
(e.g., a refrigeration-based chiller or condenser). In operation,
the second thermal circuit 402 may function at a facility level
(e.g., the entire modular data center 400) to exhaust heat from
multiple modules (e.g., the modular computing environments 100A and
100B). Other types and arrangements of the cooling system 420 are
possible.
[0069] FIG. 5 is another block diagram showing example thermal
circuits 501 that may exist in the modular computing environment
100. Various electrical circuits 501 generate heat. The
heat-generating circuits 501 may include, for example, processing
components (e.g., microprocessors, memory, storage devices, etc.)
within the modular computing environment 100 (in particular, within
the enclosure 101 of the modular computing environment 100). Heat
from the heat-generating circuits 501 may be conducted to a
corresponding circuit case or heat sink 503. For example, heat
generated within a microprocessor's circuits may be conducted to a
surface (case) of the microprocessor, or to a heat sink that is
attached to the microprocessor. Conduction is depicted in FIG. 5 by
a thin, straight arrow. From the case or heatsink ("case") 503, the
heat may be either radiated or conducted to a first cooling fluid
506. In FIG. 5, radiation is depicted as a wavy arrow. In some
implementations, the first cooling fluid is air. The first cooling
fluid 506 may be circulated within the enclosure 101 of the modular
computing environment 100, and as the first cooling fluid 506 is
circulated, the heat stored therein may be convected. Convection is
depicted in FIG. 5 by a larger, open arrow. In particular, the
first cooling fluid 506 may be circulated to a heat exchanger
(e.g., heat exchanger 116), and the heat may be transferred to the
body of the heat exchanger 509. By a process of conduction, the
heat may be transferred from the body of the heat exchanger 509 to
a second cooling fluid 512 within the heat exchanger, where it can
be removed, for example by a circulation process of the cooling
fluid to another heat exchanger that is external to the
enclosure.
[0070] In some implementations, heat may flow in other ways. For
example, some heat may be exhausted from the enclosure 101
corresponding to the modular computing environment 100 by
convection, e.g., through a fresh-air ventilation system to an
external environment 518. In many implementations, such heat
transfer has only second-order effects relative to the primary
transfer of heat to the second cooling fluid 512; that is, the
transfer of heat from the heat-generating circuits 501 to the
external environment 518 by first cooling fluid 506 that leaves the
enclosure may be negligible vis-a-vis the transfer of heat from the
heat-generating circuits to the second cooling fluid 512. As
another example, some heat may be transferred to the body 515 of
the enclosure 101 by convection, and then transferred from the body
515 of the enclosure 101 by conduction or radiation. In many
implementations, this form of heat transfer also only has
second-order effects relative to the primary transfer of heat to
the second cooling fluid 512; that is, the transfer of heat from
the heat-generating circuits 501 to the external environment 518 by
radiation from the body 515 of the enclosure may be negligible
vis-a-vis the transfer of heat from the heat-generating circuits to
the second cooling fluid 512.
[0071] For simplicity, the description above accounts for heat
transfer out of the enclosure 101. However, in some
implementations, heat may also enter the enclosure 101 from the
external environment. For example, heat may be transferred by
convection or radiation to the body 515 of the enclosure 101, and
from the body 515 of the enclosure 101 to the first cooling fluid
506. In many implementations, such heat transfer is also has only
second-order effects, and a detailed description is therefore
omitted.
[0072] FIG. 6 is a front view of an example rack 600. As shown, the
rack 600 includes a plurality of rails 603 configured to support
processor boards. The processor board 612 may be removed from the
rails 603 for service or anchored by the rails 603 for normal
operation.
[0073] In some implementations, a cable 609 (e.g., a Category 5,
Category 6, or optical fiber cable) may couple a network interface
on a process board to the patch panel 107 (see FIG. 1), a switch,
or the router 110, in order to make the processor board accessible
to the modular computing environment. In some implementations, the
cable 609 connects to the processor board via an RJ-45 plug 613 and
jack 614. In some implementations, the cable 609 may be routed
through a rack support 615 and across a cable tray 618 to the patch
panel 107, switch or router 110. The cable tray 618 may be disposed
along the bottom of the rack, as shown, or it may be disposed
elsewhere (e.g., under the walkway, along the ceiling, etc.) In
some implementations, as shown, the cable 609 may be continuous
from the processor board to the patch panel 107; in other
implementations (not shown), a short patch cable may connect each
processor board to another plug disposed on the rack support 615,
and another cable may connect the plug to the patch panel 107,
switch or router 110.
[0074] FIG. 7 is a rear view of the example rack 600 that is shown
in FIG. 6. FIG. 7 illustrates how power may be delivered to each
processor board, in some implementations. As shown, a power cable
702 is routed through a rack support 705 and connects to an
on-board power supply 708 with a plug (not shown) that is integral
to the power supply 708 and a receptacle 711.
[0075] In some implementations, the power cable 702 is connected to
a circuit that is split out in the power distribution box 131
(e.g., a 208 VAC, single-phase circuit). In some implementations,
power is distributed from the power distribution box 131 to various
rack supports along the top of the rack; in other implementations,
power may be distributed in a different manner (e.g., via buses,
along the bottom of the rack, under the walkway, etc.)
[0076] FIG. 8 is a block diagram of an example processor board 800
that may be included in the modular computing environment 100. As
shown, the processor board 800 may include one or more processors
802A and 802B, on-board memory 805, and one or more on-board
storage devices 808A and 808B. The processor board 800 also
includes an on-board power supply 811 that receives power from an
external source--for example as shown in FIG. 6. In some
implementations, the power supply 811 includes a fan that
facilitates cooling of the processor board 800 and that may help
minimize thermal inconsistencies elsewhere in an overall system. A
network interface 814 allows the processor board 800 to be coupled
to other devices within the modular computing environment 100 or
within a larger network.
[0077] In operation, the processor board 800 may execute computer
program instructions that are stored in the storage devices 808A or
808B or memory 805, or that are received via the network interface
814. In multi-processor implementations, each processor 802A or
802B may independently execute instructions, and each processor
802A or 802B may have corresponding dedicated memory or storage
devices. The instructions may, for example, cause the processors to
retrieve data from the storage devices 802A or 802B and provide it
to another device coupled to the network interface 814, process
data, or store data in the storage devices 802A or 802B or memory
805. A large data center may employ thousands of processor boards
such as the processor board 800 to perform complex data
manipulations, such as, for example, managing customer account
information within a financial institution, or providing various
users with content and information services.
[0078] FIG. 9 is perspective diagram of an example modular data
center 900 constructed with several modular computing environments
901-910. As shown, the modular computing environments 901-910 are
arranged along a connecting hub, such as, for example, a central
"spine" 913 or "backbone" of the modular data center 900. The spine
913 may route electrical power and cooling fluid to each modular
computing environment 901-910.
[0079] Electrical power may be provided to the modular data center
900 in the form of high voltage power (e.g., 13 kV) by a power
utility via utility lines 916. The voltage of the incoming power
may be reduced by a facility-level transformer 919 (e.g., to 480
VAC, three-phase) and distributed at a facility level power
distribution box 922 and routed to the various modular computing
environments within the spine 913. In some implementations, other
facility backup or supplemental power may be provided. For example,
a diesel backup generator (not shown) may be co-located with the
modular data center 900; if the utility power is compromised, the
diesel backup generator may provide interim power to the modular
data center.
[0080] A central cooling plant 925 may be provided at the facility
level to provide cooling fluid to each modular computing
environment 901-910 for the removal of heat generated within
corresponding enclosures of the modular computing environments. As
shown (and described above with reference to FIG. 4), the central
cooling plant 925 may be a chiller or condenser. In some
implementations, the central cooling plant 925 may be a cooling
tower. Cooling fluid may be distributed, within the spine 913, to
each modular computing environment 901-910.
[0081] A central communication facility 928 may also be provided at
the facility level and may couple the network interfaces of each
modular computing environment 901-910 to a larger network. An
example communication facility 928 is graphically depicted as a
satellite dish, but the communication facility could be any kind of
high-bandwidth network connection (e.g., OC-192, 10-gigabit
Ethernet, parallel T1 or T3 lines, etc.) In some implementations,
the communication facility 928 couples the modular data center 900
to a company's intranet (e.g., the internal network of a financial
institution or an Internet information provider); in some
implementations, the communication facility 928 couples the modular
data center 928 to the Internet.
[0082] In some implementations, the spine 913 provides a layer of
protection to the utilities (e.g., electrical power, cooling fluid
and a network interface) delivered to each modular computing
environment 901-910. The utilities may be delivered via flexible
"quick-connection" stubs disposed at "docking regions" (not shown)
corresponding to locations along the spine 913 that are configured
to receive modular computing environments. For example, electrical
power may be delivered from a distribution line or bus (not shown)
within the spine 913 to heavy-duty "appliance style" connectors
that can be plugged into a modular computing environment.
Similarly, cooling fluid may be delivered from large supply and
return mains (not shown in FIG. 9, but see elements 412 and 415 in
FIG. 4 for reference) to flexible hoses with threaded connectors
that can be quickly attached to a modular computing environment. A
network interface connection may also be provided by a connector
that can be quickly plugged into a modular computing
environment.
[0083] Deploying the modular data center 900 may be a simple
process. In some implementations, the process includes a) providing
(e.g., constructing) a facility level cooling plant 925; b)
providing a facility-level electrical power system 916, 919 and
922; c) providing a facility-level communication facility 928; d)
routing cooling fluid, electrical power and a connection to the
facility-level communication interface to various docking regions
(not shown), which may be optionally located in a spine 913; e)
providing two or more modular computing environments 901-910 at two
or more of the docking regions; and f) connecting the utilities to
the modular computing environments using the quick-connection stubs
disposed at the docking regions.
[0084] The example method of deploying the modular data center may
have one or more of the following advantages: modular data centers
may be quickly deployed; a modular data center 900 may be
constructed in parallel (e.g., modular computing environments may
be constructed in parallel with facility level resources); the
modular data center 900 may be easily expanded to incrementally add
computing power; large portions of the modular data center 900 may
be mobile to facilitate relocation; and portions of the modular
data center 900 may be prefabricated, possibly at lower cost than
would be possible if the modular data center 900 were serially
constructed from scratch (e.g., some manufacturing labor may be
used in place of trade labor).
[0085] Other implementations are contemplated. For example,
although ten modular computing environments 901-910 are shown, a
data center may have many more (e.g., 10-100 modular computing
environments); the modular computing environments 901-910 could
have other arrangements, such as a star configuration centered
around utility and network distribution points within a ring-shaped
connecting hub, or a stacked configuration coupled to a vertical
spine as illustrated in FIG. 10. The connecting hub need not
necessarily be either linear or circular or any other regular or
symmetric shape. In general, virtually any shape, size or
configuration of connecting hub can be used to accommodate
objectives of a system designer. As described above, backup or
supplemental generators may be provided. Cooling and communication
facilities may take forms other than those depicted.
[0086] FIG. 10 is a perspective diagram of another example modular
data center 1000 constructed with multiple modular computing
environments 1001-1008 and having a stacked configuration. Except
for configuration, the modular data center 1000 is very similar to
the modular data center 900: it includes a facility-level cooling
plant 1025, a facility-level electrical power source 1016; a
facility-level communication facility 1028; and the modular
computing environments 1001-1008. A vertical spine 1013 may route
the utilities to the modular computing environments 1001-1008, and
the facility-level cooling plant 1025 may be disposed above the
spine 1013. Facility-level power conversion equipment 1019 and
power distribution equipment (not shown) may be provided. FIG. 10
is an example of another possible data center configuration; as
described above, other configurations are contemplated. As one
additional example, a data center may be disposed onboard a ship;
electrical power may be provided by an on-ship generator, and the
cooling plant may incorporate seawater.
[0087] FIG. 11 is a block diagram of an example facility-level
electrical distribution system 1100 that may be included in a
modular data center. As shown, electrical power is provided by a
supply 1101 at a voltage V1. In some implementations, this power is
supplied by a utility company at a voltage V1 of about 13 kV. The
supply 1101 is converted by a power conversion device 1104 to a
voltage V2. In some implementations, the power conversion device
1104 is a transformer, and the voltage V2 is about 480 VAC, with
three phases provided. In some implementations, electrical power at
voltage V2 may be distributed to various modular computing
environments 1106A and 1106B. Current protection of the electrical
power at the voltage V2 may be provided by circuit protection
devices 1109 (e.g., circuit breakers, fuses, surge suppressors,
etc.)
[0088] In some implementations, each modular computing environment
1106A or 1106B includes a phase splitter 1112 that splits off each
of the three phases to single-phase legs having a voltage V3. In
some implementations, the voltage V3 is single-phase 208 VAC. The
three single-phase legs at voltage V3 may be distributed as various
circuits to loads within the modular computing environment 1106A or
1106B. For example, networking switching equipment 1115 may have a
dedicated circuit. A fan, or a number of fans 1118A and 1118B may
have a dedicated circuit (in some implementations, adjacent fans
may be supplied by different circuits for reliability). Lights 1121
may have a dedicated circuit. Groups of processing boards
1124A-1124C may have dedicated circuits (e.g., a column of
processor boards in a rack, or another portion of a rack 1125A,
1125B or 1125C may have a dedicated circuit). Other devices 1127
may also have dedicated circuits (e.g., lighting, fire/smoke
detection devices and a fire suppression system). As shown, each
circuit at the voltage V3 may have over-current devices 1130 (e.g.,
circuit breakers, fuses, surge suppressors, etc.). In some
implementations, each processor board 1124A-C includes an on-board
power supply 1133 that converts electrical power having voltage V3
to one or more voltages suitable for components on the
corresponding processor board.
[0089] Other implementations are contemplated. For example, at the
facility level, additional power sources, such as backup or
supplemental generators may be provided. Single-phase power may be
used throughout a facility in some implementations. Voltages other
than those described above may be employed.
[0090] FIG. 12 is a block diagram of an example information
provider 1201 and an environment 1200 in which the information
provider may function. In operation, the information provider 1201
may provide content and specific information services to client
devices (e.g., to the client device 1204) that access the
information provider through a network 1207, such as the Internet.
Information services may include a search engine function for
locating content that is accessible from the network 1207.
Information services may also include providing specific kinds of
information, such as time-sensitive news about particular topics
(e.g., business news, financial news, etc.), weather information
for a specific location, driving directions to navigate between two
locations, or traffic congestion information. The information
provider 1201 may access other content providers 1210A, 1210B and
1210C through the network 1207, and may index content stored on the
content providers 1210A, 1210B and 1210C, for example, in order to
provide a search engine service.
[0091] Various components of the example information provider 1201,
as shown, are now described. The information provider 1201 may
include content stores 1213A and 1213B, which may store specific
kinds of information, such as, for example, maps of various
locations, or images related to particular topics. The information
provider 1201 also includes information modules 1216A and 1216B,
which may each provide specific services such as determining
driving directions between two locations, or providing timely
weather information for a specific location. The information
provider 1201 includes a search engine 1219 that may be configured
to locate content that is internal or external to the information
provider 1201 (e.g., content in content stores 1213A or 1213B, or
content stored by the content providers 1210A-C).
[0092] To provide a search engine function, the information
provider 1201 may maintain an index 1222 (e.g., an index database)
and may further use system storage 1225 to maintain the index or
provide the search engine functions. The index database 1222 may
contain data that represents information that the information
provider 1201 provides to users. For example, the index database
1222 may contain links to information stored on content providers
1210A-C outside of the information provider 1201. Some information
referenced by entries in the index database 1222 may be stored in
the system storage 1225. For example, the system storage 1225 may
"mirror" information for which search reports are regularly
received such as, for example, breaking news stories or weather or
traffic information. The system storage 1225 may also store various
components needed for general operation of the information provider
1222, such as, applications, system parameters, and information
about users who access the system.
[0093] The index database 1222 may or may not be cached. For
example, the index database 1222 may include a separate cached
index database (not shown) to support faster access to search
results. The system storage 1225 may be local to the information
provider 1225, or it may be distributed, such as, for example, in
an external server or storage farm (not shown). Within the
information provider 1201, the search engine 1219 may operate to
produce search results that include information stored in the index
database 1222 or the system storage 1225, in response to search
requests from users.
[0094] As shown, the information provider 1201 receives and
transmits data through an external interface 1228 and an internal
interface 1231. Through the external interface 1228, which may
include one or more Web servers, the information provider 1201
receives requests and transmits responses from and to external
computing devices (e.g., the client device 1204). Through the
internal interface 1231, the information provider 1201 internally
routes information to and from the index database 1222 and the
system storage 1225 via various paths (some of which are not
shown), and to an from the information modules 1216A and 1216B and
the content stores 1213A and 1213B.
[0095] Each interface 1228 and 1231 may include one or more
distinct interfaces. For example, a first interface may receive
requests, such as, for example, requests for information or search
queries; a second interface may transmit results to requesting
devices. These interfaces may comprise interface devices for a
high-speed, high-bandwidth network such as SONET, Infiniband, or
Ethernet, or any suitable communication hardware operating under an
appropriate protocol such that the information provider 1201 can
respond to a large number of distinct requests simultaneously. The
precise design of the interfaces 1228 and 1231 or of the overall
information provider 1201 is not critical to this description and
may take any suitable form.
[0096] Information in the index database 1222 may be gathered by an
automated information gatherer 1234, such as, for example, a Web
crawler or a spider. The automated information gatherer may, for
example, continuously, or almost continuously, obtain new
information from sources connected to the network 1207 or to other
networks. The automated information gatherer 1234 may also retrieve
content in response to a search query. This content may be provided
to the index database 1222 or to the system storage 1225, or to
both. The index database 1222 may index data that is accessible
from the network 1207 or from another network.
[0097] In addition to being added to the index database 1222 or
system storage 1225 in an automated fashion, information may also
be manually loaded in or retrieved from the index database 1222 or
the system storage 1225 through a maintenance interface 1237. For
example, a maintenance interface 1237 may allow an administrator of
the information provider 1201 to manually add bulk data to the
index database 1222 or to the system storage 1225.
[0098] Data requests, such as queries received from users, may be
received and processed through a request processor 1240. The
request processor 1240 may, for example, parse requests, and, if
necessary, reformat them (e.g., from HTML (hypertext markup
language) or text format, to search term or strings that are
compatible with the search engine 1219). The request processor 1240
may operate in conjunction with the external interface 1228. In
some embodiments, the request processor 1240 may be included in the
external interface 1228.
[0099] Responses to data requests may be formatted for delivery by
a response formatter 1243. The response formatter 1243 may, for
example, format content that is responsive to data requests in a
format like HTML, XML (Extensible Markup Language), WML (Wireless
Markup Language), or some other suitable format. The response
formatter 1243 may operate in conjunction with the external
interface 1228. In some embodiments, the response formatter 1243
may be included in the external interface 1228.
[0100] The information provider 1201 may further include an ad
delivery system 1246, along with an ad database 1249 and an ad log
1252. The ad delivery system 1246 may receive query requests and
other input from the internal interface 1231. In some embodiments,
the ad delivery system 1246 may select an ad from the ad database
1249 to deliver in conjunction with other content, such as, for
example, a response to the query. The selected ad may correspond to
content of the search query. For example, in response to a user who
submits a query for a restaurant near a particular location having
a particular cuisine (e.g., That food), the information provider
may deliver, along with information about various restaurants
having that cuisine in that location, a coupon for a particular
restaurant. The response formatter 1243 may combine search results
from the search engine 1219 and ad results from the ad delivery
system 1246 into an overall response and may send that response to
the device from which the original query was receive, via the
network 1207. An ad maintenance module 1254 may be provided to add,
manage and remove ads in the ad database 1249
[0101] Some ads may solicit a response from the recipient. For
example, an ad may include a link that a user can select to receive
additional information or to perform other actions. Any such
responses may be directed back to the information provider 1201 and
subsequently logged. For example, the information provider 1201 may
receive, via the interfaces 1228 and 1231 and ad delivery system
1246, a response to a previously delivered ad. The ad delivery
system 1246 may store this response in the ad log 1252. The ad log
1252 may be subsequently used to bill sponsors of such ads or to
log various statistics related to ads and to their
effectiveness.
[0102] FIG. 13 is a block diagram 1300 illustrating one topology of
two modular computing environments 1301A and 1301B. As shown, each
modular computing environment 1301A and 1301B includes a plurality
of processing devices (e.g., processing device 1304, which may
include a processor board in a rack system) coupled to an internal
network (e.g., network 1307). The internal network 1307 may include
a patch panel, switch or router; and the internal network may
connect to another network 1310, that also connect to other modular
computing environments. Other devices may connect to the network
1310, such a client device 1313 or a server device 1313. In some
implementations, the resources of the modular computing
environments 1301A and 1301B may be accessed from either the client
device 1313 or the server device 1316. A topology similar to that
depicted in FIG. 13 may be used to implement various portions of
the information provider 1201 that is described with reference to
FIG. 12. For example, in some implementations, the modular
computing environments 1301A or 1301B may implement a portion of
the index database 1222, the search engine 1219, the content stores
1213A or 1213B, the request processor 1240, the information modules
1216A or 1216B, another component, or some combination of the
above-described components.
[0103] A number of embodiments of the invention have been
described. Nevertheless, it will be understood that various
modifications may be made without departing from the spirit and
scope of the invention. Accordingly, other embodiments are within
the scope of the following claims:
* * * * *
References