Nov 25, 2010 - Integrating Sensor Networks with Cloud Computing .... A security group should be defined that specifies what services can be accessed on.
Extending Sensor Networks into the Cloud using Amazon Web Services Dr. Kevin Lee, Murdoch University, Australia
25th November, 2010 IEEE International Conference on Networked Embedded Systems for Enterprise Applications 2010
Topic Overview The benefits and feasibility of Integrating Wireless Sensor Networks and Cloud Computing 7("&8&99$.&)96"$,&4:6";9$
$$ -G@&)9('&$"&96="C&$C694$ H)I&G(J8&$"&96="C&9$$
*6CDA6)$()>&@&)>&)4$ F&96="C&$"(C5$ H)&G@&)9('&$"&96="C&$C694$ -8D9AC$"&96="C&9$
75D4$(9$)&&>&>$ K6$J()>$45&$4:6$46B&45&"L$ 0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
0$
The Challenge 1 Integrating the Sensor Networks with Cloud Computing would mean: • Exposing Sensor Networks as a 1st class Cloud entity:
• -G@69()B$.&)96"$,&4:6";$"&96="C&9$D9$D$:&JM 9&"'(C&$ • • •
H)>('(>=D8$.&)96"9$ H)>('(>=D8$,6>&9$ .=JM)&49$6"$-)A"&$.&)96"$,&4:6";9$
• .5D"&>$DCC&99$46$.&)96"$,&4:6";$"&96="C&9$ • NDO$@&"$=9&$D)>$6)M>&?D)>$ 0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
P$
The Challenge 2 Integrating Sensor Networks with Cloud Computing would mean: • Giving Sensor networks the ability to seamlessly utilise cloud resources:
• Q)8(?(4&>$D)>$&8D9AC$946"DB&$ • Q)8(?(4&>$D)>$&8D9AC$@"6C&99()B$CD@DJ(8(4O$ • K5&$DJ(8(4O$46$@"6'(9(6)$"&96="C&9$R=(C;8O$:5&)$ D)$()C(>&)4$6CC="9$ • S6"$&GD?@8&+$D$"('&"MI6:$?6)(46"()B$D@@8(CDA6)$ CD)$R=(C;8O$@"6'(9(6)$?6"&$"&96="C&9$:5&)$D$ I66>$(9$>&4&C4&>$ 0345$,6'$0121$ !"#$%&'()$*&&+$,-.-/$0121$ T$
The Benefits Integrating Wireless Sensor Networks and Cloud Computing could enable: • • • •
dealing with dynamic loads reducing overall cost in deployment enabling the creation of new application types. Potential creation of a competitive market for sensor network providers
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
3$
Background: Sensor Networks (WSN) •
Provide a way to integrate environmental data into applications.
•
Composed of ‘motes’: embedded computers with low-power radios and simple sensors.
•
Motes form self-organizing networks that relay data to a small number of gateways.
Characteristics of Sensor Networks •
Hardware:
$6"$'("4=D8(X&>$"&96="C&9#$ -8D9AC$"&96="C&$CD@DC(4O#$ N"6B"D??DJ8&$9&8UM9&"'(C&$()4&"UDC$ Q9DB&MJD9&>$@"(C()B$?6>&8#$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
Y$
Characteristics of Cloud Computing Practically, Cloud offers:
•
• FD:$'("4=D8(X&>$"&96="C&9$ •
\H)U"D94"=C4="&$D9$D$.&"'(C&$HDD.]$
• W(B5M8&'&8$9=@@6"4&>$9&"'(C&9$ •
\N8D^6"?$D9$D$.&"'(C&]$
• $946"DB&$ •
• \.46"DB&$D9$D$.&"'(C&]$ A wide array of service types to support many types of application!
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
21$
The Vision
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
22$
Elastic Computational Capabilities for Environmental Modeling • 7&$J&8(&'&$$$J&$?694$ =9&U=8$()$C&"4D()$&)'("6)?&)4D8$D@@8(CDA6)9$ • $"&96="C&9$CD)$J&$ @"6J8&?DAC#$K5&$AB54MC6=@8()B$J&4:&&)$45&$D@@8(CDA6)$ D)>$45&$&)'("6)?&)4$>(9C6="DB&9$"&96="C&$D'D(8DJ(8(4O`$ • &"$D$S866>M?6)(46"()B$D@@8(CDA6)`$ • • • •
,6"?D88O+$86:$DCA'(4O+$86:$@"6C&99()B$"&R=("&?&)49+$A?&8()&99$6U$ I66>M@"&>(CA6)9$)64$C"(ACD8#$ Q)>&"$I66>$C6)>(A6)9+$I66>M@"&>(CA6)9$J&C6?&$C"(ACD8#$ Q9()B$4"D>(A6)D8$?&456>686B(&9+$"&96="C&9$:6=8>$J&$?D)=D88O$ @"6'(9(6)&>$D)>$?DO$)64$J&$DJ8&$46$"&9@6)>$UD94$&)6=B5#$ H)4&B"DA)B$$8O#$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
20$
Can Cloud Computing Resources Support the Dynamic resource provisioning needed by environmental applications?
1. How can this be realised? 2. Can we quantify the support for this?
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2P$
Realising Computational Elasticity in Amazon EC2 /?DX6)$-$?D)DB&$'("4=D8$?DC5()&9$6)$ /?DX6)a9$()U"D94"=C4="&$ K5"6=B5$:&JM9&"'(C&$/NH9+$46689$6"$D$:&JMJD9&>$ ?D)DB&?&)4$C6)968$ NDO$@&"M=9&+$D)>$U6"$45&$4O@&$6U$?DC5()&$J&()B$=9&>#$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2T$
Realising Computational Elasticity in Amazon EC2 K5&$@"6C&99$6U$8D=)C5()B$()94D)C&9$6)$-#$ ,6"?D8$D)>$?(C"6$()94D)C&9$U6"$D$'D"(&4O$6U$$V&?6"O$ C6)EB="DA6)9#$$ *D"B&$"D)B&$D'D(8DJ8&+$()C8=>()B$4569&$@"&@D"&>$JO`$
• /?DX6)$ • K5(">M@D"A&9$ • J=(84$JO$45&$=9&"+$U6"$&GD?@8&+$46$5694$D$9@&C(EC$?6>&8#$ 1.2*3!-45*956=8>$J&$C"&D4&>$45D4$D886:$9&C="&$C6))&CA6)9$46$J&$?D>&$46$45&$()94D)C$ '*5."$4-&2*04#$3*956=8>$J&$>&E)&>$45D4$9@&C(E&9$:5D4$9&"'(C&9$CD)$J&$DCC&99&>$6)$ 45&$()94D)C&9$D)>$:56$CD)$DCC&99$45&?#$ 6!$%",-%0$6U$45&$()94D)C&$4D;&9$@8DC$b)C&$45&$()94D)C&$(9$9&4=@$D$=9&"$CD)$C6))&C4$ 46$(4$\B#$=9()B$..W]$D)>$C6)EB="&$(4$U6"$45&$"&R=("&>$D@@8(CDA6)$9C&)D"(6#$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
23$
Using EC2: Selecting a machine
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2c$
Using EC2:Launching instances
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2d$
Using EC2:Monitoring running instances
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2[$
Amazon EC2 Web-service Interface • K5&$-$WKKN$CD889$$ :(45$956"4$eV*$>&9C"(@A6)9$6"$WKKN$f=&"(&9#$ K5&$?694$8(B54:&(B54$6U$45&9&$J&()B$WKKN$f=&"(&9$=9()B$ 45&$F-.KU=8$@D"D>(B?#$$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
2Y$
Amazon EC2 Web-service Interface • S6"$&GD?@8&+$46$8(94$D88$D'D(8DJ8&$QJ=)4=$/VH9$45&$ WKKN$f=&"O$:6=8>$J&$45&$U6886:()B`$ !"#$%%&'()*+*,-.*/0)'-+%1 23'4-.56&0'789&:+*;&01 5?9?.@?1 :(>45$ • :(>45$ $ C7&*'!6?7*4-.$(9$45&$D?6=)4$6U$A?&$45&$?&4"(C$(9$"&R=("&>$46$ J&$DJ6'&$45&$4"(BB&"$@6()4$J&U6"&$94D"A)B$D)645&"$()94D)C$
A trigger is configured to respond to an increase/decrease in utilization through the use of two parameters:
By adjusting these parameters:
• • •
/=46$9CD8()B$CD)$"&9@6)>$"D@(>8O$46$9?D88$C5D)B&9$ b"+$86)BM4&"?$4"&)>9$ !&@&)>&)4$6)$45&$=9&"a9$'D8=&$6U$@&"U6"?D)C&$'9$C694g$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
0P$
Using EC2 to support dynamism •
Example (1) defines a new autoscaling group with a maximum of 10 instances and (2) sets the scaling algorithm to increase the number of instances when the CPU utilization reaches 50% for more than 300 seconds.
http://autoscaling.amazonaws.com/ ?AutoScalingGroupName=sensorappn &LaunchConfigurationName=e480aa90 &MinSize=0 &MaxSize=10 &Cooldown=0 &AvailabilityZones.member.1=eu-west-1b &Action=CreateAutoScalingGroup http://autoscaling.amazonaws.com/ ?AutoScalingGroupName=sensorapn ?TriggerName=50percent &MeasureName=CPUUtilization &Statistic=Average &Namespace=AWS/EC2 &Period=50 &LowerThreshold=0 &LowerBreachScaleIncrement=-1 &UpperThreshold=50 &UpperBreachScaleIncrement=1 &BreachDuration=300 &Action=CreateOrUpdateScalingTrigger
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
0T$
Using EC2 to support dynamism $>D4D$D"&$()5&"&)48O$@D"D88&8$ D)>$:6=8>$45=9$J&)&E4$U"6?$=9()B$C86=>$"&96="C&9#$$ • 75&)$()C"&D9()B$45&$)=?J&"$6U$()94D)C&9$U6"$D)$D@@8(CDA6)$45&$C=""&)4$()94D)C&$D)>$94D4&$ (9$)64$"&@8(CD4&>+$()94&D>+$U"&95$/VH9$D"&$()94D)AD4&>$D)>$>D4D$>("&C4&>$46$45&?$=9()B$45&$ 86D>MJD8D)C()B$D8B6"(45?#$$ • K5&$/VH$?=94$J&$C=946?(9&>$46$9=@@6"4$45&$D@@8(CDA6)#$K5(9$CD)$J&$C6?@8&4&>$()$4:6$ :DO9#$ • O$46$"&C&('&$>D4D#$ • $DC"699$?=8A@8&$()94D)C&9i$ • ()C"&D9()B$D)>$>&9C"&D9()B$45&$)=?J&"$6U$()94D)C&9$D9$)&C&99D"O$
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
03$
Quantifying Elasticity of Amazon EC2 • •
To evaluate if EC2 can effectively: • 86D>MJD8D)C&$:5&)$9=Jj&C4&>$46$>O)D?(C$86D>9$ • *D=)C5$)&:$()94D)C&9$()$D$A?&8O$UD95(6)$ Experiment Setup • /$9()B8&$/VH$:D9$@"&@D"&>$=9()B$QJ=)4=$21#1Tk/@DC5&$ 0#0#2T#$$.46"&>$6)$/?DX6)$.(?@8&$.46"DB&$.&"'(C&$\.P]$ "&D>O$46$J&$()94D)AD4&>$JO$-$6)$D$ 94D)>D">$?DC5()&$4O@&$?2#8D"B&$:5(C5$5D9$T$cTMJ(4$-$(9$94D"4&>$D4$45&$8&'&8$6U$21$C6)C=""&)4$ "&R=&949$@&"$9&C6)>$ • 45&)$()C"&D9&>$&'&"O$21$?()=4&9$6'&"$[1$ ?()=4&9#$$ • K5&$8&'&8$6U$45&$86D>$D@@8(&>+$45&$DC4=D8$&8$N"6GO$?D)DB&9$M K(&"$9=@@6"4$U6"$45&$ /@@8(CDA6)$$ • $4&C5)(CD8$@D"4$ • ,&G4$94&@9$(9$46$&G@86"&$/@@8(CDA6)$.C&)D"(69$D)>$45&("$ &o&C4$6)$45&$V6>&8g$
•
•
•
Happy to hear your thoughts during the conference!
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
PT$
Summary Integrating Wireless Sensor Networks and Cloud Computing will enable: • dealing with dynamic loads • reducing overall cost of deployment • enabling the creation of new application types.
0345$,6'$0121$
!"#$%&'()$*&&+$,-.-/$0121$
P3$
Any Questions?