a scalable and distributed electrical power monitoring system utilizing cloud computing

19
A Scalable and Distributed Electrical Power Monitoring System U:lizing Cloud Compu:ng Ryousei Takano , Hidemoto Nakada, Toshiyuki Shimizu, Tomohiro Kudoh Informa(on Technology Research Ins(tute, Na(onal Ins(tute of Advanced Industrial Science and Technology (AIST), Japan CUTE2013@Da Nang, Vietnam, Dec. 19 2013

Upload: ryousei-takano

Post on 12-May-2015

422 views

Category:

Technology


0 download

DESCRIPTION

CUTE 2013

TRANSCRIPT

Page 1: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

A  Scalable  and  Distributed    Electrical  Power  Monitoring  System  

 U:lizing  Cloud  Compu:ng  

Ryousei  Takano,  Hidemoto  Nakada,  Toshiyuki  Shimizu,  Tomohiro  Kudoh  

 Informa(on  Technology  Research  Ins(tute,    

Na(onal  Ins(tute  of  Advanced  Industrial  Science  and  Technology  (AIST),  Japan  

CUTE2013@Da  Nang,  Vietnam,  Dec.  19  2013  

Page 2: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Outline  

• Background  • System  overview  • Design  of  hardware  and  soKware  • Deployment  at  AIST  campus  • Summary  

2  

Page 3: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Background  

3  

•  The  power  consumpPon  of  data  centers  and  networks  becomes  an  issue  of  vital  importance  to    IT  industries.  

 •  In  Japan,  there  is  an  urgent  need  for  green  IT  technologies  aKer  March  11th,  2011.  

Fukushima  Daiichi  Nuclear  Power  Plant  

Google  data  center  in  the  Dalles,  Oregon    

Page 4: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Mo:va:on  •  VisualizaPon  is  key  to  plan  power  savings.  •  The  total  system  cost  and  scalability  are  problem.  

–  Our  server  room  has  over  100  racks.  –  Our  campus  are  geographically  distributed  in  Japan.  

•  The  system  has  to  be  low-­‐cost,  scalable,  and  also  ease  to  develop  applicaPons.  ➡ Cheap  power  monitoring  hardware  ➡ Power  monitoring  soKware  uPlizing  cloud  compuPng  ➡ A  simple  REST  API  

4  

Page 5: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

System  overview  

5  

Data collecting unit

Viewer

Alert

Google App Engine

Data store

Update power usage using REST w/ JSON Retrieve data using

REST w/ JSON

Data collecting unit

1

2

3

4

Power measuring unit

Data collecting unit

Applications

Data store

Observe the state of power consumption Plan power saving

Page 6: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Small  start  Go  big  

6  

4  sensors   32  ports   =  128  sensors  

Google  App  Engine  

Data  store  

...  

Google  App  Engine  

Data  store  

Sensors  can  be  incrementally  installed.  

Page 7: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Low-­‐cost  power  measuring  unit    

7  

RJ-­‐45  port  

Signal  processing  board  (dsPIC30F3013)  

Clamp-­‐on  current  sensor  (max:  4)  

•  Send  data  to  data  collecPng  unit  every  second.  •  The  producPon  cost  is  approximately  120  USD,  including  the  cost  of  4  current  sensors.  

Page 8: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Data  collec:ng  unit  (1/2)  

8  

•  Gather  data  from  up  to  32  power  measuring  unit  •  Push  data  to  GAE  

–  Can  be  placed  behind  NAT  

To  power  measuring  unit  (Not  Ethernet,  data  transfer  and  power  supply)  

To  GAE  via  the  Internet  (Ethernet/100BaseT)  

Page 9: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Data  collec:ng  unit  (2/2)  

9  

RJ-45 ports x 32

Power Measuring Unit

CPU board (T-SH7706LSR) -  SH3 Linux -  Buildroot 2011.05 -  pmon.py Serial/Parallel converter

(Xilinx Spartan-3E)

data store

Page 10: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Data  store  and  REST  protocol  

10  

pmon.py pmon.py

pmon.py Application

Application

config DB

power logging DB Configuration

Update

Get, query

Power measuring unit “sensor”

Current sensor “probe”

Data collecting unit “unit”

data store

Page 11: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Google  App  Engine  

•  PaaS  cloud  service  for  web  applicaPons  –  Java,  Python,  and  Go  are  supported  –  Your  applicaPon  will  have  URL  like    hgp://XXXX.appspot.com  

•  Scalable  and  stable  data  storage  –  Data  are  replicated  to  5  different  datacenters  –  Allows  2  of  them  to  be  lost  during  operaPon  

•  Maintenance  free  –  No  need  to  manage,  almost  

•  Cost  effecPve  –  Almost  free  of  charge  

11  

Page 12: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

REST  API  

12  

path   method   descrip:on  

/update   POST   Upload  data  

/latest   GET   Get  all  data  for  the  last  minute  

/latest,N   GET   Get  all  data  for  the  last  N  minutes  

/summary.s/YYYYmmDDHHMMSS,N   GET   Get  all  data  for  each  second  start  from  YYYYmmDDHHMMSS,  for  N  seconds  

/summary.m/YYYYmmDDHHMM,N   GET   Get  all  data  for  each  minute  start  from  YYYYmmDDHHMM,  for  N  minutes  

/summary.h/YYYYmmDDHH,N   GET   Get  all  data  for  each  hour  start  from  YYYYmmDDHH,  for  N  hours  

/summary.d/YYYYmmDD,N   GET   Get  all  data  for  each  day  start  from  YYYYmmDD,  for  N  days  

/query.s/LOC/YYYYmmDDHHMMSS,N   GET   Get  data  for  locaPons  that  name  start  with  LOC  

/query.m/LOC/YYYYmmDDHHMM,N   GET  

/unit-­‐config/UNIT_ID   GET   Get  configuraPon  data  

/unit-­‐config/UNIT_ID   PUT   Set  configuraPon  data  

Page 13: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Update  from  Data  collec:ng  unit  

Each  data  collecPon  unit  sends  data  every  20  seconds  

–  POST  the  following  JSON  string  

13  

{            "id":      "UNIT_ID"            "Pme":  "1319837460”    /*  elapsed  seconds  from  the  UNIX  epoch  Pme  */            "power":    {                                      /*  data  for  the  last  20  seconds  per  measurement  point  */                  "sensor0.0":  [VAL0,  VAL1,  VAL2,  VA3,  ...,  VAL19],                  "sensor0.1":  [VAL0,  VAL1,  VAL2,  VA3,  ...,  VAL19],                  "sensor1.0":  [VAL0,  VAL1,  VAL2,  VA3,  ...,  VAL19],                  ....            }  }  

GAE  xxx.appspot.com/update  

Page 14: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Data  retrieval  

An  applicaPon  periodically  (e.g.,  1  min)  gets  data  from  GAE  

–  GET  the  following  JSON  string  

14  

GAE  xxx.appspot.com/latest,N  

{          "Pme":  “1319837460”  /*  epoch  Pme  */          "PmeStr":  “201110290631”  /*  human  readable  Pme  in  JST  */          "power":  {                  "LOCATION0":  [1234]                  "LOCATION1":  [1234]                  "LOCATION2":  [1234]                  "LOCATION3":  [1234]                  "LOCATION4":  [1234]                  ...          }  }  

Viewer  applicaPon  

Page 15: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Deployment  at  AIST  campus  

15  

…  

…  

Server  room  

Data  collecPng  unit  

10  data  collecPng  units  177  power  measuring  units  620  measurement  points  

Sensor  module  Clamp-­‐on  current  transformer  

Data  collecPon  unit  

Clean  room  

Viewer  

billing  service  

Google  App  Engine  

Datastore  

Update  power  usage  using  REST  w/  JSON     Retrieve  data  using  

REST  w/  JSON  

Data  collecPng  unit  

1

1

2

2 3

1

2

3

gather  

store  

view  

Page 16: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Installa:on  in  AIST  server  room  

16  

Power measuring unit

Clamp-on current sensor

Power distribution board

data store

Data collecting unit in free access floor

Page 17: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Summary  •  Our  proposed  system  helps  reduce  total  system  cost  and  

improve  scalability  by  employing  low-­‐cost  power  measuring  units  (30  USD  per  measurement  point),  and  uPlizing  cloud  compuPng.  –  The  development  of  the  system  was  completed  within  3  months.  –  We  have  successfully  operated  it  over  2  years  for  to  provide  an  

electricity  billing  service  and  evaluate  the  power  efficiency  of  data  processing  middleware.  

•  Future  work  –  Tolerance  for  network  failures.  –  More  applicaPons:  e.g.,  server  consolidaPon  on  a  private  cloud  to  

reduce  power  consumpPon.  

17  

Page 18: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Q&A  

18  

This  research  was  parPally  supported  by  the  NEDO  research  project  enPtled  “Research  and  Development  Project  for  Green  Network/System  Technology  (Green  IT  Project).”  

Thanks  for  your  agenPon!  

Page 19: A Scalable and Distributed Electrical Power Monitoring System Utilizing Cloud Computing

Visualiza:on  Applica:ons  

19  

less  than  90%  of  the  upper  limit  less  than  95%  of  the  upper  limit  more  than  95%  of  the  upper  limit  Offline  

(a)  Web  applicaPon   (b)  Desktop  applicaPon