Ahmed Alharthi April 25, 2014 UCCS

Ahmed- Public Cloud Location in Africa

1

Ahmed AlharthiApril 25, 2014

UCCSMaster’s Project Report

4/25/2014

Methodology for Choosing Datacenter for Cloud Providers in West Africa Region


2

• Introduction

• Why Africa

• The proposed Methodology

• Selection Results

• Lessons Learned

• Future Work

• Conclusion

4/25/2014

Agenda


3

Research and process for selecting the best location to build data centers for cloud providers in West Africa.

Main Considerations: Network performance, Costs, Submarine cables

Other factors : Environmental, Economical and Politics stabilities.

Introduction

4/25/2014


4

Lack of information regarding to data centers in Africa.

Demand of the services

Globalizations moment

Cloud provider numbers

4/25/2014

Problems


54/25/2014

Cloud providers


6

McConnell A., Parr G.P., Mc Clean S.I., Khan, K [38]

Khosrow Ebrahimi, Gerard F. Jones, and Amy S. Fleischer [11]

In terms of Africa’s network connectivity, M. Zennaro, E. Canessa, K.R. Sreenivasan , A.A. Rehmatullah and R.L. [12]

Sanjay M. Parekh, Robert B. Friedman, Neal K. Tibrewala, and Benjamin Lutch [28]

4/25/2014

Related Work


7

Number of data centers

Why Africa ?

4/25/2014


84/25/2014

Number of Data Centers around the World


9

Growing demand

4/25/2014

Growing Demand from Africa


104/25/2014

Africa Submarine Cables


11

Geographical & Climate Characteristics of the region

Climate & topography of the continent Tornados & Hurricanes & Tsunami Earthquake Politic Systems on the region Economic Status

4/25/2014

Why Africa ?


12

Intel Considerations are : WAN infrastructure availability of water for cooling Also, costs

workforce, water, power or landscape.

HP Considerations are : size, price and geographical quality of the land, the availability of power and communication network in the region. Cost of Operations [ electricity, water, network access, Free cooling]

4/25/2014

How Intel/HP Select their Data Center Locations


13

Methodology is mix of Intel , HP and IBM approaches

Main considerations :◦ Costs of Powering, Cooling and landscape. ◦ Communication network and their capabilities.Secondary considerations :

Politics status Geographical & Climate location Submarine cables Economy

4/25/2014

The Proposed Methodology


144/25/2014

Decision Process for Selecting Countries to Host Public Cloud

Senegal, Guinea, Liberia, Ghana, Togo, Benin, Sierra Leone, Nigeria, Cameron, Cote d'ivoire, Congo, Angola, Democratic Republic of the Congo and Namibia.


154/25/2014



16

Network measurements

◦ Latency ◦ Round Trip Time◦ Packet loss◦ Bandwidth

4/25/2014



17

Network measurements

Amazon Linux AMI 2014.03.1 machines In Singapore, Sydney and Sao PauloLinux commands : Ping and Traceroute

Tool used : Nmap ( Network mapper) Visual route

4/25/2014

System and Tools for Measurement


18

<where in North America?>

4/25/2014

Network measurements from North America

Ghana South Africa Senegal Nigeria 0

50

100

150

200

250

300

23 27 30 27

220

265

182 182

Latecny RTT

Country Latency Round trip time Packet loss Number of Hops

Ghana 197.253.65.36 23 ms 220 ms 10 % 17

South Africa 41.74.178.82 27 ms 265 ms 0% 20

Senegal 41.83.255.255 30 ms 182 ms 0% 17

Nigeria 41.222.211.58 27 ms 182 ms 0% 18


194/25/2014

Network measurements from East Asia (Singapore)

Country Latency Round trip time

Packet loss Number of Hops

Ghana 197.253.65.36 28 ms 270 ms 0% 12South Africa 41.74.178.82 48 ms 476 ms 0% 20Senegal 41.83.255.255 24 ms 236 ms 0% 13Nigeria 41.222.211.58 27 ms 274 ms 0% 19


100150200250300350400450500

28 4824 27

270

476

236274

RTT Latency


204/25/2014

Network measurements from East Asia (Sydney)

Country Latency Round trip time Packet loss Number of Hops

Ghana 197.253.65.36 44 ms 440 ms 0% 17South Africa 41.74.178.82 44 ms 493 ms 0% 25Senegal 41.83.255.255 42 ms 428 ms 0% 17Nigeria 41.222.211.5 42 ms 448 ms 0% 23


100

200

300

400

500

600

44 44 42 42

440493

428 448

Latency RTT


214/25/2014

Network measurements from South America (Brazil)

Country Latency Round trip time Packet loss Number of HopsGhana 197.253.65.36 30 ms 312 ms 0% 14South Africa 41.74.178.82 36 ms 343 ms 0% 16Senegal 41.83.255.255 31 ms 281 ms 0% 14Nigeria 41.222.211.58 32 ms 297 ms 0% 12

Ghana South Africa 4 Senegal Nigeria 0

50

100

150

200

250

300

350

400

30 36 31 32

312343

281 297

Latency RTT


22

[aalharth@walrus cs701]$ ping -c 10 197.253.65.36 PING 197.253.65.36 (197.253.65.36) 56(84) bytes of data. 64 bytes from 197.253.65.36: icmp_seq=1 ttl=43 time=981

ms 64 bytes from 197.253.65.36: icmp_seq=2 ttl=43 time=983








ms --- 197.253.65.36 ping statistics --- 10 packets transmitted, 9 received, 10% packet loss, time

8997ms rtt min/avg/max/mdev = 960.481/975.940/1003.880/11.900

ms, pipe 2

4/25/2014

Detailed Stat Related to Ghana Ghana Observation

core4.LDN-London.as6453.net

Goes to WASC submarine cables

TATA


23

Estimating the Coast of Space, Cooling and Powering Data center

Estimation based on HP “Cost model for planning, development and operation of a data center”

Assumptions Facility size is 30000 sq ft 10 MW datacenter

4/25/2014

Formula for Estimating Costs


24

Cost of Space

4/25/2014

Cost of Renting Spaces

Country Average rent (per sq ft)

Lagos, Nigeria

$6.48 – 8.00

Accra, Ghana

$5.7-10.5

Dakar, Senegal

$ 5.5 – 9.5

Cape Town, South Africa

$ 10 - 14

Lagos, Nigeria Accra, Ghana Dakar, Senegal Cape Town, SA

$36 $41

$38

$60

Space Cost / sqft


254/25/2014

Cost of Power Delivery

Country

Rate/sqft / Unit

Lagos, Nigeria

$ 0.072/ft

Accra, Ghana

$ 0.115 /ft

Dakar, Senegal

$0.151 /ft

Cape Town, SA

$ 0.172


$0.07

$0.12

$0.15

$0.17

Cost of Power /sqft


264/25/2014

Cost of Cooling

Country

Rate / sqft

Lagos, Nigeria

$0.017/ft

Accra, Ghana

0.030/ft

Dakar, Senegal

0.040/ft

Cape Town, SA

0.047/ft


$0.02

$0.03

$0.04

$0.05

Cooling Cost / sqft


274/25/2014

Total Cost


$36.44

$41.15 $37.44

$60.22

Total Cost /sq ft


28

Country Nigeria Senegal South Africa

Ghana

North America

1 1 4 3

Singapore 2 1 4 3

Sydney 2 1 4 3

Brazil 2 1 4 3

4/25/2014

Selection ResultsRanking the countries based on RTT


29

Bandwidth

4/25/2014

Selection Results

Ghana Nigieria Senegal South Africa0

50

100

150

200

250

300

MaxMin

Country

Band

wid

th K

B/s

X


30

Total Space, Power and Cooling costs

4/25/2014

Selection Results


$36.44 $41.15

$37.44

$60.22

Total Cost /sq ft


31

Difficults to find … data The real estate are eager to help. …

4/25/2014

Lessons Learned


32

Relevant criteria to select the best location of data center for a cloud provider should be identified.

The data gathered from latency, round trip time and bandwidth should be gained from relevant data center of a cloud provider in real world.

Review Ping and trace utility data multiple time in different times.

4/25/2014

Future Work


33

Senegal, Nigeria, Ghana and South Africa are the ideal locations to host data center for cloud provider.

Nigeria has the most suitable features to host one. Africa has a good opportunity to host datacenter for

a cloud provide. Costs of datacenter may be vary from city to other

one.

4/25/2014

Conclusion


34

Amazon Web Services. (2014). Global Infrastructure. Retrieved from http://aws.amazon.com/about-aws/globalinfrastructure/ Rath, John. DATA CENTER SITE SELECTION. Tech. N.p.: n.p., n.d. Print. "BlackSwan Zine." Data Center: 10 Places You Don’t Want to Build. Blackswanzine, 2010. Web. 2 Jan. 2014. [4] Chang, Shin-Jyh Frank, Susmit Harihar Patel, and James Marc Withers. "An optimization model to determine data center

locations for the army enterprise."Military Communications Conference, 2007. MILCOM 2007. IEEE. IEEE, 2007. [5] Hugos, M. H., & Hulitzky, D. (2011). Business in the cloud: What every business needs to know about cloud computing.

New York: Wiley. [6] Lanier, J. (2013). Who owns the future?. London: Allen Lane. [7] Wang, L. (2012). Cloud computing: Methodology, systems, and applications. Boca Raton, FL: CRC Press. [8] Agrawal, G. P. (2010). Fiber-optic communication systems (Vol. 222). John Wiley & Sons. [9] "How to Meet a Growing Need for More Data Center Capacity."Http://www.ibm.com/us/en/. IBM, Sept. 2012. Web. Oct.

2013. [10] Nozick, Linda K., and Mark A. Turnquist. "A two-echelon inventory allocation and distribution center location

analysis." Transportation Research Part E: Logistics and Transportation Review 37.6 (2001): 425-441. [11] Ebrahimi, Khosrow, Gerard F. Jones, and Amy S. Fleischer. "A review of data center cooling technology, operating

conditions and the corresponding low-grade waste heat recovery opportunities." Renewable and Sustainable Energy Reviews 31 (2014): 622-638.

[12] Zennaro, M., et al. "Scientific Measure of Africa's Connectivity." Information Technologies & International Development 3.1 (2006).

[13] Nozick, Linda K., and Mark A. Turnquist. "A two-echelon inventory allocation and distribution center location analysis." Transportation Research Part E: Logistics and Transportation Review 37.6 (2001): 425-441.

[14] Cloud Computing in Africa Situation and Perspectives. Tech. ITU, 20 Apr. 2012. Web. Feb.-Mar. 2014. [15] Data Center Map." - Colocation, Cloud, Managed Hosting Etc. N.p., n.d. Web. Mar. 2014.

<http://www.datacentermap.com/>. [16] Submarine Telecoms INDUSTRY REPORT. Rep. Terabit Consulting, 2013. Web. [17] African Studies Center, University of Pennsylvania. African Studies Center, University of Pennsylvania. Web. 15 Feb.

2014. [18] Pappa, Farrah. "The Do's and Dont's of Data Center Planning." The Do's and Dont's of Data Center Planning. Web. 15

Feb. 2014. [19] Map of Earthquakes Today." Map of Earthquakes Today. N.p., n.d. Web. 15 Mar. 2014. [20] Earthquakes .Web. Feb. 2014. <Earthquake Hazards Program>.

4/25/2014

References


35

[21] P&S. "Tsunami Strikes African Coast." Tsunami Strikes African Coast. N.p., n.d. Web. 15 Mar. 2014. [22] "Nigeria Economy Set to Leapfrog South Africa on Data Revamp."Bloomberg.com. Bloomberg, 7 May

2014. Web. 11 Apr. 2014. [23] Nigeria. Official Website. Web. Feb. 2014. <http://www.nigeria.gov.ng/index.php>. [24] "Selecting A Data Center Site: Intel's Approach." N.p., Feb. 2014. Web. Feb. 2014.

<http://connectedsocialmedia.com/intel/7426/it-best-practices-data-center-site-selection/> [25] Selecting a Location for an HPC Data Centre. Tech. N.p.: n.p., 2013. Web.

<http://webcache.googleusercontent.com/search?q=cache:http://www.prace-project.eu/IMG/pdf/hpc-centre-location-whitepaper-2.pdf>.

[26] "Africa’s International Bandwidth Growth to Lead the World." Africa's International Bandwidth Growth to Lead the World. N.p., n.d. Web. 29 Mar. 2014.

[27] Hurricanes VS Tornadoes. Diffen, n.d. Web. Mar. 2014. [28] Parekh, Sanjay M., et al. "Systems and methods for determining collecting and using geographic

locations of Internet users." U.S. Patent No. 6,757,740. 29 Jun. 2004. [29] Nmap Free Security Scannar. Web. Jan. 2014. <http://nmap.org/>. [30] TATA, n.d. Web. Apr. 2014. <http://lg.as6453.net/bin/lg.cgi>. [31] Patel, Chandrakant D., and Amip J. Shah. "Cost model for planning, development and operation of a data

center." (2005). [32] "Nigeria Becomes Africa's Biggest Economy." BBC. N.p., 06 Apr. 2014. Web. 06 Apr. 2014.

<www.bbc.com%2Fnews%2Fbusiness-26913497%E2%80%8E>. [33] "Cloud Uptake Set to Explode Across Africa." Cisco. Web. 26 Mar. 2014.

<http://www.cisco.com/web/ZA/press/2013/112813.html>. [34] Hanson, Stephanie. "Nigeria’s Creaky Political System." Council on Foreign Relations, Apr. 2007. Web.

<http://www.cfr.org/nigeria/nigerias-creaky-political-system/p13079>. [35] "Top 10 Internet Countries in Africa." PC Tech Magazine. N.p., Feb. 2014. Web. Apr. 2014.

<http://pctechmag.com/2014/02/top-10-internet-countries-in-africa/>. [36] "Political System." Wikipedia. Web. Mar. 2014. <http://en.wikipedia.org/wiki/Political_system>. [37] "GLOBAL BROADBAND." Netindex. N.p., n.d. Web. Feb. 2014. <www.netindex.com>. [38] McConnell, A., et al. "Selecting a Data C based Location."

4/25/2014

References

Ahmed Alharthi April 25, 2014 UCCS

Documents

public cloud senegal

best location

ahmed alharthiapril

communication network

africa11why africa

network access

network performance

intel considerations