Nov 22, 2013 - In this case must be decided which server from outside the datacenter (another site) will allocate to that source server in cloud rapidly with ...
Effective Technique for Allocating Servers to Support Cloud using GPS and GIS Science and Information Conference (SAI), 2013 at 7-9 Oct. 2013
Dr. Ayad Ghany Ismaeel Professor Assistant, Computer Science Specialist Computer & Network Department of Information Systems Engineering- Erbil Technical College, Hawler Polytechnic University (previous FTE- Erbil), Iraq
2 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
Overview • Cloud must be provided services to customers in all situation (at any time) with high quality. • Cloud associated with market's applications cannot be delayed, postponed for another time, etc because that will lead to considerable material losses. • Cloud involve a group of servers (datacenters) may work as distributed system (like Google Apps engine, Amazon, etc) these servers can support each other rapidly and effective cost at peak time, failure, backup, etc to offer continuous services for clients. • A big problem appears when there is not idle server at the same datacenter, which has source server needs support.
3 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
Problem of research • There is no idle server at the same datacenter to support the source server, which needed support.
• In this case must be decided which server from outside the datacenter (another site) will allocate to that source server in cloud rapidly with effective cost. • This paper proposed a new technique for allocating servers which is idle relatively with a minimum cost to support the source server within cloud.
4 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
Motivating of Research • Overcome the shortages in the techniques which based on queuing theory to enhance the performance (throughput) of cloud servers. • Needed to reach a new technique for allocating servers to support cloud by allocate servers relatively idle to support source server using queue theory [2, 4], at the same time employs GIS with GPS techniques [5] using Haversine equation [6]. • To satisfy the follow: 1. Services to cover all users cloud. 2. Support another server within cloud, when needed that at peak times (to avoid reject/lost requests of clients). 3. The ability to replace the source server in case of failure or the backup, etc.
5 Effective Technique for Allocating Servers to Cloud using GPS & GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
PROPOSED OF EFFECTIVE TECHNIQUE FOR ALLOCATION SERVERS TO SUPPORT CLOUD
Reveals in the flowchart as shown in Figure.
11/22/2013
6 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
PROPOSED OF EFFECTIVE TECHNIQUE FOR ALLOCATION SERVERS ON TWO IMPORTANT TASKS: 1. Study Stead Queuing Model • The queuing model M/M/1, i.e. jobs enter the system according to an
BASE
independent Poisson and system with n=1 trunk (i.e. single server). Steady state queuing model which relates with these three variables L, λ, W as follow [2, 4]: L = λW …………………..…………. (1-a)
• The general and simplest setup is that the items arrive at an average Poisson process with rate of (λ) items per unit time. The items stay in the system an average of (W) units of time; finally, there is an average of (L) units in the queue system at any one time. Again can compute (L) as follow: L= λ / (µ- λ) ……………………………... (1-b) • The traffic intensity (busy/utilities) will define the following equation [2]: ρ=λ/µ .................................... (2)
• Where ρ is called the traffic intensity of queuing system, while: π0 = 1-ρ …….............………….…. (3) • Where π0 is called the %idle this formula will used to compute the rate of the idle server
7 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
PROPOSED OF EFFECTIVE TECHNIQUE FOR ALLOCATION SERVERS BASE ON TWO IMPORTANT TASKS: 2. Algorithm of Haversine Equation • It gives great-circle (shortest) distance between two points on a sphere from their longitudes and latitudes to compute the nearest idle server which play role to avoid the problems of network specially when the servers have large distance. The algorithm of Haversine equation shown as follow [1, 6]:
R: (mean radius = 6,371km)
8 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
EXPERIMENTAL RESULTS • Implement and applying the proposed technique need several important software and technologies as follow: 1. Microsoft visual studio 2010 Service pack 1. 2. C# language. 3. Google Maps API V3.1.
4. Microsoft SQL server 2008 R: Used to create the database called Allocation as a relational database contains 3 tables (Location, arrival service and Idle server), as shown in Fig. A, B and C respectively. 5. Language-Integrated Query (LINQ): to retrieve the data from Allocation database .
11/22/2013
9 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
EXPERIMENTAL RESULTS (CONT.) Suppose there are multiple datacenters distributed in the cloud and previously determined the Latitude and longitude, e.g. these datacenters are India-Asia (A) is the source server and the coordinates of that server is (e.g. 22.593726N, 78.398438E), China-Asia (B), Pakistan-Asia (C), GermanyEurope (D), Egypt- Africa (E) and ItalyEurope (F). As shown in Fig. No
Datacenter_ID
GPS Coordinate Latitude Longitude
Servers_ID
%idle
1
China-Asia
Server B
N37.1603170
E102.304688
65.9
2
Pakistan-Asia
Server C
N28.613459
E66.445313
60.4
3
Germany-Europe
Server D
N50.064192
E10.195313
55.6
4
Egypt-Africa
Server E
N27:99:44.01
E28:12:50.00
53.5
5
Italy-Europe
Server F
N44.263200
E11.440300
51.3
No.
Source Server
% idle
Distance in Km
1
Server A
Server B from China data center
65.9
7322
2
Server A
Server C from Pakistan data center
60.4
2129
3
Server A
Server D from Germany data center
55.6
8129
4
Server A
Server E from Egypt data center
53.5
5204
5
Server A
Server F from Italy data center
51.3
8520
Idle Server
Proposed Technique
Queuing Model
10 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
Discussion Feature
Proposed Technique of Allocation Servers to Support Cloud
Allocation of idle server by using
Queue theory for determining the idle% server and the shortest distance between the idle server and source server using Haversine equation
Using GPS and GIS Yes via Google Maps More an effective because addition to Results (selecting determine the idle server to support a the % idle server source server will select the nearest idle for supporting the server (economically means lowest cost source server) and avoiding the problems of network to satisfy fast reply) Shows more fast in finding the Using Google Maps distances between idle server and source server than GIS packages [5] Response to support the cloud
Higher relative to other techniques, because it used Haversine equation to compute the nearest idle server
Ayad Ghany Ismaeel and Suha Adham [2]
Michele Mazzucco, and others [3]
Queue theory only
Queue theory only
No
No
Based on the idle server only
Based on the idle server only
No, uses
No, uses
Do not bother to nearest distance between the idle sever and source server
Do not bother to nearest distance between the idle sever and source server
11 Effective Technique for Allocating Servers to Cloud using GPS and GIS. Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
Conclusions • The proposed technique for support cloud is more effective cost when it merges between more than one feature (idle server using Study Stead queuing model at the same time closer to source server (using Haversine algorithm). • This proposed technique is flexible, because it can find a list of servers which are idles (more than one option) outside the datacenter and at the same time select from it the nearest server (means this technique is economic). • The proposed technique for allocating (may allocate the lowest rate idle server) to support the source server, and that will not cause a big difference/changing because all those idle servers are supercomputer and with higher resource. • The proposed technique for allocating servers shows a high response relative to other techniques because it doing navigation to allocate the idle server which is closer to the source server, so the throughput and the performance better than other techniques.
12 Effective Technique for Allocating Servers to Cloud using GPS and GIS.. Dr Ayad Ghany Ismaeel, Professor Assistant in Computer Science
11/22/2013
References 1. Simulation of Haversine Formula URL : http://www.movabletype.co.uk/scripts/latlong.html 2. Ayad Ghany Ismaeel and Suha Adham Abdul-Rahman, " NEW METHOD OF MEASURING TCP PERFORMANCE OF IP NETWORK USING BIO-COMPUTING", International Journal of Distributed and Parallel Systems (IJDPS) Vol.3, No.3, May 2012, Pages 167-183. 3. Michele Mazzucco, Dmytro Dyachuk and Ralph Deters , "Maximizing Cloud Providers Revenues via Energy Aware Allocation Policies "arXiv:1102.3058v1 [cs.DC] 15 Feb 2011, http://www.eia.doe.gov/. 4. János Sztrik, "Basic Queueing Theory", University of Debrecen, Faculty of Informatics, publication Debrecen, 2012. URL: http://irh.inf.unideb.hu/~jsztrik/education/16/SOR_Main_Angol.pdf 5. Ayad Ghany Ismaeel, "New Technique for Proposing Network's Topology using GPS and GIS", International Journal of Distributed and Parallel Systems (IJDPS) Vol.3, No.2, March 2012, Pages 53-65. 6. Sinnott, R. W. "Virtues of the Haversine". Sky and Telescope, 68(2), 159. (1984).
13 Effective Technique for Allocating Servers to Cloud using GPS and GIS. 11/22/2013
Dr. Ayad Ghany Ismaeel, Professor Assistant in Computer Science Department of Information Systems Engineering- Erbil Technical College, Hawler Polytechnic University (previous FTE- Erbil), Kurdistan-Iraq.
