M’SILA University Information Communication Sciences and technology SUBJECT
I State Of The Art of Cloud Computing
1 What is cloud computing ?
2 Cloud computing architecture To understand the Cloud Computing architecture, we better adopt the description provided by NIST.
2.1 CHARACTERISTICS On-demand self-service : Resources provided any time and anywhere. Broad network access : resources available and accessed by a thick ( desktop, PC) or thin ( mobile phones, tablets, laptops, so on ). Resource pooling : clients are not able to know the exact physical location. Rapid elasticity : Users should be able to rapidly allocate and release capacity as required. Measured service : ( hours, days, months, years/ CPU, Memory, Storage ) as they are need.
2.2 DEPLOYMENT MODELS Community cloud: between two or more organizations Private Cloud: provided within a single organization.. Public Cloud: provided via the Internet, which can be accessed by any user. Hybrid Cloud: both public and private
2.3 DELIVERY MODELS 1 2 3
Virtualization Virtualization makes it possible to run multiple OS and multiple applications on the same server at the same time VMs are controlled by a layer of software called a Hypervisor. Hypervisor
Benefits using cloud No software required. Resources available on Network. On-demand-self service. Manipulate & configure apps online. Applications as service over internet. The cloud offer us several benefits, the most important are :
Service Broker Polices and load Balancing Algorithms II
II.1 Service Broker Polices Service broker controls the traffic routing between user bases (Clients) and data centers (DC). Closest data center CDC Optimize response time ORT DC CDC ORT
is a method to distribute workload on the multiple VMs for minimizing overall response time. II.2 II.2 Load Balancer Load Balancing
II.2 Load Balancing Algorithms Each MV is allotted with a time slice and has to wait for their turn. Round robin(RR) Equally spread current execution (ESCE) Equally distributed among all the VMs. ESCE
1.The client makes a request. 2.The load balancer scans the index table from top until the first VM Available if (Available) MV is allotted of the request client. else (All VMs are Busy) the request waits in the queue. 3. Back to step 2. Load Balancing Throttled
Analyst and Results III
Cloud-Analyst is useful tool to analyze cloud computing environment and applications. Designed on Java Swing, Melbourne University Australia. Main Screen Main Conf DC ConfAdvanced Conf
A:: Scenario1 Closest Data Center Scenario2 Optimize Response Time B:: Scenario3 Closest Data Center Scenario4 Optimize Response Time C: Scenario5 Closest Data Center Scenario6 Optimize Response Time RR ESCE TLB DC1 (20 VM) DC2 (50 VM) DC1 (8 VM) DC2 (20 VM) DC1 (2 VM) DC2 (5 VM)
Average response time a Singular Equilibrium Point manifests by breaking our intuition of performance at Conf B
CDC-ORT
Minimum response time
Maximum response time
Thanks a lot