presentation on reducing Cost in Cloud Computing


Published on

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

presentation on reducing Cost in Cloud Computing

  1. 1. Presentation onThe Cost of a Cloud in DataCenter Network
  2. 2. Presented To: Sir Uzair IshtiaqDepartment of Information Technology
  3. 3. Presented By:Farrukh Shahzad Muhammad Muhammad Ali Taha Rehman Faheem-ul-Hassan Roll# 09-13 Roll# 09-32 Roll# 09-11 Roll# 09-20 Department of Information Technology
  4. 4. Farrukh Shahzad Introduction Roll# 09-13 To The Cost of Cloud In Data Centers
  5. 5. In recent years Large investments have been made in massive datacenters supporting cloud services by companies such as eBay,Facebook.The investigation of the research paper starts with the question.“Where does the cost go in today’s cloud service data centers?”Here consider that a data centers housing on the order of50,000 servers that would be built based on currently well-understood techniques, using good quality, highly availableequipment.
  6. 6. TableAmortized Cost Components Sub Components ~45% Servers CPU, Memory, Storage System ~25% Infrastructure Power Distribution ~15% Power draw Electrical Utility Costs ~15% Network Links, Transfer, equipment
  7. 7. Cloud Service Data Centers are  It is natural to ask why existing solutions for the enterprise data center do not work for cloud service data centers.  First and foremost, the leading cost in the enterprise is operational staff. In the data center, such costs are so small (under5%Different due to automation) . In a well-run Enterprise , a typical ratio of IT staff members to servers is 1:100.
  8. 8. Cloud Service Data Centers are  In a well run datacenter, a typical ratio of staff members to servers is 1:1000.Different
  9. 9. Muhammad Server Cost Problem and AliRoll# 09-11 Solution
  10. 10. Cost Breakdown Server CostThe greatest data center costs go to servers. Forexample, assuming 50,000 servers , a relativelyaggressive price of $3000 per serverWith prices this high , achieving high utilization ,i.e. useful work accomplished per dollar invested,is an important goal . Unfortunately, utilizationin the data center can turn out to be remarkablylow near about 10%.
  11. 11. Structural Reasons For increasing Server CostUneven Application FitA server integrates CPU, memory, Uncertainty in Demand ForecastsNetwork and (often) storage Cloud service demands can spikecomponents. It is often the case quickly, especially for newthat The application fit in the services,server does not fully utilize one ormore of these components.
  12. 12. Structural Reasons For increasing Server Cost • Purchases, whether for upgrades or new builds, tend to be large, with components bought inLong Provisioning bulk. Infrastructure is typically meant to last very Time Scales long time periods • If successful, a service creator might reason, demand could ramp up beyond the capacity ofRisk Management the resources Allocated to the service ( and demand, as noted, can be hard to forecast).
  13. 13. Problem
  14. 14. Taha Rehman Infrastructure & Power cost Roll# 09-32 Problems and Solutions
  15. 15. Infrastructure Cost
  16. 16. As Table indicates, the aggregate cost is substantial.As depicted in Figure, Drawing power from theutility leads to capital investments in large Scalegenerators, transformers, and Uninterruptible PowerSupply (UPS) systems.
  17. 17. SolutionDriving the price of the infrastructure to these highlevels is the requirement for delivering consistentpower. resilienceled to scale-out data center designsbased on very large numbers of commodity , low costservers, with resilience in the system even though thecomponents have relatively high failure rates.
  18. 18. Power Cost ProblemTo track where the power goes, postulate application ofstate-of-the art practice based on currently well understoodtechniques and implementation based on good quality butwidely available equipment. The Green Grid provides ametric to describe data center Power Usage Efficiency(PUE) asPUE= (Total Facility Power)/(IT Equipment Power).
  19. 19. SolutionDecreasing the power draw of each server is clearly hasthe largest impact on the power cost of a data center,and it would additionally benefit infrastructure cost bydecreasing the need for infrastructure equipment. Thoseimprovements are most likely to come from hardwareinnovation, including use of high efficiency powersupplies and voltage regulation modules
  20. 20. Networking Cost ProblemsThe capital cost of networking gear for datacenters is a significant fraction of the cost ofnetworking, and is concentrated primarily inswitches , routers, and load balancers. Theremaining networking costs are concentrated inwide area networking:(1) peering, where traffic is handed off to the Internet Service Providers that deliver packets to end users,(2) the inter data center links carrying traffic between geographically distributed data centers
  21. 21. SolutionWide area networking costs are sensitive tosite selection, and to industry dynamics.Accordingly, clever design of peering andtransit strategies, combined with optimalplacement of micro and mega data centers,all have a role to play in reducing networkcosts.
  22. 22. MuhammadFaheem-ul-Hassan Agility to till Conclusion Roll# 09-20
  23. 23. Networking in Current Data Centers
  24. 24. Networking in Current Data Centers:Multiple applications run inside a single datacenter, typically with each application hosted onits own set of(potentially virtual) server machines.A single data center network supports two typesof traffic: (a) traffic flowing between external end systemsand internal servers(b) traffic flowing between internal servers.
  25. 25. Networking in Current Data Centers This spreading is typically performed by a specialized hardware load balancer. Using conventional load-balancer terminology, the IP address to which requests are sent is called a virtual IP address(VIP) and the IP addresses of the servers over which the requests are spread are known as direct IP addresses(DIPs).
  26. 26. Design Objectives In order to achieve agility within a data center, we argue the network should have the following properties:Services should use location independentaddresses that decouple the server’s location inthe DC from its address. This enables any serverto become part of any server pool whilesimplifying configuration management.
  27. 27. Uniform Bandwidth and Latency: Uniform bandwidth, combined with uniformlatency between any two servers would allowservices to achieve same performanceregardless of the location of their servers. Security and Performance Isolation: If any server can become part of anyservice, then it is important that services aresufficiently isolated from each other that oneservice cannot impact the performance andavailability of another.
  28. 28. INCENTING DESIRABLE BEHAVIORDesigning mechanisms to implementeconomic incentives that encourage efficientbehavior is a rich area for study and impact.Trough Filling:Periods of peak usage of net work and powerare relatively expensive to a data center bothresources are typically charged based onpercentiles of usage, meaning that the cost isdetermined by the height of the peaks“Binpacking” opportunities to manage services
  29. 29. GEO-DISTRIBUTION Speed and latency matter. There is substantial empirical evidence suggesting that performance directly impacts revenue.Geo-D st r i but i ng St at e iThe state-of-the art is that every service implements its ownsolution for geo distribution. For example, Facebook replicatesdata with all writes going through a single master data center.Yahoo! mail partitions data across DCs based on user
  30. 30. CONCLUSIONSData center costs are concentrated in servers,infrastructure, power requirements, and networking, in thatorder .Though costs are steep, utilization can beremarkably low.First, we need to increase internal data center networkagility, to fight resource fragmentation and to get morework out of fewer servers reducing costs across theboard.Second, we need to pursue the design of algorithms andmarket mechanisms for resource consumption shapingthat improve data center efficiency.Finally, geo diversifying data centers can improve end toend performance and increase reliability. To reapeconomic benefits from geo diversity, we need to designand manage data center and network resources as a jointoptimization, and we need new systems to manage the
  31. 31. Questions