A View of Cloud Computing- Armbrust et al. (5. Top 10 Obstacles and …
A View of Cloud Computing- Armbrust et al.
Developers with innovative ideas for new Internet services no longer require the large capital outlays in hardware to deploy their service or the human expense to operate it.
Using 1,000 servers for one hour costs no more than using one server for 1,000 hours
Defining Cloud Computing
Cloud computing refers to both the applications delivered as services over the Internet and the hardware and systems software in the data centers that provide those services.
The services themselves have long been referred to as Software as a Service (SaaS).a Some vendors use terms such as IaaS (Infra-structure as a Service) and PaaS (Plat-form as a Service) to describe their products, but we eschew these because accepted definitions for them still vary widely.
Similarly, the related term “grid computing,” from the high-performance computing community, suggests protocols to offer shared computation and storage over long distances, but those protocols did not lead to a software environment that grew beyond its community
The data center hardware and soft-ware is what we will call a cloud.
When a cloud is made available in a pay-as-you-go manner to the general public, we call it a public cloud; the service be-ing sold is utility computing. We use the term private cloud to refer to internal data centers of a business or other or-ganization, not made available to the general public, when they are large enough to benefit from the advantages of cloud computing that we discuss here
cloud computing is the sum of SaaS and utility computing, but does not include small or medium-sized data centers, even if these rely on virtualization for management.
People can be users or providers of SaaS, or us-ers or providers of utility computing. We focus on SaaS providers (cloud us-ers) and cloud providers, which have received less attention than SaaS us-ers
From a hardware provisioning and pricing point of view, three aspects are new in cloud computing
i- The appearance of infinite com-˲puting resources available on demand, quickly enough to follow load surges, thereby eliminating the need for cloud computing users to plan far ahead for provisioning.
ii- The elimination of an up-front ˲commitment by cloud users, thereby allowing companies to start small and increase hardware resources only when there is an increase in their needs
iii- The ability to pay for use of com-˲puting resources on a short-term basis as needed (for example, processors by the hour and storage by the day) and re-lease them as needed, thereby reward-ing conservation by letting machines and storage go when they are no longer useful.
We argue that the construction and operation of extremely large-scale, commodity-computer data centers at low-cost locations was the key neces-sary enabler of cloud computing, for they uncovered the factors of 5 to 7 decrease in cost of electricity, network bandwidth, operations, software, and hardware available at these very large economies of scale.
We therefore believe that including traditional data centers in the definition of cloud com-puting will lead to exaggerated claims for smaller, so-called private clouds, which is why we exclude them
here we describe how so-called private clouds can get more of the ben-efits of public clouds through
hybrid cloud computing
Classes of Utility Computing
Any application needs a model of com-putation, a model of storage, and a model of communication.
Our view is that different utility computing of-ferings will be distinguished based on the cloud system software’s level of ab-straction and the level of management of the resources
Cloud Computing Economics
We see three particularly compelling use cases that favor utility computing over conventional hosting.
i- A first case is when demand for a service varies with time.
ii- A second case is when demand is unknown in advance.
iii- Finally, organizations that perform batch analytics can use the “cost asso-ciativity” of cloud computing to finish computations faster: using 1,000 EC2 machines for one hour costs the same as using one machine for 1,000 hours
Although the economic appeal ofcloud computing is often described as “converting capital expenses to operat-ing expenses” (CapEx to OpEx), we be-lieve the phrase “pay as you go” more directly captures the economic benefit to the buyer
Hours purchased via cloud computing can be distributed non-uni-formly in time (for example, use 100 server-hours today and no server-hours tomorrow, and still pay only for 100); in the networking community, this way of selling bandwidth is already known as usage-based pricing.c In addition, the absence of up-front capital expense allows capital to be redirected to core business investment.
Therefore, even if Amazon’s pay-as-you-go pricing was more expensive than buying and depreciating a com-parable server over the same period, we argue that the cost is outweighed by the extremely important cloud com-puting economic benefits of elasticity and transference of risk, especially the risks of overprovisioning (underutiliza-tion) and underprovisioning (satura-tion)
While the cost of overprovisioning is easily mea-sured, the cost of underprovisioning is more difficult to measure yet potential-ly equally serious
Table 1. Comparing public clouds and private data centers.
figure 2. (a) Even if peak load can be correctly anticipated, without elasticity we waste resources (shaded area) during nonpeak times. (b) underprovisioning case 1: potential revenue from users not served (shaded area) is sacrificed. (c) underprovisioning case 2: some users desert the site permanently after experiencing poor service; this attrition and possible negative press result in a permanent loss of a portion of the revenue stream
5. Top 10 Obstacles and
Opportunities for Cloud Computing
-The first three affect adop-tion, the next five affect growth, and the last two are policy and business ob-stacles.
Number 1. Business Continuity
and Service Availability
Availability/business Continuity ------Use Multiple Cloud Providers
Number 2. Data Lock-In
Data Lock-in----Standardize APis; Compatible SW to enable Surge or hybird Cloud Computing
Number 3. Data
Data Confidentiality and Auditability --- Deploy encryption, VLANs and firewalls
The primary security mechanism in today’s clouds is virtualization.
However, not all resources are virtualized and not all virtualization environments are bug-free. Virtualiza-tion software has been known to con-tain bugs that allow virtualized code to “break loose” to some extent.
Data Transfer Bottlenecks
Data Transfer bottlenecks ---- Fedexing Disks; higher bW Switches
Performance Unpredictability-- mproved vM Support; Flash Memory; Gang Schedule vMs
Number 6: Scalable Storage
Scalable Storage--- invent Scalable Store
Number 7: Bugs in LargeScale
bugs in Large Distributed Systems--- Invent Debugger that relies on Distributed vMs
Number 8: Scaling Quickly
Scaling Quickly --- invent Auto-Scaler that relies on ML; Snapshots for Conservation
Number 9: Reputation Fate Sharing
Reputation Fate Sharing--- Offer reputation-guarding services like those for email
Number 10: Software Licensing
Software Licensing --- Pay-for-use licenses
Regardless of whether a cloud provider sells services at a low level of abstraction like EC2 or a higher level like AppEngine, we believe computing, storage, and networking must all focus on horizontal scalability of virtualized resources rather than on single node performance.Moreover;
i- Applications software needs to both scale down rapidly as well as scale up, which is a new requirement.
ii- Infrastructure software must be aware that it is no longer running on bare metal but on VMs.
iii- Hardware systems should be designed at the scale of a container (at least a dozen racks), which will be the minimum purchase size.