use the article provided to answer questions
50 CommuniCAtionS of thE ACm | APRiL 2010 | vOL. 53 | nO. 4 practice CLOUd COMPUting, the long-held dream of computing as a utility, has the potential to transform a large part of the IT industry, making software even more attractive as a service and shaping the way IT hardware is designed and purchased. 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. They need not be concerned about overprovisioning for a service whose popularity does not meet their predictions, thus wasting costly resources, or underprovisioning for one that becomes wildly popular, thus missing potential customers and revenue. Moreover, companies with large batch-oriented tasks can get results as quickly as their programs can scale, since using 1,000 servers for one hour costs no more than using one server for 1,000 A View of Cloud Computing Doi:10.1145/1721654.1721672 Clearing the clouds away from the true potential and obstacles posed by this computing capability.
B y miChAEL ARmBRuS t, ARmAnDo fox, REAn GRiffith, Anthony D. JoSEPh, RAnD y KAtz, AnD y KonWinSKi, Gunho LEE, D AViD PAttERSon, ARiEL RABKin, ion StoiCA, AnD mA tEi zAhARiA hours. This elasticity of resources, with- out paying a premium for large scale, is unprecedented in the history of IT.
As a result, cloud computing is a popular topic for blogging and white papers and has been featured in the title of workshops, conferences, and even magazines. Nevertheless, confu- sion remains about exactly what it is and when it’s useful, causing Oracle’s CEO Larry Ellison to vent his frustra- tion: “The interesting thing about cloud computing is that we’ve rede- fined cloud computing to include ev- erything that we already do…. I don’t understand what we would do differ- ently in the light of cloud computing other than change the wording of some of our ads.” Our goal in this article is to reduce that confusion by clarifying terms, pro- viding simple figures to quantify com- parisons between of cloud and con- ventional computing, and identifying the top technical and non-technical obstacles and opportunities of cloud computing. (Armbrust et al 4 is a more detailed version of this article.) Defining Cloud Computing Cloud computing refers to both the applications delivered as services over the Internet and the hardware and sys- tems 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. The line between “low-level” infrastructure and a higher-level “plat- form” is not crisp. We believe the two are more alike than different, and we consider them together. Similarly, the a For the purposes of this article, we use the term Software as a Service to mean applica- tions delivered over the Internet. The broadest definition would encompass any on demand software, including those that run software locally but control use via remote software li- censing. APRiL 2010 | vOL. 53 | nO. 4 | CommuniCA tionS of thE ACm 51 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. Thus, 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. Figure 1 makes provider-user re- lationships clear. In some cases, the same actor can play multiple roles. For instance, a cloud provider might also host its own customer-facing services on cloud infrastructure.
From a hardware provisioning and pricing point of view, three aspects are new in cloud computing. 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.
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. b 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.
b Note, however, that upfront commitments can still be used to reduce per-usage charges.
For example, Amazon Web Services also offers long-term rental of servers, which they call re- served instances. ILLUstratIon by Jon han 52 CommuniCAtionS of thE ACm | APRiL 2010 | vOL. 53 | nO. 4 practice 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. These factors, combined with statistical multiplexing to increase utilization compared to tra- ditional data centers, meant that cloud computing could offer services below the costs of a medium-sized data cen- ter and yet still make a good profit. Our proposed definition allows us to clearly identify certain installations as examples and non-examples of cloud computing. Consider a public-facing Internet service hosted on an ISP who can allocate more machines to the ser- vice given four hours notice. Since load surges on the public Internet can hap- pen much more quickly than that (An- imoto saw its load double every 12 hours for nearly three days), this is not cloud computing. In contrast, consider an internal enterprise data center whose applications are modified only with sig- nificant advance notice to administra- tors. In this scenario, large load surges on the scale of minutes are highly un- likely, so as long as allocation can track expected load increases, this scenario fulfills one of the necessary conditions for operating as a cloud. The enterprise data center may still fail to meet other conditions for being a cloud, however, such as the appearance of infinite re- sources or fine-grained billing. A pri- vate data center may also not benefit from the economies of scale that make public clouds financially attractive. Omitting private clouds from cloud computing has led to considerable de- bate in the blogosphere. We believe the confusion and skepticism illustrated by Larry Ellison’s quote occurs when the advantages of public clouds are also claimed for medium-sized data centers. Except for extremely large data centers of hundreds of thousands of machines, such as those that might be operated by Google or Microsoft, most data centers enjoy only a subset of the potential advantages of public clouds, as Table 1 shows. 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. How- ever, here we describe how so-called private clouds can get more of the ben- efits of public clouds through surge computing or hybrid cloud computing.
Classes of utility Computing Any application needs a model of com- putation, a model of storage, and a model of communication. The statisti- cal multiplexing necessary to achieve elasticity and the appearance of infi- nite capacity available on demand re- quires automatic allocation and man- agement. In practice, this is done with virtualization of some sort. 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. Amazon EC2 is at one end of the spectrum. An EC2 instance looks much like physical hardware, and us- ers can control nearly the entire soft- ware stack, from the kernel upward. This low level makes it inherently dif- ficult for Amazon to offer automatic scalability and failover because the semantics associated with replication and other state management issues are highly application-dependent. At the other extreme of the spectrum are application domain-specific platforms such as Google AppEngine, which is targeted exclusively at traditional Web applications, enforcing an applica- tion structure of clean separation be- tween a stateless computation tier and a stateful storage tier. AppEngine’s impressive automatic scaling and high-availability mechanisms, and the proprietary MegaStore data storage available to AppEngine applications, all rely on these constraints. Applica- tions for Microsoft’s Azure are written using the .NET libraries, and compiled to the Common Language Runtime, a language-independent managed en- vironment. The framework is signifi- cantly more flexible than AppEngine’s, but still constrains the user’s choice of storage model and application struc- ture. Thus, Azure is intermediate be- tween application frameworks like AppEngine and hardware virtual ma- chines like EC2.
Cloud Computing Economics We see three particularly compelling use cases that favor utility computing over conventional hosting. A first case is when demand for a service varies with time. For example, provisioning a data center for the peak load it must sustain a few days per month leads to underutilization at other times.
Instead, cloud computing lets an or- ganization pay by the hour for com- puting resources, potentially leading to cost savings even if the hourly rate to rent a machine from a cloud pro- vider is higher than the rate to own one. A second case is when demand is unknown in advance. For example, a Web startup will need to support a spike in demand when it becomes popular, followed potentially by a re- duction once some visitors turn away.
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 of figure 1. users and providers of cloud computing. We focus on cloud computing’s effects on cloud providers and SaaS providers/cloud users. the top level can be recursive, in that SaaS providers can also be a SaaS users via mashups. Web applications Utility computing SaaS user Cloud Provider SaaS Provider/ Cloud user practice APRiL 2010 | vOL. 53 | nO. 4 | CommuniCA tionS of thE ACm 53 cloud 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). We start with elasticity. The key ob- servation is that cloud computing’s ability to add or remove resources at a fine grain (one server at a time with EC2) and with a lead time of minutes rather than weeks allows matching resources to workload much more closely. Real world estimates of average server utilization in data centers range from 5% to 20%. 15,17 This may sound c Usage-based pricing is not renting. Renting a resource involves paying a negotiated cost to have the resource over some time period, whether or not you use the resource. Pay-as- you-go involves metering usage and charging based on actual use, independently of the time period over which the usage occurs. shockingly low, but it is consistent with the observation that for many services the peak workload exceeds the aver- age by factors of 2 to 10. Since few us- ers deliberately provision for less than the expected peak, resources are idle at nonpeak times. The more pronounced the variation, the more the waste.
For example, Figure 2a assumes our service has a predictable demand where the peak requires 500 servers at noon but the trough requires only 100 servers at midnight. As long as the aver- age utilization over a whole day is 300 servers, the actual cost per day (area un- der the curve) is 300 × 24 = 7,200 server hours; but since we must provision to the peak of 500 servers, we pay for 500 × 24 = 12,000 server-hours, a factor of 1.7 more. Therefore, as long as the pay-as- you-go cost per server-hour over three years (typical amortization time) is less than 1.7 times the cost of buying the server, utility computing is cheaper. In fact, this example underestimates the benefits of elasticity, because in ad- dition to simple diurnal patterns, most services also experience seasonal or other periodic demand variation (for example, e-commerce in December and photo sharing sites after holidays) as well as some unexpected demand bursts due to external events (for ex- ample, news events). Since it can take weeks to acquire and rack new equip- ment, to handle such spikes you must provision for them in advance. We al- ready saw that even if service operators predict the spike sizes correctly, capac- ity is wasted, and if they overestimate the spike they provision for, it’s even worse. They may also underestimate the spike (Figure 2b), however, accidental- ly turning away excess users. While the cost of overprovisioning is easily mea- sured, the cost of underprovisioning is more difficult to measure yet potential- ly equally serious: not only do rejected users generate zero revenue, they may never come back. For example, Friend- ster’s decline in popularity relative to competitors Facebook and MySpace is believed to have resulted partly from user dissatisfaction with slow response times (up to 40 seconds). 16 Figure 2c aims to capture this behavior: Users will desert an underprovisioned service until the peak user load equals the data center’s usable capacity, at which point users again receive acceptable service. For a simplified example, assume that users of a hypothetical site fall into two classes: active users (those who use the site regularly) and defectors (those who abandon the site or are turned away from the site due to poor perfor- mance). Further, suppose that 10% of active users who receive poor service due to underprovisioning are “perma- nently lost” opportunities (become de- fectors), that is, users who would have remained regular visitors with a better experience. The site is initially provi- sioned to handle an expected peak of 400,000 users (1,000 users per server × 400 servers), but unexpected positive press drives 500,000 users in the first hour. Of the 100,000 who are turned away or receive bad service, by our as- sumption 10,000 of them are perma- nently lost, leaving an active user base of 390,000. The next hour sees 250,000 new unique users. The first 10,000 do fine, but the site is still overcapacity by 240,000 users. This results in 24,000 additional defections, leaving 376,000 permanent users. If this pattern con- tinues, after lg(500,000) or 19 hours, the number of new users will approach zero and the site will be at capacity in steady state. Clearly, the service op- erator has collected less than 400,000 users’ worth of steady revenue during those 19 hours, however, again illustrat- ing the underutilization argument—to say nothing of the bad reputation from the disgruntled users. Do such scenarios really occur in practice? When Animoto 3 made its ser- vice available via Facebook, it experi- enced a demand surge that resulted in growing from 50 servers to 3,500 serv- ers in three days. Even if the average t able 1. Comparing public clouds and private data centers.
Advantage Public Cloud Conventional Data Center Appearance of infinite computing resources on demand Yes no elimination of an up-front commitment by Cloud users Yes no Ability to pay for use of computing resources on a short-term basis as needed Yes no e conomies of scale due to very large data centers Yes Usually not higher utilization by multiplexing of workloads from different organizations Yes Depends on company size Simplify operation and increase utilization via resource virtualization Yes no 54 CommuniCAtionS of thE ACm | APRiL 2010 | vOL. 53 | nO. 4 practice tomers will be reluctant to migrate to cloud computing without a business- continuity strategy for such situations.
We believe the best chance for inde- pendent software stacks is for them to be provided by different companies, as it has been difficult for one company to justify creating and maintain two stacks in the name of software depend- ability. Just as large Internet service providers use multiple network provid- ers so that failure by a single company will not take them off the air, we believe the only plausible solution to very high availability is multiple cloud comput- ing providers.
number 2. Data Lock-in Software stacks have improved interop- erability among platforms, but the stor- age APIs for cloud computing are still essentially proprietary, or at least have not been the subject of active stan- dardization. Thus, customers cannot easily extract their data and programs from one site to run on another. Con- table 2. t op 10 obstacles to and opportunities for growth of cloud computing. obstacle opportunity 1 Availability/business Continuity Use Multiple Cloud Providers 2 Data Lock-in Standardize APis; Compatible SW to enable Surge or hybird Cloud Computing 3 Data Confidentiality and Auditability Deploy encryption, vLAns, Firewalls 4 Data Transfer bottlenecks Fedexing Disks; higher bW Switches 5 Performance Unpredictability improved vM Support; Flash Memory; Gang Schedule vMs 6 Scalable Storage invent Scalable Store 7 bugs in Large Distributed Systems invent Debugger that relies on Distributed vMs 8 Scaling Quickly invent Auto-Scaler that relies on ML; Snapshots for Conservation 9 Reputation Fate Sharing Offer reputation-guarding services like those for email 10 Software Licensing Pay-for-use licenses utilization of each server was low, no one could have foreseen that resource needs would suddenly double every 12 hours for three days. After the peak sub- sided, traffic fell to a lower level. So in this real-world example, scale-up elas- ticity was not a cost optimization but an operational requirement, and scale- down elasticity allowed the steady-state expenditure to more closely match the steady-state workload.
top 10 obstacles and opportunities for Cloud Computing Table 2 summarizes our ranked list of critical obstacles to growth of cloud computing. The first three affect adop- tion, the next five affect growth, and the last two are policy and business ob- stacles. Each obstacle is paired with an opportunity to overcome that obstacle, ranging from product development to research projects.
number 1. Business Continuity and Service Availability Organizations worry about whether utility computing services will have adequate availability, and this makes some wary of cloud computing. Ironi- cally, existing SaaS products have set a high standard in this regard. Google Search has a reputation for being high- ly available, to the point that even a small disruption is picked up by major news sources. 11 Users expect similar availability from new services, which is difficult to do. Table 3 shows recorded outages for Amazon Simple Storage Service (S3), AppEngine and Gmail in 2008, and explanations for the outages. Note that despite the negative publicity due to these outages, few enterprise IT infrastructures are as good. Techni- cal issues of availability aside, a cloud provider could suffer outages for non- technical reasons, including going out of business or being the target of regu- latory action (a recent example of the latter occurred last year, as we describe later).
Although they have not done so, cloud vendors could offer specialized hardware and software techniques in order to deliver higher reliability, pre- sumably at a high price. This reliability could then be sold to users as a service- level agreement. But this approach only goes so far. The high-availability com- puting community has long followed the mantra “no single point of failure,” yet the management of a cloud com- puting service by a single company is in fact a single point of failure. Even if the company has multiple data centers in different geographic regions using dif- ferent network providers, it may have common software infrastructure and accounting systems, or the company may even go out of business. Large cus- 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. Resources time (days) 1 23 Capacity demand (a) Provisioning for peak load Resources 1 23 time (days) Capacity demand (b) underprovisioning 1 Resources 1 23 time (days) Capacity demand (c) underprovisioning 2 practice APRiL 2010 | vOL. 53 | nO. 4 | CommuniCA tionS of thE ACm 55 cern about the difficulty of extracting data from the cloud is preventing some organizations from adopting cloud computing. Customer lock-in may be attractive to cloud computing provid- ers, but their users are vulnerable to price increases, to reliability problems, or even to providers going out of busi- ness.
For example, an online storage ser- vice called The Linkup shut down on Aug. 8, 2008 after losing access as much as 45% of customer data. 6 The Linkup, in turn, had relied on the online stor- age service Nirvanix to store customer data, which led to finger pointing be- tween the two organizations as to why customer data was lost. Meanwhile, The Linkup’s 20,000 users were told the service was no longer available and were urged to try out another storage site. One solution would be to standard- ize the APIs d in such a way that a SaaS developer could deploy services and data across multiple cloud computing providers so that the failure of a single company would not take all copies of customer data with it. One might worry that this would lead to a “race-to-the- bottom” of cloud pricing and flatten the profits of cloud computing provid- ers. We offer two arguments to allay this fear. First, the quality of a service matters as well as the price, so customers may not jump to the lowest-cost service.
Some Internet service providers today cost a factor of 10 more than others because they are more dependable and offer extra services to improve usabil- ity. Second, in addition to mitigating data lock-in concerns, standardization of APIs enables a new usage model in which the same software infrastruc- ture can be used in an internal data center and in a public cloud. Such an option could enable hybrid cloud com- puting or surge computing in which the public cloud is used to capture the extra tasks that cannot be easily run in the data center (or private cloud) due to temporarily heavy workloads.
This option could significantly expand the cloud computing market. Indeed, open-source reimplementations of proprietary cloud APIs, such as Euca- d Data Liberation Front; http://dataliberation.org lyptus and HyperTable, are first steps in enabling surge computing.
number 3. Data Confidentiality/Auditability Despite most companies outsourcing payroll and many companies using external email services to hold sensi- tive information, security is one of the most often-cited objections to cloud computing; analysts and skeptical companies ask “who would trust their essential data out there somewhere?” There are also requirements for audit- ability, in the sense of Sarbanes-Oxley and Health and Human Services Health Insurance Portability and Accountabil- ity Act (HIPAA) regulations that must be provided for corporate data to be moved to the cloud.
Cloud users face security threats both from outside and inside the cloud.
Many of the security issues involved in protecting clouds from outside threats are similar to those already facing large data centers. In the cloud, however, this responsibility is divided among potentially many parties, including the cloud user, the cloud vendor, and any third-party vendors that users rely on for security-sensitive software or con- figurations. The cloud user is responsible for application-level security. The cloud provider is responsible for physical security, and likely for enforcing exter- nal firewall policies. Security for inter- mediate layers of the software stack is shared between the user and the oper- ator; the lower the level of abstraction exposed to the user, the more respon- sibility goes with it. Amazon EC2 us- ers have more technical responsibility (that is, must implement or procure more of the necessary functionality themselves) for their security than do Azure users, who in turn have more re- sponsibilities than AppEngine custom- ers. This user responsibility, in turn, can be outsourced to third parties who sell specialty security services. The ho- mogeneity and standardized interfaces of platforms like EC2 make it possible for a company to offer, say, configura- tion management or firewall rule anal- ysis as value-added services.
While cloud computing may make external-facing security easier, it does pose the new problem of internal- facing security. Cloud providers must guard against theft or denial-of-service attacks by users. Users need to be pro- tected from one another. The primary security mechanism in today’s clouds is virtualization. It is a powerful defense, and protects against most attempts by users to attack one another or the underlying cloud infra- structure. 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. Incorrect network virtualization may allow user code access to sensitive portions of the provider’s infrastructure, or to the re- sources of other users. These challeng- es, though, are similar to those involved in managing large non-cloud data cen- ters, where different applications need to be protected from one another. Any large Internet service will need to en- sure that a single security hole doesn’t compromise everything else. One last security concern is protect- ing the cloud user against the provider.
The provider will by definition con- trol the “bottom layer” of the software stack, which effectively circumvents most known security techniques. Ab- sent radical improvements in security technology, we expect that users will use contracts and courts, rather than clever security engineering, to guard against provider malfeasance. The one important exception is the risk of inad- vertent data loss. It’s difficult to imag- ine Amazon spying on the contents of virtual machine memory; it’s easy to t able 3. outages in AWS, AppEngine, and gmail service and outage duration date.
Service and outage Duration Date S3 outage: authentication service overload leading to unavailability 17 2 hours 2/15/08 S3 outage: Single bit error leading to gossip protocol blowup 18 6–8 hours 7/20/08 Appengine partial outage: programming error 19 5 hours 6/17/08 Gmail: site unavailable due to outage in contacts system 11 1.5 hours 8/11/08 56 CommuniCAtionS of thE ACm | APRiL 2010 | vOL. 53 | nO. 4 practice bandwidth of 5Mbits/sec to 18Mbits/ sec. Suppose we get 20Mbits/sec over a WAN link. It would take 10 * 1012 Bytes / (20×106 bits/second) = (8×1013)/(2×107) seconds = 4,000,000 seconds, which is more than 45 days. If we in- stead sent 10 1TB disks via overnight shipping, it would take less than a day to transfer 10TB, yielding an effective bandwidth of about 1,500Mbit/sec. For example, AWS 8 recently started offering such a service, called Import/Export.
number 5. Performance unpredictability Our experience is that multiple vir- tual machines (VMs) can share CPUs and main memory surprisingly well in cloud computing, but that network and disk I/O sharing is more problematic.
As a result, different EC2 instances vary more in their I/O performance than in main memory performance. We mea- sured 75 EC2 instances running the STREAM memory benchmark. 14 The mean bandwidth is 1,355Mbytes/ sec., with a standard deviation across instances of just 52MBytes/sec, less than or about 4% of the mean. We also measured the average disk bandwidth for 75 EC2 instances each writing 1GB files to local disk. The mean disk write bandwidth is nearly 55Mbytes per sec- ond with a standard deviation across instances of a little over 9MBytes/sec, or about 16% of the mean. This demon- strates the problem of I/O interference between virtual machines. One opportunity is to improve ar- chitectures and operating systems to efficiently virtualize interrupts and I/O channels. Note that IBM mainframes and operating systems largely over- came these problems in the 1980s, so we have successful examples from which to learn. Another possibility is that flash memory will decrease I/O interference.
Flash is semiconductor memory that preserves information when powered off like mechanical hard disks, but since it has no moving parts, it is much faster to access (microseconds vs. mil- liseconds) and uses less energy. Flash memory can sustain many more I/Os per second per gigabyte of storage than disks, so multiple virtual machines imagine a hard disk being disposed of without being wiped, or a permissions bug making data visible improperly.
This is a problem in non-cloud con- texts as well. The standard defense, user-level encryption, is also effective in the cloud. This is already common for high-value data outside the cloud, and both tools and expertise are readily available. This approach was success- fully used by TC3, a health care com- pany with access to sensitive patient records and health care claims, when moving their HIPAA-compliant appli- cation to AWS. 3 Similarly, auditability could be add- ed as an additional layer beyond the reach of the virtualized guest OS, pro- viding facilities arguably more secure than those built into the applications themselves and centralizing the soft- ware responsibilities related to confi- dentiality and auditability into a single logical layer. Such a new feature rein- forces the cloud computing perspec- tive of changing our focus from specific hardware to the virtualized capabilities being provided.
number 4. Data t ransfer Bottlenecks Applications continue to become more data-intensive. If we assume applica- tions may be “pulled apart” across the boundaries of clouds, this may compli- cate data placement and transport. At $100 to $150 per terabyte transferred, these costs can quickly add up, mak- ing data transfer costs an important issue. Cloud users and cloud providers have to think about the implications of placement and traffic at every level of the system if they want to minimize costs. This kind of reasoning can be seen in Amazon’s development of its new cloudfront service. One opportunity to overcome the high cost of Internet transfers is to ship disks. Jim Gray found the cheap- est way to send a lot of data is to ship disks or even whole computers. 10 While this does not address every use case, it effectively handles the case of large delay-tolerant point-to-point transfers, such as importing large data sets. To quantify the argument, assume that we want to ship 10TB from U.C.
Berkeley to Amazon in Seattle, WA. Gar- finkel 9 measured bandwidth to S3 from three sites and found an average write ACM Journal on Computing and Cultural Heritage � ���� JOCCH publishes papers of significant and lasting value in all areas relating to the use of ICT in suppor t of Cultural Heritage, seek ing to combine the best of computing science with real attention to any aspec t of the cultural heritage sector. � ���� www.acm.org/jocch www.acm.org/subscribe practice APRiL 2010 | vOL. 53 | nO. 4 | CommuniCA tionS of thE ACm 57 with conflicting random I/O workloads could coexist better on the same physi- cal computer without the interference we see with mechanical disks.
Another unpredictability obstacle concerns the scheduling of virtual ma- chines for some classes of batch pro- cessing programs, specifically for high- performance computing. Given that high-performance computing (HPC) is used to justify government purchases of $100M supercomputer centers with 10,000 to 1,000,000 processors, there are many tasks with parallelism that can benefit from elastic computing.
Today, many of these tasks are run on small clusters, which are often poorly utilized. There could be a significant savings in running these tasks on large clusters in the cloud instead. Cost as- sociativity means there is no cost pen- alty for using 20 times as much com- puting for 1/20th the time. Potential applications that could benefit include those with very high potential financial returns—financial analysis, petroleum exploration, movie animation—that would value a 20x speedup even if there were a cost premium. The obstacle to attracting HPC is not the use of clusters; most parallel computing today is done in large clus- ters using the message-passing inter- face MPI. The problem is that many HPC applications need to ensure that all the threads of a program are run- ning simultaneously, and today’s virtu- al machines and operating systems do not provide a programmer-visible way to ensure this. Thus, the opportunity to overcome this obstacle is to offer some- thing like “gang scheduling” for cloud computing. The relatively tight timing coordination expected in traditional gang scheduling may be challenging to achieve in a cloud computing environ- ment due to the performance unpre- dictability just described.
number 6: Scalable Storage Earlier, we identified three properties whose combination gives cloud com- puting its appeal: short-term usage (which implies scaling down as well as up when demand drops), no upfront cost, and infinite capacity on demand.
While it’s straightforward what this means when applied to computation, it’s less clear how to apply it to persis- tent storage. There have been many attempts to answer this question, varying in the richness of the query and storage API’s, the performance guarantees offered, and the resulting consistency seman- tics. The opportunity, which is still an open research problem, is to create a storage system that would not only meet existing programmer expecta- tions in regard to durability, high avail- ability, and the ability to manage and query data, but combine them with the cloud advantages of scaling arbitrarily up and down on demand.
number 7: Bugs in Large- Scale Distributed Systems One of the difficult challenges in cloud computing is removing errors in these very large-scale distributed systems. A common occurrence is that these bugs cannot be reproduced in smaller config- urations, so the debugging must occur at scale in the production data centers. One opportunity may be the reliance on virtual machines in cloud comput- ing. Many traditional SaaS providers de- veloped their infrastructure without us- ing VMs, either because they preceded the recent popularity of VMs or because they felt they could not afford the per- formance hit of VMs. Since VMs are de rigueur in utility computing, that level of virtualization may make it possible to capture valuable information in ways that are implausible without VMs.
number 8: Scaling Quickly Pay-as-you-go certainly applies to stor- age and to network bandwidth, both of which count bytes used. Computation is slightly different, depending on the virtualization level. Google AppEngine automatically scales in response to load increases and decreases, and us- ers are charged by the cycles used. AWS charges by the hour for the number of instances you occupy, even if your ma- chine is idle. The opportunity is then to auto- matically scale quickly up and down in response to load in order to save money, but without violating service- level agreements. Indeed, one focus of the UC Berkeley Reliable Adaptive Distributed Systems Laboratory is the pervasive and aggressive use of statis- tical machine learning as a diagnostic and predictive tool to allow dynamic scaling, automatic reaction to perfor- Just as large iSPs use multiple network providers so that failure by a single company will not take them off the air, we believe the only plausible solution to very high availability is multiple cloud computing providers. 58 CommuniCAtionS of thE ACm | APRiL 2010 | vOL. 53 | nO. 4 practice mance and correctness problems, and automatically managing many other aspects of these systems.
Another reason for scaling is to con- serve resources as well as money. Since an idle computer uses about two-thirds of the power of a busy computer, care- ful use of resources could reduce the impact of data centers on the environ- ment, which is currently receiving a great deal of negative attention. Cloud computing providers already perform careful and low-overhead accounting of resource consumption. By impos- ing fine-grained costs, utility comput- ing encourages programmers to pay attention to efficiency (that is, releas- ing and acquiring resources only when necessary), and allows more direct measurement of operational and de- velopment inefficiencies. Being aware of costs is the first step to conservation, but configuration hassles make it tempting to leave machines idle overnight so that startup time is zero when developers return to work the next day. A fast and easy-to-use snapshot/re- start tool might further encourage con- servation of computing resources.
number 9: Reputation f ate Sharing One customer’s bad behavior can af- fect the reputation of others using the same cloud. For instance, black- listing of EC2 IP addresses 13 by spam- prevention services may limit which applications can be effectively hosted.
An opportunity would be to create rep- utation-guarding services similar to the “trusted email” services currently offered (for a fee) to services hosted on smaller ISP’s, which experience a mi- crocosm of this problem. Another legal issue is the question of transfer of legal liability—cloud com- puting providers would want custom- ers to be liable and not them (such as, the company sending the spam should be held liable, not Amazon). In March 2009, the FBI raided a Dallas data cen- ter because a company whose services were hosted there was being investi- gated for possible criminal activity, but a number of “innocent bystander” companies hosted in the same facility suffered days of unexpected downtime, and some went out of business. 7 number 10: Software Licensing Current software licenses commonly restrict the computers on which the software can run. Users pay for the software and then pay an annual main- tenance fee. Indeed, SAP announced that it would increase its annual main- tenance fee to at least 22% of the pur- chase price of the software, which is close to Oracle’s pricing. 17 Hence, many cloud computing providers originally relied on open source software in part because the licensing model for com- mercial software is not a good match to utility computing. The primary opportunity is either for open source to remain popular or simply for commercial software com- panies to change their licensing struc- ture to better fit cloud computing. For example, Microsoft and Amazon now offer pay-as-you-go software licensing for Windows Server and Windows SQL Server on EC2. An EC2 instance run- ning Microsoft Windows costs $0.15 per hour instead of $0.10 per hour for the open source alternative. IBM also announced pay-as-you-go pricing for hosted IBM software in conjunction with EC2, at prices ranging from $0.38 per hour for DB2 Express to $6.39 per hour for IBM WebSphere with Lotus Web Content Management Server.
Conclusion We predict cloud computing will grow, so developers should take it into ac- count. 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: Applications software needs to 1.
both scale down rapidly as well as scale up, which is a new requirement. Such software also needs a pay-for-use li- censing model to match needs of cloud computing. Infrastructure software must be 2.
aware that it is no longer running on bare metal but on VMs. Moreover, me- tering and billing need to be built in from the start. Hardware systems should be de- 3.
signed at the scale of a container (at least a dozen racks), which will be the minimum purchase size. Cost of oper- ation will match performance and cost of purchase in importance, rewarding energy proportionality 5 by putting idle portions of the memory, disk, and net- work into low-power mode. Processors should work well with VMs and flash memory should be added to the mem- ory hierarchy, and LAN switches and WAN routers must improve in band- width and cost.
Acknowledgments This research is supported in part by gifts from Google, Microsoft, Sun Mi- crosystems, Amazon Web Services, Cisco Systems, Cloudera, eBay, Face- book, Fujitsu, Hewlett-Packard, Intel, Network Appliances, SAP, VMWare, Ya- hoo! and by matching funds from the University of California Industry/Uni- versity Cooperative Research Program (UC Discovery) grant COM07-10240 and by the National Science Founda- tion Grant #CNS-0509559. t he authors are associated with the UC berkeley reliable adaptive distributed s ystems Laboratory (rad Lab).
References 1. amazon.com Ceo Jeff bezos on animoto (apr. 2008); http://blog.animoto.com/2008/04/21/amazon-ceo-jeff- bezos-on-animoto/.
2. amazon s3 t eam. amazon s3 a vailability event (July 20, 2008); http://status.aws.amazon.com/s3- 20080720.html.
3. amazon Web services. t C3 health Case study; http:// aws.amazon.com/solutions/case-studies/tc3-health/.
4. a rmbrust, M., et al. above the clouds: a b erkeley view of cloud computing. t ech. rep. UCb/eeCs-2009-28, ee Cs department, U.C. berkeley, Feb 2009.
5. barroso, L.a., and holzle, U. the case for energy- proportional computing. IEEE Computer 40, 12 (dec. 2007).
6. b rodkin, J. Loss of customer data spurs closure of online storage service ’the Linkup.’ Network World (aug. 2008).
7. Fink, J. FbI agents raid dallas computer business (apr. 2009); http://cbs11tv.com/local/Core.
8. Freedom oss. Large data set transfer to the cloud (apr.
9. g arfinkel, s. a n evaluation of amazon’s grid Computing services: e C2, s3 and sqs. t ech. rep. tr-08-07, harvard University, aug. 2007.
10. gray, J., and Patterson, d . a conversation with Jim gray. ACM Queue 1, 4 (2003), 8–17.
11. helft, M. google confirms problems with reaching its services (May 14, 2009).
12. Jackson, t . We feel your pain, and we’re sorry (aug.
2008); http://gmailblog.blogspot.com/2008/08/we- feel-your-pain-and-were-sorry.html.
13. Krebs, b . amazon: hey spammers, get off my cloud!
Washington Post (July 1 2008).
14. McCalpin, J. Memory bandwidth and machine balance in current high performance computers. IEEE Technical Committee on Computer Architecture Newsletter (1995), 19–25.
15. rangan, K. the Cloud Wars: $100+ billion at stake. tech. rep., Merrill Lynch, May 2008.
16. rivlin, g . Wallflower at the Web Party (oct. 15, 2006).
17. s iegele, L. Let It rise: a s pecial report on Corporate It . The Economist (oct. 2008).
18. stern, a. Update from amazon regarding Friday’s s3 downtime. Centernetworks (Feb. 2008); http://www.
19. Wilson, s. appengine outage. CIo Weblog (June 2008); http://www.cio-weblog.com/50226711/appengine\ outage.php.
© 2010 a CM 0001-0782/10/0400 $10.00