How Cloud Applications Let Me Trust You

Last April, Satoshi Nakajima, founder of Big Canvas Inc. in Washington, ardently invited new users to use the company's flagship PhotoShare, which allows users to swap iPhone photos for free. However, just three months later, Nakajima expressed its sincere apologies to users. He apologized for not having a problem with the product, nor was it the poor quality of customer service or other problems often experienced by start-ups, but because PhotoShare and its 50,000 users were victims of cloud computing services problems. PhotoShare service was interrupted for seven hours on July 20 when Amazon S3 cloud service was interrupted for some reason. That was the second interruption that the PhotoShare service encountered in 2008.

Nakajima pays Amazon a monthly fee of $ 900 for Amazon cloud computing services. In order to gain flexible computing power and unlimited data storage space, he applied to Amazon for the Elastic Compute Cloud Services (EC2) and Simple Storage Service (S3) services. Amazon claimed service interruption was due to poor performance of communications servers, with the result that Big Canvas lost 50 clients' photo files. Nakajima personally apologized for calling each user.

Nakajima recalls: "We said to those customers at the time: 'The latest photos you uploaded are missing, and we're sorry for that. Please resubmit them or forgive our fault."

Accountability for cloud services

In addition to Nakajima, many other entrepreneurs are also forced to clean up the mess left by the disruption of cloud computing services. In February, Google's Gmail email service was interrupted for hours, affecting about 113 million users. Last month, 14% of Google Apps users ran into problems with their services due to a blockage of network traffic.

July 18 last year, Apple's online cloud service MobileMe also interrupted for a long time, the result caused extreme dissatisfaction with the users.

But this mess did not prevent more and more businesses from using cloud services. IDC, a market research firm, expects IT spending worldwide for cloud services to nearly tripling to $ 42 billion by 2012. However, with the growth of service relevance, the reliability issue of cloud computing service is getting more and more attention. People want to know if famous vendors such as Amazon, Google and Apple are responsible for service interruption.

Pros and cons of cloud services

The benefits of cloud computing services are generally recognized are:

1, the configuration quickly and easily

2, do not have to pay

3, to reduce internal staff demand and reduce costs

The biggest deficiency of cloud computing service:

1, safety

2, performance

3, availability

4, it is difficult to integrate with the internal IT

5, can not be customized

Customers most want to get from cloud computing services:

1, a competitive price

2, performance assurance

3, the company's business and industry understanding

4, able to cloud computing services in the data back to the local service

In 2008, IDC conducted a survey of 244 chief information officers and business managers, with over 63% of executives claiming performance and availability as two of the top three challenges in cloud computing services. Nearly 75% of executives claim they think safety is a serious problem.

There are many smaller businesses that have experienced cloud computing service defects like Big Canvas. While startups are often hard hit by service interruptions, even the Amazon Times cloud broke down in February 2008, even the New York Times. The New York Times uses the S3 service to save and extract articles from historical databases.

John Sloan, analyst at Info-Tech Research Group, a market-analysis firm, said: "Disruptions to mission-critical applications can cost millions of dollars for a while."

For the current cloud computing service providers, who is responsible for ensuring service operations and clean up the turmoil is a very tricky issue. Because in fact almost all cloud service providers can not guarantee service level agreement (SLA) can reach 99.99% uptime, or because of service interruptions to reduce service costs. Vendors insist that ensuring 99.99% uptime is costly.

Sloan said: "In order to ensure 99.99% service level, the supplier may charge you more." He added that other additional conditions may also include you have to sign a multi-year service contract with the supplier.

In the division of responsibilities and accountability, there will often be some phenomenon of biting the word. For example, Google's SLA states that Google Apps SLAs do not apply to performance issues that result from other factors beyond Google's control. Communications server failure or DDoS attacks should be excluded from outside the control of Google has been the topic of debate among lawyers, small companies who simply can not afford the costs.

Cloud service compensation

Developed a comprehensive terms of service, taking all precautionary measures, cloud computing service provider reputation and records may be the best indicator of service reliability. On the plus side, many suppliers can do a better job of notifying customers on timely service interruptions. For example, Google released the Google Apps Status Dashboard in February, which gives users instant access to current availability of applications such as Google Gmail, Video and Docs. Applications that are out of service are marked with a red "X", while they are marked with a green check mark during normal operation.

Some vendors are even willing to compensate certain customers for service interruptions. Google, for example, provided Google Apps Enhanced Edition paid users with 15 days of service to compensate customers in February 2008 for service interruptions. But users receive different compensation. For services disrupted by Big Canvas, Nakajima said Amazon did not charge for seven hours of service disruption, but that is equivalent to less than a 1% discount on monthly subscriptions.

"Manufacturers should give us 99.9% uptime, and if they can not deliver that promise, they should pay us back for a full month," he said.

Ray Wang, an analyst at Forrester, a market research firm, thinks it is impossible to expect large payments from service providers.

Wang pointed out: "All existing cloud service agreements do not address the issue of compensation to customers, you can get compensation for future service hours, but there is absolutely no way to compensate for your direct losses."

Not all users agree that cloud service providers' compensation policies are not reasonable. Peter Sanchez is the best example. He is the founder of Los Angeles startup SmartJabber.com, which sells automated customer service software such as virtual chat agent software for online retailers. Since its launch in April last year, SmartJabber has been using Amazon S3 service to save image files, JavaScript files and website images. However, in July last year, a serious disruption occurred in the S3 service, leaving SmartJabber and his clients unreadable for up to 6 hours.

Sanchez recalled: "When the service was interrupted, neither the client's chat window nor the website image was properly loaded, causing the site to look rather bad and many users complained."

Despite complaints from about 15 users, Sanchez unexpectedly suffered a severe test, saying he had never considered requiring Amazon compensation for his losses. According to Sanchez, the monthly service charge is only 35 U.S. dollars, because the costs of occasional disruption should be low. If you use Amazon's cloud storage services, and use their own internal server, then he would need to pay almost every month to reach 500 US dollars.

Sanchez said: "Terms of service are long and vendors do not intend to hide anything. You either accept the agreement or go to other vendors that provide a higher level of service."

He believes that cloud computing services does not mean that users have to surrender all control of the system. Sanchez said SmartJabber can download the data stored in S3 to its own local storage server in minutes when the service is down. He said: "This is not the best solution, but it allows us to continue working."

Nakajima also has a similar contingency plan. Today, as a precautionary measure, Big Canvas's EC2 server temporarily saves those photos in the cache before transferring the user's photos to the S3 server.

Despite some complaints, internal IT departments are still under greater pressure because they must provide internal services that are service standards that are current to the services provided by many vendors.

"If you want to compare the uptime provided by cloud providers with the uptime provided by your in-house IT team, you will find that cloud service providers offer relatively higher levels of service because of the higher number of censors Higher. "

Finally, it is up to the companies themselves how much to risk and the willingness to accept the loss of cloud service disruption that may offset the cost savings and convenience benefits of cloud computing service commitments.

Sloan, of Info-Tech Research, said: "If your company can afford to be interrupted by occasional service interruptions, then cloud computing services are solid and you are just starting a business. You do not invest a lot and you can try your luck. It may be possible to start a business. "

Before businesses migrate to the cloud, they should first consider the following five questions:

1, your cloud service provider is how to define "good customer service"? The advanced nature of the service itself is important, but you have to figure out what kind of service is a specific service is a good service, as well as services are What is included, such as compensation policies, technical assistance, etc.

2, Are you satisfied with the physical facilities of the cloud service provider? Due diligence includes checking the physical facilities of the manufacturer, paying attention to the matters related to the maintenance of the manufacturer's building, and the equipment maintenance plan of the manufacturer and the staff in the building Quantity and so on.

3, you should know what type of service interruption will occur. There are times when vendors have to turn off some updates or upgrades because those activities can seriously hamper your cloud computing service. Find out how often the vendor plans to schedule a routine overhaul, and what advance notification it will provide to the user.

4. How fast is a cloud provider growing? You have to know if your vendor is able to accept more new users without compromising service levels.

5, cloud service providers prepared for those follow-up procedures? Once the service interruption occurs, the important thing is how quickly it can identify the root cause of the problem.

【Editor's Choice】