What should be evaluated for cloud migration

As cloud infrastructure services in the enterprise mature and computing slowly melts into people's lives, companies are increasingly eager to migrate important workloads to cloud computing-public, private, or mixed clouds. But how do you migrate mission-critical applications? How do you build an application on a traditional architecture? It can be said that this is a very difficult thing.

Dvae Leclair, senior strategic director of Stratus Technologies, which specializes in high-availability solutions, says many companies are planning to migrate important parts of their infrastructure into the cloud.

A survey by Stratus, North Bridge Venture and Gigaom Research found that 75% of companies are using some form of cloud platform, in the 2013 statistics, The world's available cloud computing market will reach $158.8 billion trillion, a 126.5% increase from 2011.

"The enthusiasm for cloud computing continues to warm up, this enthusiasm transcends security concerns," Leclair said. "This year's findings show that the pace of cloud deployment continues to accelerate, and companies want suppliers to provide a consistently viable infrastructure to support their more critical business applications running in this new environment. ”

Leclair says companies need to carefully examine what applications they are putting in the cloud, and then consider the factors involved in managing this shift from the perspective of resources, skills, costs, and complexity. We know these considerations are comprehensive, and rewriting the cloud-appropriate applications is not the way to solve most problems.

Availability issues for mission-critical applications in the cloud

The value proposition that migration applies to the cloud seems clear, and this can greatly improve the flexibility and scalability of the application. In many cases, the enterprise's mission-critical applications are best able to benefit from the cloud infrastructure. But usability issues remain "misfortune".

The purpose of the cloud computing architecture is primarily scalability and resilience. A single cloud component may fail and cannot be replaced. If the application of the enterprise is unable to deal with these failure problems, it may face serious problems.

"We see a lot of basic apps migrating to Cloud computing," Leclair says. "We also see a lot of new applications built in the cloud, but we don't see the most important web apps moving around." ”

High cost of applying downtime to critical tasks

Downtime of critical tasks can cause business paralysis. For example, the software that United Airlines controls its ground operations had failed before Thanksgiving last year, resulting in a two-hour outage. This caused united passengers to delay or miss the flight. In addition, on Christmas Eve, the Amazon cloud computing services infrastructure failure caused Netflix failed to load media content to millions of users, so that users can not watch their favorite programs and movies.

Aberdeen Group Studies show that the average cost of downtime is now estimated at $138,888 per hour.

"Half of it decision-makers generally want to do less than 30 minutes of downtime per year," Leclair said, "but in reality they are far from reaching that level." ”

Eventually, Leclair points out, some applications may never migrate to the cloud because of the very high cost and risk involved. They will remain in bare-metal, virtualized, and non-cloud environments. These applications may require specialized hardware to address hardware or performance issues, or to lock these applications into a secure environment with some specification.

Organizations need to evaluate each application to determine if the application is best suited to the physical, virtualized, private cloud, public cloud, or mixed cloud environments. In each case, the enterprise should weigh the pros and cons.

Three important usability considerations

"Whether you're looking for a cloud or an enterprise's internal deployment environment, we recommend that you first evaluate the cost of applying downtime," he says. "In the case of public security applications, downtime costs can be measured in dollars, and reputational damage can even be calculated on the basis of loss of life." This allows you to understand the level of availability you actually need to better deploy your application. ”

If a business decides to move to a cloud service provider, Leclair recommends that the enterprise carefully examine the service level agreement (SLA) to determine what to do when the vendor's commitment to availability is not available.

"Some service-level agreements may say, ' We guarantee 100% uptime ', but if you look closely at the details of the contract, you may see, ' If we can't guarantee that, we'll return the 20% point in the next month's bill, ' Stratus's promise is to repay 50,000 dollars, However, the cost per hour of the most critical network application downtime is 150,000 dollars, so the offer is simply not worth mentioning. ”

Leclair also said companies need to consider data protection.

"Companies should protect applications at the transport level, but also need to consider other types of downtime that might occur-what if the tsunami overturned the office building?" or how often do I respond to data? "

"Edit Recommendation"

Public cloud migration is difficult cloud migration case: Harvard Business School Publishers Cloud migration annoying to crack key technical challenges how cloud migration affects existing data center Infrastructure "executive editor: Xiao Yun TEL: (010) 68476606"