Why is public cloud a rose with thorns?

Public cloud has always been a controversial and competitive market. Public cloud service providers, including IaaS, PaaS and SaaS services, have sprung up like mushrooms after a rain. In the absence of a finalized market structure, customers are confused about the choice of services. Let's follow the network communication editor to learn more about the relevant content.

Speaking of public cloud services and markets, we have to mention Amazon, the leader in the entire public cloud market, which is located in Seattle, Washington. Amazon was founded in 1995. Its cloud data centers use a central and edge architecture, with 28 central cloud data centers and 52 edge data centers. It currently provides public cloud services with no business differentiation experience to 11 regions around the world.

In the field of cloud computing and public cloud, many of Amazon's services have become de facto standards (mainly including S3 and EC2). The specifications of its cloud data center hardware facilities (such as data rooms, cooling, power supply, storage, servers, networks and monitoring, etc.) are also used as textbooks for reference by different cloud service providers. AWS service business covers the entire IaaS, PaaS and SaaS layers.

However, from the emergence of public cloud to today, the competition in the market has never diminished, but under the fierce competition and market baptism, the market structure of public cloud has become increasingly clear. Amazon's AWS, Microsoft's Azure, IBM's SoftLayer and Google's Cloud Platform lead the entire public cloud market, followed by Oracle Cloud public cloud services; Alibaba Cloud has also become the leader of domestic cloud services. The development of public cloud services and technologies and the increase in market share have also shrunk the market space for traditional IT hardware manufacturers. But from the current point of view, is the public cloud really reliable? How far is it from the reliability we need, including mature public cloud management specifications.

In fact, the public cloud has always been a controversial and highly competitive market. Public cloud service providers, including IaaS, PaaS and SaaS services, have sprung up like mushrooms after a rain. As the market structure has not yet been finalized, customers are also confused about their choice of services.

In 2013, public cloud service provider Nirvanix lost in competition, forcing nearly 1,000 customers to retrieve or migrate data originally stored in the Nirvanix cloud in a short period of time. Nirvanix's sudden closure brought considerable turbulence and impact to customers' businesses.

Now let’s talk about the reliability of public clouds. For example, Amazon, the largest cloud service provider, has reliability incidents almost every year. In March 2017, the S3 storage service in the US-EAST-1 region was interrupted, and Imgur, Medium, and Docker Registry Hub, which rely on S3 services, as well as Runkeeper, Trello, and Yahoo Webmail, were interrupted.

In March 2016, cloud service giant Amazon suffered another serious service outage, which lasted for 20 minutes and caused huge economic losses. When the incident occurred, Amazon's main e-commerce website was inaccessible, and other service industries were affected, including Amazon's cloud computing services for external services, and digital content services such as music provided to free shipping members.

In July 2015, Amazon Web Services (AWS) experienced a large-scale outage, causing many apps such as Slack, Asana, Netflix, Pinterest, and many websites using AWS services to become unresponsive.

In October 2012, Amazon's cloud computing platform in Northern Virginia data center crashed again, causing many well-known websites to be inaccessible for a long time. Well-known websites and information service providers were also affected by this outage. This was the fifth outage of Amazon's cloud computing in the past year and a half.

Microsoft Azure public cloud service also experienced a service outage on March 16, affecting 26 of the 28 data centers. Users in the affected areas were unable to create, update, or delete data resources. Fortunately, the Azure cloud service has returned to normal. In 2012, the Azure cloud service also experienced a major outage, which took half a day to restore normal service.

Alibaba Cloud also has frequent accidents. On the morning of July 6, 2016, Blue Whale TMT reported that the internal network of Alibaba Cloud's Beijing data center failed and it took nearly an hour to return to normal, causing a large number of Internet companies to be affected. Alibaba Cloud has had many failures. In September 2015, Alibaba Cloud's normal customer files were mistakenly isolated, resulting in all basic commands being unable to run. In June 2015, Alibaba Cloud also had an incident in Hong Kong, China where services were suspended for more than 12 hours. In fact, Alibaba Cloud had experienced many network failures before that.

All kinds of downtime and service interruptions show that shared cloud has big problems in reliability at this stage. Of course, users and service providers cannot avoid the problems of cloud computing itself, and cannot deny it completely because of a failure; but they should not over-praise cloud computing. In order to make public cloud better serve the public, shared cloud service providers can consider making improvements from technical means, and users should also re-examine what kind of business is suitable for shared cloud.

For cloud service providers, in addition to considering data backup and disaster recovery, and providing different levels of services for different businesses, they must also consider providing effective data emergency measures in the event of a major disaster or service suspension to ensure that customers can retrieve or migrate important data. They must use security technology to ensure the risk of data leakage. By providing professional service capabilities, they can formulate data migration strategies and priorities.

In addition, it is also necessary to consider how to find a new data home (new cloud service provider) for customers in the short term to store and protect the cloud data; this requires the cloud service provider to be relatively open first and to enable cross-cloud data sharing and data migration.

For enterprise and organizational customers, when deploying business on the public cloud and purchasing cloud services, they cannot rely on just one service provider. They need to involve multiple clouds to achieve better flexibility; ensure that data has multiple copies in multiple regions, while ensuring data reliability and enjoying the performance advantages of localized data access; it is recommended that data be connected to the public cloud through a hybrid cloud, and the public cloud is only used as a backup or non-critical business.

Public cloud services are indeed a good thing for enterprises and organizations. They can provide low-cost investment without maintenance and can also be invested and allocated on demand to meet different business requirements. However, everything has two sides, and public cloud is also a double-edged sword. To make better use of public cloud services, both customers and cloud service providers must bear their own responsibilities and obligations.

The above is the detailed introduction of the network communication editor of eeworld electronic engineering network on why public cloud is a rose with thorns. I hope that through the editor's explanation, it can bring new understanding to everyone. Pay attention to eeworld and electronic engineering, and you will be introduced to more knowledge about semiconductors.