Former MySQL CEO: Cloud will redefine everything about developers

The rules of software since its inception have not changed, that is, software running on the platform. However, the design and deployment of the underlying software is often changed, and even 8630.html ">" is sometimes completely changed.

For example, in the 80 's when the PC, x86 architecture, and Client/server model were born, the design of the application changed dramatically. The change is happening again in the middle of the http://www.aliyun.com/zixun/aggregation/16333.html ">90 era, caused by a burst of web and Open-source technology. Whenever such changes occur, developers are forced to rethink how they develop and deploy their software.

Figure: Marten Mickos

Now, again, we see a huge increase in infrastructure capabilities, and AWS is leading the industry. There is no doubt that AWS is a global leader in cloud infrastructure, and that running applications on AWS is naturally different from running applications in a private environment, even through virtualized private environments. But the design of cloud applications is significantly different from the past, both now and in the future.

Extension

In the past, extensions meant vertical scaling (scaling up), storing a lot of data, and you usually buy larger servers.

Now, the extension is typically scale-out (scaling out). You don't need to buy a bigger machine, you just need to add the same machine. In the cloud world, these machines are virtual machines.

Elastic

In the past, software was considered to be unreliable, and software resilience depended on the hardware layer.

Today, the underlying infrastructure is considered very fragile, and these virtual machines do not guarantee total availability, and applications must be prepared for this.

Netflix, for example, is a leading cloud user and is using this new architecture. They're running chaosmonkey.
, Chaosmonkey can easily turn off virtual machines for the underlying application load. Why is Netflix doing this? Because they need to ensure the virtual machine uptime and resilience: By eliminating the use of randomly generated fault virtual machines, they force developers to build more flexible applications.

Outbreak

In the past, consideration of statistics and the cost of all through the application, the application of the load is strong and predictable. It is easy to know how many users the system supports and how many records may be running at a certain point in time.

Today, the load becomes diverse and unpredictable. New software architectures are needed to meet the unpredictable complexity of a single application. We have now implemented this function in the cloud, but it is obviously very rudimentary.

Software diversity

In the past, there were not so many types of software. Each application is written in one language and uses a database. Companies tend to use only one or several operating systems.

In the present cloud world it is quite different. An application often uses many programming languages, calls many libraries and toolset, and uses many database products.

From virtual machine to cloud

Hypervisor is different from the current cloud computing considerations. The leading virtualization vendor VMware, the hypervisor is essentially a continuation of previous physical machines.

However, in the cloud world, the virtual machine is not the representative of the physical server, but the unit amount.

User Patience

In the past, users were as obedient as patients when they faced a doctor. The system takes a long time to handle a simple recovery or upgrade, and it is slow to add functionality to the application, sometimes without updating at all.

Today, users are not "patients." They can hardly tolerate delays or waiting, and they expect improvements in service providers every week. This evidence can be found in the enterprise's Self Service IT system, which provides flexibility to provide internal resource requirements.