Unlock the 7 mysteries of cloud computing (2)

5. Cloud computing is no big deal

Veil: Cloud computing is just another way of saying virtual computing, which is not unusual.

For the definition of cloud computing, it seems that it is out of place to raise objections here. Cloud computing is not just about virtual computing environments. Cloud computing is the task of using a virtual computing environment and other technologies to accomplish what a single technology can't do.

This misunderstanding may be due to misinterpretations by some cloud service providers when explaining cloud computing to developers. For example, Amazon's EC2 is described as a practical system with fixed storage, memory, and computing power. The reason for this explanation is simple: in order to make developers better understand, cloud computing is a combination of different hardware that they are already familiar with.

But this does not mean that cloud computing is always just an aggregation of different hardware. With the continuous development of cloud computing technology and the development of distributed/parallel computing, there will be better ways to describe the services provided by cloud computing, and will no longer be limited to the combination of known hardware. In the future cloud computing platform may be like this, from the programming language to the development environment itself can understand that the resulting code will be deployed in a flexible manner.

Among the voices of many opponents is Oracle's Larry Ellison, mostly because of the fact that for him, cloud computing and Oracle's other business models and computing models (such as SaaS or IaaS) draw a clear line. But this did not prevent Oracle from creating its own "application network" system, which allows Oracle's database to be deployed to Amazon EC2 cloud services. Obviously, Oracle is not very resistant to cloud services, he just wants to choose his own way. Who is not like this?

6. Network and storage restrictions

Veil: Cloud computing has limitations in terms of network and storage

This question depends entirely on the size and scope of your deployment of cloud services, the I/O bindings, the scalability of the programs involved, and the process of your deployment.

I/O inside and outside the cloud computer is still the main bottleneck, unless we can achieve mega-fiber connection speed, and soon after Godot is released, it can reach this speed. At the same time, the smartest thing to do is to build your own cloud infrastructure, including the cloud front end and back end, so you only need to move the necessary data to the cloud. Otherwise, you can only move jobs that need to be done on the data, and this obviously has more computational and programming difficulties, which of course is more flexible than moving all the data.

On the other hand, there is no consideration, that is, the limitation of network speed on the back end and the restriction on the connection of the client to the server. The bandwidth limitation on the back end is not as much as the customer's limit, but the data density grows faster than the network speed, we will face the PB level running synchronously on multiple hosts.

The solution we need is not a breakthrough in bandwidth, but a better use of existing networks, such as through the use of innovative technologies such as low-bandwidth file systems. The most important thing is that you must try your own efforts first.

These questions should lead to more discussion on how to make the most efficient use of the network and what types of customers can be paired in the cloud as a server/backend.

7. Scalability

Veil: It is difficult to take advantage of the scalability of cloud computing

There is a common complaint among programmers that cloud computing is difficult to program effectively. Cloud computing can be very easy to run and provide functionality, but when it comes to scale issues, it's another matter. Ask the Twitter staff to know that they are starting to build Ruby on Rails applications and are gradually rewriting the services in Scala.

This part of the problem mainly comes down to the parallel programming ability that is still in its infancy. Writing applications or algorithms across multiple systems is very difficult. The most common languages ​​(like PHP) are not designed for extensions, but to be able to deploy features quickly and well. The aforementioned Scala is a language specifically designed to scale programs. The interoperability of Scala and Java means that Java programmers can do a lot less now. That is to say, learning another language has never been a simple matter, and these complaints from programmers reflect the need for cloud computing to actively improve.

Scale control of the most urgently needed operations is more cumbersome because most operations are not scalable.

Yahoo's Apache Software Foundation's Hadoop and Google's MapReduce technologies all access and aggregate results from large data. These data are distributed across different nodes, sending calculations to these different nodes and then returning the results.

However, MapReduce can only be run in certain operating environments and is not a comprehensive solution, so it is often necessary to adjust the problem based on existing solutions. Another solution is Swarm, an application framework that allows calculations rather than data to be scaled at different nodes. This concept and mode of operation is still in its infancy, but it is worthy of attention.

Unraveling the mystery of these cloud computing, it is king to deploy cloud computing services suitable for your own business according to the specific situation.

Previous page

3D Panel Fence

3D Panel Fence

3D Panel Fence,3D Mesh Fence,3D Wire Fence,3D Wire Mesh Fence

HeBei Bosen Metal Products Co.,Ltd , https://www.hbbosenfence.com