English text:Google Is Challenging AWS How
Abstract: the first wave cloud computing boom, as the company's products Google made easy to use but not flexible App Engine, and as a platform for AWS is not very good but very flexible cloud computing services, Zhandexianji; now Google on the open source Kubernetes and AI provide a variety of services can regain a bureau?
Catch up withSuite G, not only grow faster, but also won the customer.
Nevertheless, in order to Microsoft and Office 365 success, the real giant cloud computing, that is, the future of enterprise computing, as usual, is full of unknowns: the same year, Google decided to use MicrosoftAmazon launched Web Services Amazon
The original nature of AWS
Earlier this year inAmazon Tax TheI explained how Amazon's AWS strategy is through the same approach that makes the company a success in the first:
The company consists of a number of relatively independent team organizations, each team has its own profit and loss, responsibility and distributed decision-making. [Everything Store TheAuthor Stone Brad] explains the early Bezos program:
Steven Sinofsky pointed out that love organizations tend to release their organization chart, although I began to suggest Amazon repeat AWS model, but the fact that the AWS model is the representative of the Amazon itself in many ways (like Apple's iPhone reflects the unified organization in many aspects): to create a stack of the original, then abandoned, as this nothing has happened.
Google is a product company
Platform companies and product companies are very different, just like the difference between the cloud services and hardware (I am in theHave been discussed in the middle. To create an ideal product, whether it is a smart phone or a search box, in order to give the user a good experience, both in the design and engineering efforts. But these efforts are the end users can not see. And the integration of the product is precisely the opposite, which is why Google's consumer centric servicesReason for integration at the back end, this is the same as iPhone.
IT timesExcellent platform company's Microsoft in the practice of API Win32 also like this. Although the Windows so designed to bring the ultimate user experience can not be compared to OS Mac, but it is more powerful, and can be extended. And AWS's flexibility and modularity has also become a major factor in breaking the Google (App Engine Google) in 2008. In addition, AWS has a clear advantage is that you can let you build your own needs, and the use of Engine App you need to accept some of the rules of Google.
Anti platform strategy of Google
When it comes to the transformation of Google mode, Windows provides an example of a guiding significance, that is, to build a large-scale ecological system around the API. Windows is a very strong place is that the application for the Windows build is not easy to be ported to other operating systems. Moreover, there is a huge business network between Windows and partners, it makes Windows become an essential tool for enterprises. Therefore, the Amazon is now moving in this direction.
For consumers and enterprise users, do not use Windows, now also is feasible, and the reason for this is that Web Technology: now there is a new operation, we can put it in the Windows to run, but it does not depend on the Windows system, and this kind of operation made the Google the biggest winner in the camp of consumers. In fact, the rise of the browser for the interpretation of AWS: any new business applications for the network (including the application of API operation based on network) construction, and can be accessed on any device.
Facts have proved that in the past few years, Google has used a browser approach to achieve enterprise computing.In 2014Google released Kubernetes, a Google based internal Borg service source container cluster manager, large-scale infrastructure extracted Google, so that all Google services can be instant access to all computing power they need, without worrying about the details. It is the focus of the container,I wrote in 2014Engineer to establish a standard interface, simply to retain its full flexibility, and do not need to know the underlying hardware and operating system related content.
The difference between Kubernetes and Borg is the full portability of it: it can be run in AWS and Azure, can run on the Google Cloud Platform infrastructure and internal deployment, and even run at home. More related to this paper is that it is a perfect start for ten years in the AWS infrastructure as a service of the correction: Although Google infrastructure in terms of their products has made great progress, but the particularity of kubernetes and based on the wide applicability of vessel development, make you no matter which is not the infrastructure provider effect of the use of AWS. No wonder it is one of the fastest growing projects: because there is no lock.
But how can you help Google? After all, even though Kubernetes has become the standard for enterprise cloud, Amazon's widespread ecosystem lock still exists (the company has its own container strategy to further lock customers into AWS); it looks like. Google must take different measures.
Cost and experience
The desktop also played a pointing role: the network is running on a platform independent browser, and its openness does not make Google a success. On the contrary, the opening of the Internet has created the winning conditions for the best technology. Google's advantage is not only because of the best search engine, but also because of its reliance on connection makes the network bigger and bigger, which Google do better than competitors.
I think this is an idea that can be widely used. As a matter of fact, it isPolymerization theoryWith the decrease of the cost of distribution (or exchange), the importance of the user experience is increased. In other words, when you can access to all services, whether it is news, car sharing, or video or search, the best way to win is not to win in the first advantage, or win after the advantages of composite.
Google in the cloud, said the company said: Kubernetes open source shows that Google is trying to effectively build a browser on the cloud infrastructure, thereby reducing the cost of exchange. The company and Search Google will be the nature of machine learning.
Machine learning and data
It is almost certain that machine learning will be more and more inclined to cloud services: both are dealing with large data. But only a few of the giants have the financial power to build the infrastructure and to hire the best machines to learn from the world's engineers. This means that the majority of enterprises from machine learning will be the first from their data whether differentiation in the cloud (there will be internal deployment solutions, but I want them to fall more and more time in the past), followed by their choice of cloud service providers.
This increases the risk of cloud computing vendors own excellent products; machine learning is sustainable: better products to attract more customers, so as to get more data, and the data is the raw material of machine learning. Because of the data, Google has become the biggest threat to AWS.
Google's biggest advantage is the large amount of data it has collected over the last twenty years, as well as a powerful machine learning algorithm developed over the last few years. The key to all of the problems lies in the data, last year, Google's open source tensorflow is the best example: as I am in theTensorFlow and monetization of intellectual property rightsSaid, Google is willing to share its way, is implicitly recognized its superior data processing infrastructure is a sustainable advantage.
We have been able to see the advantages of the application to Google cloud products, Google before ThanksgivingReleased a series of products, it can be clearly seen that the use of its own data advantage:
The first three API apparently originated in a variety of Google consumer products, API Jobs quad core and Google web property statistics, like the data, based on the creation of Google internal tools. These Google have spent several years to hone its algorithm, so that it is applied to the results of a corporate data set is likely to be better than or at least far less than the training funnel. I hope this advantage can persist, and produce more meaning.
However, Google had to do more, top AI expert Li Feifei, Li Jia led the team will publish the construction of machine learning and training intelligent application customization model cloud machine learning framework Google Learning: Cloud Machine, the team will be responsible for building a new machine learning business API, in other words, their task is to let Google machine learning to product.
Of course, AWS, IBM and Microsoft all have their own machine learning API. Microsoft may be particularly powerful in this regard: not only have many years of research and commercial product technology specific experience; and Google consumer long-term concern may become the obstacles, while the equally popular Kubernetes, can be said to beGoogle hasn't eaten its own dog food..
Anyway, Google will be a strong competitor: because it has a full strategy, and now to find a new business line is far more pressing than in 2006. And most importantly, it has only just begun to move to the cloud, and Amazon seems to have developed into the distant future, the future is not yet known, it will look at Google trying to change the rules, it must be very interesting.