Are developers needed in the future?
It is time for entertaining stories. To begin with, I propose to stock up on cookies and a large mug of toxication fluid. All taken? I present you an interview with Dmitry Zavalishin, the founder of the DZ Systems group of companies, and Alexander Lozhechkin, the head of the Microsoft strategic technologies department in Central and Eastern Europe. In it, you will learn what Microsoft has become today, how we are competing with Amazon, and the most interesting thing that will happen to programmers is if artificial intelligence enslaves the world. Under the cut, you will find its text decoding.
Dmitry Zavalishin is a Russian programmer, author of the concept of OS Phantom, organizer and member of the OS Day conference program committee, founder of the DZ Systems group of companies. In 1990–2000 he took an active part in the creation of the Russian segments of the Internet (Relcom) and Fidonet, in particular, ensured the transparent interaction of these networks. In 2000-2004, he was responsible for the design, development and development of the Yandex company portal, created the Yandex.Guru service (hereinafter Yandex.Market). You can read more on the Wiki .
Alexander Lozhechkin - Head of Microsoft Strategic Technologies in Central and Eastern Europe, Member of the Board of Directors of the company. In his free time, he maintains a personal blog on Medium , in which he reflects on various topics, ranging from evaluating the work of employees to philosophical discourses about delicious mediocrity.
We are accustomed to the fact that Microsoft is a technologically highly closed global monster, which, in general, is quite tough on technology synergies and seeks to completely drag people into its picture who have begun to use them. It seems that today is not quite so, and Microsoft has changed quite a bit. Alexander, tell me about it, please.
I have been working at Microsoft for 14 years, and have watched many stages in the transformation of a company. Indeed, for some time it was rather closed and focused on its technologies, but this has changed a long time ago. We love “non-Microsoft” technologies and understand that a world built on the technologies of one company will never exist.
Therefore, the winner is not the one who makes the technology that can capture the whole world, but the one who can best integrate with the technologies of other companies, manufacturers, communities. Long ago we set ourselves the goal of becoming a company that is most open and integrates best with other technologies. Open source, java, anything, everything is working with us, with us and we are trying to develop these areas.
If you look at the structure of Microsoft products, you can see that the company has always been, conditionally, at the bottom of the stack: operating systems, compilers, DBMS tools, in other words, tools that lie behind the programmer. And the topic of digital transformation, conventionally, is at the other end of this scale. It is more business and concerns the ability to apply a set of IT-tools to the real business. Why are you there? And in what format are you there?
Microsoft is really a platform company. We make and continue to make a lot of products that, as you said, are at the bottom of the stack - these are our cloud technologies, compilers, database programming languages.
In addition, we make productivity products on the basis of these products, for example, Office. Although one might argue here, is Office a product or a platform? Because you can expand it to infinity. Plus, we have a large line of business solutions - Dynamics. In principle, all parts of the stack are closed, but digital transformation, it seems to me, can work at any of these levels.
Here I would recall a rather distinctive attribute of all Microsoft products and technologies - their availability. Microsoft was not the company that invented the PC, but Microsoft made it so that the PC was on every desktop. You can blame Microsoft for this, you can not think of Microsoft technology as perfect, but the fact that they were able to spread around the world is a fact.
Now we are trying to do the same with the technologies of artificial intelligence, which are often behind the decisions in the digital transformation. Our goal is not so much to make them the most perfect (although we are also trying to do this), our goal is the democratization of AI, the democratization of information technologies. We want to make the cloud available to any company on the market, to make artificial intelligence and ML technologies available to any company. This will allow companies to transform very quickly.
And what does this mean practically? Roughly speaking, if we consider a developer who has set himself the task of providing the IT component of some new or traditional business, why would he come to Microsoft today? And what would you recommend him to consider first? What building blocks and which bricks could he use? (On the vskidku)
With us, I would first of all start with cloud technologies, since this is now what really changes the world around. I love to use here such an analogy with the famous quote that God made people, and Mr. Colt made them equal. I shift it to business: God created companies, and cloud technologies make them equal. Thanks to cloud technologies, the capabilities of computers that previously were available only to large companies are made available to any company, be it a small startup or company in a small or medium business. Today there is no need to build datacenters, invest huge resources in their support, and so on. In the cloud, you can easily use the latest technologies, the newest algorithms ML, the same AI.
Therefore, to your question about bricks, I would recommend using cloud technologies. It is worth noting that our approach here is not that we dictate our own rules, what a cloud should be; we give companies the opportunity to choose a stack suitable for them. You can use our cloud technologies in public clouds, that is, in our data centers, or you can, with a high degree of symmetry, build a similar infrastructure in your data center. But, in any case, it all revolves around the clouds, and then there appear ML and AI technologies, such as, for example, the Cognitive Toolkit.
That is, these are some software components that a developer can use. And if he is deploying his application in the Microsoft cloud, then these components exist for him in the finished form?
Yes, that's right. And here the use of technologies of other companies is not canceled in any way. For example, the same TensorFlow works fine on virtual machines in the Microsoft Azure cloud. There are no restrictions and there are a large number of applications of third-party services. And this is probably the key attribute of what Microsoft has become now. We do not impose a choice, we provide it for everyone.
How do you compete with Amazon? After all, the Amazon cloud occupies quite strong positions. I think he was almost the first in this market. But it appeared a long time ago, at least. You still came a little later. Do you have price competition? Functional competition? What makes you significantly different from the strongest competitor?
I would probably have addressed this issue a little wider and started with what we understand by cloud technologies. For example, the same hosting providers call themselves cloud providers, and this is probably true. There are several players - Microsoft, Amazon and Google, which are present all over the world, and there are a large number of slightly smaller companies hosting companies, which also, in general, offer cloud technologies may not be so developed, not so serious, not so big. the number of implementations and so on, but still it is cloud technology.
And if we look at this market more broadly, it turns out that, firstly, there is no one player in any dominant position, and the market is very fragmented. And secondly, the market is growing very, very quickly, so whoever is the leader of this market today, whatever definitions we use, is not a fact that it will be like this because the market is growing significantly every year. For us the main thing here is to grow at a fast pace to how the market grows. Returning to your question, what is our difference ...
Differentiation with Amazon - we try to keep up, first. And if there appear any new technologies, we are trying to implement them in our country. We are trying to offer something that our main competitors do not have, and here again, I would not forget not only about Amazon, but also about Google, and about everyone else.
That is, the ML components for you are such a distinguishing advantage?
Initially, in general, Amazon has emerged as a service of providing virtual machines. Infrastructure as a Service (IaaS) is what they started with, plus storage. From the very beginning, we installed Platform as a Service (PaaS). That is, we offered not just basic virtual machines, but some services built from above. And still, this is a very big focus for us.
A little later, Amazon entered the basic infrastructure market, we had virtual machines a little later, but now it can be said to be comparable. This is precisely the area in which we are trying to keep up and, maybe, in something, to be ahead. Our main focus is PaaS and AI as one of the components.
And here I may be, not so much from the point of view of an employee of Microsoft, but in general from the point of view of a person who is interested in the topic, said how interesting the competition has now moved from the field of software to the area of ​​hardware.
Microsoft recently introduced a project that was once a Microsoft Research project on the use of programmable matrices for ML algorithms. It has already been deployed at many of our data centers. This is a thing that allows you to speed up typical ML algorithms very much. And an example that we showed not so long ago was when we translated from the Russian language into English on the “War and Peace” scene, and it turned out that if we use the same algorithms on the underlying virtual machines, this, conditionally speaking, takes a day . When used in programming matrices, it takes seconds or minutes. That is, there is an increase hundreds of times, and it seems to me (again, from my own particular point of view) that the competition will continue to go where various providers of cloud services will compete in providing something unique that is available only from them, and this means that not only software, but also hardware.
That is, roughly speaking, this means that in those cogs that you provide, there are these subsystems available for software. At the same time, I understand correctly, maybe we will bury too deeply now, but in reality it is very interesting. Roughly speaking, if we take a standard neural network, which is a set of layers, then part of the layers physically degenerates into hardware components that AI cheats?
Yes, that's right. And due to this it turns out just a dramatic increase in productivity.
Not surprising. This, of course, is a very serious application.
This is the first moment by which we try to differentiate ourselves, and the second moment is what I have already said: we originally had hybrid clouds. We are not only talking about public clouds, like our competitors, we are saying that clouds should be in private data centers, or at a hosting provider, or in data centers, and we try to maintain a very high degree symmetry between what is available in public clouds and what the customer can deploy.
And in the FPGA can deploy?
No, here in the FPGA can not. Naturally, there will never be full symmetry, but at least we are trying to ensure the highest possible degree of symmetry, and this is exactly what a product like the Azure Stack is aimed at - when you deploy an infrastructure in your data center that allows you to without rewriting, without changes, without adaptation, deploy a large number of applications both in the public cloud and in your own data center.
And of course they can be software integrated?
Yes, they are integrated, and it turns out that a public cloud can be such an extension of a private one. When you keep something in your home, say, you cannot transfer any data to a public cloud. And the second cloud is needed in order to very quickly increase and scale the load.
In this case, quite often there is such a fairly obvious topic as public and closed sharding. That is, roughly speaking, we want to store the data of this user here publicly or in this country, and the data of this one is either non-public or simply in another country. Is there any support for this?
There are technologies that allow us to do these things, and also, we have many partner solutions, for example, from startups. One of the options for implementing this is when the data that you do not want to give is replaced with some conditional hashtags and, accordingly, only these hashtags are placed into the public cloud; Then there are technologies that allow you to simply mirror some data, that is, separate them by identifier and separate them by public and private storages.
This suggests a rather obvious picture, when the cloud would suggest a conditional interface to the database, into which you can embed a predicate that describes, in fact, where this data falls. But this probably does not exist.
Yes, this is not basic functionality, but this is something that can actually be implemented, and this is just one of the goals of creating hybrid clouds - this is when opportunities are divided, for example, according to the degree of sensitivity of data to movement.
I want to ask also about the ML technologies that you offer. After all, if I correctly understand the current situation in this market, then it looks like this: there is a fairly wide range of developments, and, in essence, the success of a particular team that makes a real decision based on ML solutions is to choose the right build these developments. And the spectrum is quite large. Probably, if they turn to your technology, then this range is somehow narrowed? Or does not taper? Or you can build Microsoft solutions with solutions from other vendors in this place?
Yes, we just have a set of ready-made blocks. If you want to deploy Google's TensorFlow from Azure, please do it, it's all possible, it all works. Here, our approach is the same as it was once with a PC, our task is to make it accessible to everyone, that is, we are democratizing artificial intelligence technologies. As often happens, the Pareto law has not been canceled - the problem is solved with the help of 20% of efforts. And we try to give such technologies, which even people who are not superspecialists in AI technologies could use.
Here I have an example of our customer, this is a company engaged in baking bread. They had a very big problem: bread is a product that needs to be consumed fresh, and they have written off a lot of bread. Since the worst thing that any commercial company can have is unmet demand, they always brought back with a stock, and then scrapped up to 20%, because it was stale, and it was a direct loss.
To solve this problem, we applied Azure ML - this is a product in which, literally, you can leave the data with the mouse, test the model, then he learns to calculate this model, and then we check this model on the control sample, see what this model is learned In other words, this is a product that is very easy to use and teach it to anyone, well, the programmer will precisely master it in one day.
For example, no one on the customer’s team was an ML specialist, or Deep Neural Network specialist. There were just ordinary techies who after a week began using this tool. And they managed to reduce losses by half, simply because they learned how to correctly calculate the amount of bread to be brought to each bakery every day based on previous data.
Now you say terrible things, in fact! Terrible! After all, in fact, pay attention, a new elite of programmers has already begun to form. A very clear wave happened: there are programmers who are familiar with the ML theme, and there are programmers who have not yet crawled. I often hear on this topic of discussion. For example, when the first cars appeared, and all had horses, those who drive the horse are no longer needed, and drivers appeared. And now you say that Microsoft brought in this place a steering robot for a car before people learned to control it with their hands.
This is a good analogy, very beautiful. But what's so scary? This is wonderful!
Why, so many people in this place will lose their jobs ...
There was just a discussion on this topic in which I participated. My point of view on this matter is quite radical, but I believe that even if robots can replace people in some professions, this will not lead to any dramatic consequences. I call it the time of the new antiquity, as in my time in ancient Greece. True, thanks to the not very ethical use of slave labor, the citizens of the Greek poles were able not to do hard work from morning to night, but were able to do art, philosophy, self-knowledge. Why not expect the same from robots now?
That is, thanks to Microsoft, we will all enter such an era, when we drink wine, sitting at a table with fruit, and next to the wall computers controlled by artificial intelligence will steer by plants, create wine for us and grow fruit for us?
Here I was rightly objected to such an analogy that we can all be a society like in ancient Greece, engage in art and philosophy, or, in fact, corrupted by Nero in ancient Rome. Therefore, it only depends on us, and it will be interesting to understand what kind of society we have: the one that is capable of becoming the society of the new antiquity, or it will be the society of the new corrupted system of Patricia.
As always, with any instrument, the effect depends on the people who use it. It's clear. But let's go back to the subject of the use of tools that Microsoft has now appeared in real life. We started by saying that there are several cases of digital transformation in which Microsoft participated, and they showed good results.
, , «» , , , . , , , : , , . , , , , . 50% ! . - ?
, . , , , - , , , . , , , , .
, – Sarafan, ML , -, . , Instagram - , , ML , . , , .
, , . , , ML .
, : ML. , , , .
? , , Microsoft , , , , .
, , . , - , , ( «1984»), - . , . , , , , .
: (), - , ( ). , , . , , .
… , , : , . . , , ML: – ?
Yes.
, , Agile. , . ? . ? , - , - , ?
, . , , , , . , ! , , , , . KPI , , , . ! , , , , , . , ? , , - , .
, .
, - . , , – , , .
, FPGA ? ?
FPGA – , - . , , , , .
ML?
, , , , , 1000 . , .
private cloud FPGA , - ? ?
, , , . - , , . , , . , , , , ? . , , … Microsoft , , .
, ML . - .
, Microsoft , . Intel- ARM , - , , , . , Windows ARM, ? , ?
Windows Snapdragon, ARM. , . , , win32 , Windows ARM, , Windows.
?
, , ARM , . , , , – , .
Why?
ARM , , , , . , ARM , , , Windows ARM – .
, x86 native , Windows, , , ?
, , . :
— -?
— Yes, of course.
— .
— , .
, windows .
, ? ARM x86, , , .
, . , : java-script x86, , java-script, Linux . ! , ! , , , , , . , , , , , win32 … -, . , , Apple , , !
, win32 . , , ( ) Cobol- . . , win32 , , Microsoft … . , win32 !
, legacy , , , , , - , . - digital transformation.
, , , , . , - , ?
, , , . ( ), . – , . , , – .
. DoDo Pizza. , , , - .
, ( . , « ») - . , «, , - , , , , . , -, : „ : -, , IT-, - , , IT?“ – . - , .
, IT. ? , , , !
, , , ( . — «. ») . , . IT-, – IT-, ? , . , , , . – , . , Uber, Airbnb – .
, , – Microsoft – , , Yota Devices.
YotaPhone?
, YotaPhone, - Microsoft, . «ICEBERG», ML . , . , , . , : , , . ( ), , - - , : , , . , , , .
.
, , . Cortana (Cortana, , , — , Windows, Siri, OK Google).
?
, . Cortana 15 16 .
, -, « » ? ?
, , … , , . , , , , . , . , , , .
( , , !). . , ., , . , - Retail, , , Retail.
, . , , 10 . , , ! -, , -, . . ? ? , - -?
, , . ., – Amazon, , Amazon ., , . , 30% , 70% , . . Amazon : 70% 30% . , . , commodity service (), , .
, , , , . ! , , , , , . , , Kickstarter. crowdfunding-, Amazon , , ( ) . .
, , retail- , retail , , retail- , , , , - .
, , , , , , , , , , ! Thank you very much! Microsoft – , !
, ! !
We remind you that you can try Azure for free .
Minute advertising . If you want to try new technologies in your projects, but do not reach the hands, leave the application in the program Tech Acceleration from Microsoft. Its main feature is that together with you we will select the required stack, we will help to realize the pilot and, if successful, we will spend maximum efforts so that the whole market will know about you.
Series of Digital Transformation articles
Technological articles:
1. Start .
2. Blockchain in the bank .
3. We learn the car to understand human genes .
4. Machine learning and chocolates .
5. Loading ...
')
A series of interviews with Dmitry Zavalishin on DZ Online :
1. Alexander Lozhechkin from Microsoft: Do we need developers in the future?
2. Alexey Kostarev from “Robot Vera”: How to replace HR with a robot?
3. Fedor Ovchinnikov from Dodo Pizza: How to replace the restaurant director with a robot?
4. Andrei Golub from ELSE Corp Srl: How to stop spending a lot of time on shopping trips?
Who is the interview with?
Dmitry Zavalishin is a Russian programmer, author of the concept of OS Phantom, organizer and member of the OS Day conference program committee, founder of the DZ Systems group of companies. In 1990–2000 he took an active part in the creation of the Russian segments of the Internet (Relcom) and Fidonet, in particular, ensured the transparent interaction of these networks. In 2000-2004, he was responsible for the design, development and development of the Yandex company portal, created the Yandex.Guru service (hereinafter Yandex.Market). You can read more on the Wiki . Alexander Lozhechkin - Head of Microsoft Strategic Technologies in Central and Eastern Europe, Member of the Board of Directors of the company. In his free time, he maintains a personal blog on Medium , in which he reflects on various topics, ranging from evaluating the work of employees to philosophical discourses about delicious mediocrity.Interview
We are accustomed to the fact that Microsoft is a technologically highly closed global monster, which, in general, is quite tough on technology synergies and seeks to completely drag people into its picture who have begun to use them. It seems that today is not quite so, and Microsoft has changed quite a bit. Alexander, tell me about it, please.
I have been working at Microsoft for 14 years, and have watched many stages in the transformation of a company. Indeed, for some time it was rather closed and focused on its technologies, but this has changed a long time ago. We love “non-Microsoft” technologies and understand that a world built on the technologies of one company will never exist.
Therefore, the winner is not the one who makes the technology that can capture the whole world, but the one who can best integrate with the technologies of other companies, manufacturers, communities. Long ago we set ourselves the goal of becoming a company that is most open and integrates best with other technologies. Open source, java, anything, everything is working with us, with us and we are trying to develop these areas.
If you look at the structure of Microsoft products, you can see that the company has always been, conditionally, at the bottom of the stack: operating systems, compilers, DBMS tools, in other words, tools that lie behind the programmer. And the topic of digital transformation, conventionally, is at the other end of this scale. It is more business and concerns the ability to apply a set of IT-tools to the real business. Why are you there? And in what format are you there?
Microsoft is really a platform company. We make and continue to make a lot of products that, as you said, are at the bottom of the stack - these are our cloud technologies, compilers, database programming languages.
In addition, we make productivity products on the basis of these products, for example, Office. Although one might argue here, is Office a product or a platform? Because you can expand it to infinity. Plus, we have a large line of business solutions - Dynamics. In principle, all parts of the stack are closed, but digital transformation, it seems to me, can work at any of these levels.
Here I would recall a rather distinctive attribute of all Microsoft products and technologies - their availability. Microsoft was not the company that invented the PC, but Microsoft made it so that the PC was on every desktop. You can blame Microsoft for this, you can not think of Microsoft technology as perfect, but the fact that they were able to spread around the world is a fact.
Now we are trying to do the same with the technologies of artificial intelligence, which are often behind the decisions in the digital transformation. Our goal is not so much to make them the most perfect (although we are also trying to do this), our goal is the democratization of AI, the democratization of information technologies. We want to make the cloud available to any company on the market, to make artificial intelligence and ML technologies available to any company. This will allow companies to transform very quickly.
And what does this mean practically? Roughly speaking, if we consider a developer who has set himself the task of providing the IT component of some new or traditional business, why would he come to Microsoft today? And what would you recommend him to consider first? What building blocks and which bricks could he use? (On the vskidku)
With us, I would first of all start with cloud technologies, since this is now what really changes the world around. I love to use here such an analogy with the famous quote that God made people, and Mr. Colt made them equal. I shift it to business: God created companies, and cloud technologies make them equal. Thanks to cloud technologies, the capabilities of computers that previously were available only to large companies are made available to any company, be it a small startup or company in a small or medium business. Today there is no need to build datacenters, invest huge resources in their support, and so on. In the cloud, you can easily use the latest technologies, the newest algorithms ML, the same AI.
Therefore, to your question about bricks, I would recommend using cloud technologies. It is worth noting that our approach here is not that we dictate our own rules, what a cloud should be; we give companies the opportunity to choose a stack suitable for them. You can use our cloud technologies in public clouds, that is, in our data centers, or you can, with a high degree of symmetry, build a similar infrastructure in your data center. But, in any case, it all revolves around the clouds, and then there appear ML and AI technologies, such as, for example, the Cognitive Toolkit.
That is, these are some software components that a developer can use. And if he is deploying his application in the Microsoft cloud, then these components exist for him in the finished form?
Yes, that's right. And here the use of technologies of other companies is not canceled in any way. For example, the same TensorFlow works fine on virtual machines in the Microsoft Azure cloud. There are no restrictions and there are a large number of applications of third-party services. And this is probably the key attribute of what Microsoft has become now. We do not impose a choice, we provide it for everyone.
How do you compete with Amazon? After all, the Amazon cloud occupies quite strong positions. I think he was almost the first in this market. But it appeared a long time ago, at least. You still came a little later. Do you have price competition? Functional competition? What makes you significantly different from the strongest competitor?
I would probably have addressed this issue a little wider and started with what we understand by cloud technologies. For example, the same hosting providers call themselves cloud providers, and this is probably true. There are several players - Microsoft, Amazon and Google, which are present all over the world, and there are a large number of slightly smaller companies hosting companies, which also, in general, offer cloud technologies may not be so developed, not so serious, not so big. the number of implementations and so on, but still it is cloud technology.
And if we look at this market more broadly, it turns out that, firstly, there is no one player in any dominant position, and the market is very fragmented. And secondly, the market is growing very, very quickly, so whoever is the leader of this market today, whatever definitions we use, is not a fact that it will be like this because the market is growing significantly every year. For us the main thing here is to grow at a fast pace to how the market grows. Returning to your question, what is our difference ...
Differentiation with Amazon - we try to keep up, first. And if there appear any new technologies, we are trying to implement them in our country. We are trying to offer something that our main competitors do not have, and here again, I would not forget not only about Amazon, but also about Google, and about everyone else.
That is, the ML components for you are such a distinguishing advantage?
Initially, in general, Amazon has emerged as a service of providing virtual machines. Infrastructure as a Service (IaaS) is what they started with, plus storage. From the very beginning, we installed Platform as a Service (PaaS). That is, we offered not just basic virtual machines, but some services built from above. And still, this is a very big focus for us.
A little later, Amazon entered the basic infrastructure market, we had virtual machines a little later, but now it can be said to be comparable. This is precisely the area in which we are trying to keep up and, maybe, in something, to be ahead. Our main focus is PaaS and AI as one of the components.
And here I may be, not so much from the point of view of an employee of Microsoft, but in general from the point of view of a person who is interested in the topic, said how interesting the competition has now moved from the field of software to the area of ​​hardware.
Microsoft recently introduced a project that was once a Microsoft Research project on the use of programmable matrices for ML algorithms. It has already been deployed at many of our data centers. This is a thing that allows you to speed up typical ML algorithms very much. And an example that we showed not so long ago was when we translated from the Russian language into English on the “War and Peace” scene, and it turned out that if we use the same algorithms on the underlying virtual machines, this, conditionally speaking, takes a day . When used in programming matrices, it takes seconds or minutes. That is, there is an increase hundreds of times, and it seems to me (again, from my own particular point of view) that the competition will continue to go where various providers of cloud services will compete in providing something unique that is available only from them, and this means that not only software, but also hardware.
That is, roughly speaking, this means that in those cogs that you provide, there are these subsystems available for software. At the same time, I understand correctly, maybe we will bury too deeply now, but in reality it is very interesting. Roughly speaking, if we take a standard neural network, which is a set of layers, then part of the layers physically degenerates into hardware components that AI cheats?
Yes, that's right. And due to this it turns out just a dramatic increase in productivity.
Not surprising. This, of course, is a very serious application.
This is the first moment by which we try to differentiate ourselves, and the second moment is what I have already said: we originally had hybrid clouds. We are not only talking about public clouds, like our competitors, we are saying that clouds should be in private data centers, or at a hosting provider, or in data centers, and we try to maintain a very high degree symmetry between what is available in public clouds and what the customer can deploy.
And in the FPGA can deploy?
No, here in the FPGA can not. Naturally, there will never be full symmetry, but at least we are trying to ensure the highest possible degree of symmetry, and this is exactly what a product like the Azure Stack is aimed at - when you deploy an infrastructure in your data center that allows you to without rewriting, without changes, without adaptation, deploy a large number of applications both in the public cloud and in your own data center.
And of course they can be software integrated?
Yes, they are integrated, and it turns out that a public cloud can be such an extension of a private one. When you keep something in your home, say, you cannot transfer any data to a public cloud. And the second cloud is needed in order to very quickly increase and scale the load.
In this case, quite often there is such a fairly obvious topic as public and closed sharding. That is, roughly speaking, we want to store the data of this user here publicly or in this country, and the data of this one is either non-public or simply in another country. Is there any support for this?
There are technologies that allow us to do these things, and also, we have many partner solutions, for example, from startups. One of the options for implementing this is when the data that you do not want to give is replaced with some conditional hashtags and, accordingly, only these hashtags are placed into the public cloud; Then there are technologies that allow you to simply mirror some data, that is, separate them by identifier and separate them by public and private storages.
This suggests a rather obvious picture, when the cloud would suggest a conditional interface to the database, into which you can embed a predicate that describes, in fact, where this data falls. But this probably does not exist.
Yes, this is not basic functionality, but this is something that can actually be implemented, and this is just one of the goals of creating hybrid clouds - this is when opportunities are divided, for example, according to the degree of sensitivity of data to movement.
I want to ask also about the ML technologies that you offer. After all, if I correctly understand the current situation in this market, then it looks like this: there is a fairly wide range of developments, and, in essence, the success of a particular team that makes a real decision based on ML solutions is to choose the right build these developments. And the spectrum is quite large. Probably, if they turn to your technology, then this range is somehow narrowed? Or does not taper? Or you can build Microsoft solutions with solutions from other vendors in this place?
Yes, we just have a set of ready-made blocks. If you want to deploy Google's TensorFlow from Azure, please do it, it's all possible, it all works. Here, our approach is the same as it was once with a PC, our task is to make it accessible to everyone, that is, we are democratizing artificial intelligence technologies. As often happens, the Pareto law has not been canceled - the problem is solved with the help of 20% of efforts. And we try to give such technologies, which even people who are not superspecialists in AI technologies could use.
Here I have an example of our customer, this is a company engaged in baking bread. They had a very big problem: bread is a product that needs to be consumed fresh, and they have written off a lot of bread. Since the worst thing that any commercial company can have is unmet demand, they always brought back with a stock, and then scrapped up to 20%, because it was stale, and it was a direct loss.
To solve this problem, we applied Azure ML - this is a product in which, literally, you can leave the data with the mouse, test the model, then he learns to calculate this model, and then we check this model on the control sample, see what this model is learned In other words, this is a product that is very easy to use and teach it to anyone, well, the programmer will precisely master it in one day.
For example, no one on the customer’s team was an ML specialist, or Deep Neural Network specialist. There were just ordinary techies who after a week began using this tool. And they managed to reduce losses by half, simply because they learned how to correctly calculate the amount of bread to be brought to each bakery every day based on previous data.
Now you say terrible things, in fact! Terrible! After all, in fact, pay attention, a new elite of programmers has already begun to form. A very clear wave happened: there are programmers who are familiar with the ML theme, and there are programmers who have not yet crawled. I often hear on this topic of discussion. For example, when the first cars appeared, and all had horses, those who drive the horse are no longer needed, and drivers appeared. And now you say that Microsoft brought in this place a steering robot for a car before people learned to control it with their hands.
This is a good analogy, very beautiful. But what's so scary? This is wonderful!
Why, so many people in this place will lose their jobs ...
There was just a discussion on this topic in which I participated. My point of view on this matter is quite radical, but I believe that even if robots can replace people in some professions, this will not lead to any dramatic consequences. I call it the time of the new antiquity, as in my time in ancient Greece. True, thanks to the not very ethical use of slave labor, the citizens of the Greek poles were able not to do hard work from morning to night, but were able to do art, philosophy, self-knowledge. Why not expect the same from robots now?
That is, thanks to Microsoft, we will all enter such an era, when we drink wine, sitting at a table with fruit, and next to the wall computers controlled by artificial intelligence will steer by plants, create wine for us and grow fruit for us?
Here I was rightly objected to such an analogy that we can all be a society like in ancient Greece, engage in art and philosophy, or, in fact, corrupted by Nero in ancient Rome. Therefore, it only depends on us, and it will be interesting to understand what kind of society we have: the one that is capable of becoming the society of the new antiquity, or it will be the society of the new corrupted system of Patricia.
As always, with any instrument, the effect depends on the people who use it. It's clear. But let's go back to the subject of the use of tools that Microsoft has now appeared in real life. We started by saying that there are several cases of digital transformation in which Microsoft participated, and they showed good results.
, , «» , , , . , , , : , , . , , , , . 50% ! . - ?
, . , , , - , , , . , , , , .
, – Sarafan, ML , -, . , Instagram - , , ML , . , , .
, , . , , ML .
, : ML. , , , .
? , , Microsoft , , , , .
, , . , - , , ( «1984»), - . , . , , , , .
: (), - , ( ). , , . , , .
… , , : , . . , , ML: – ?
Yes.
, , Agile. , . ? . ? , - , - , ?
, . , , , , . , ! , , , , . KPI , , , . ! , , , , , . , ? , , - , .
, .
, - . , , – , , .
, FPGA ? ?
FPGA – , - . , , , , .
ML?
, , , , , 1000 . , .
private cloud FPGA , - ? ?
, , , . - , , . , , . , , , , ? . , , … Microsoft , , .
, ML . - .
, Microsoft , . Intel- ARM , - , , , . , Windows ARM, ? , ?
Windows Snapdragon, ARM. , . , , win32 , Windows ARM, , Windows.
?
, , ARM , . , , , – , .
Why?
ARM , , , , . , ARM , , , Windows ARM – .
, x86 native , Windows, , , ?
, , . :
— -?
— Yes, of course.
— .
— , .
, windows .
, ? ARM x86, , , .
, . , : java-script x86, , java-script, Linux . ! , ! , , , , , . , , , , , win32 … -, . , , Apple , , !
, win32 . , , ( ) Cobol- . . , win32 , , Microsoft … . , win32 !
, legacy , , , , , - , . - digital transformation.
, , , , . , - , ?
, , , . ( ), . – , . , , – .
. DoDo Pizza. , , , - .
, ( . , « ») - . , «, , - , , , , . , -, : „ : -, , IT-, - , , IT?“ – . - , .
, IT. ? , , , !
, , , ( . — «. ») . , . IT-, – IT-, ? , . , , , . – , . , Uber, Airbnb – .
, , – Microsoft – , , Yota Devices.
YotaPhone?
, YotaPhone, - Microsoft, . «ICEBERG», ML . , . , , . , : , , . ( ), , - - , : , , . , , , .
.
, , . Cortana (Cortana, , , — , Windows, Siri, OK Google).
?
, . Cortana 15 16 .
, -, « » ? ?
, , … , , . , , , , . , . , , , .
( , , !). . , ., , . , - Retail, , , Retail.
, . , , 10 . , , ! -, , -, . . ? ? , - -?
, , . ., – Amazon, , Amazon ., , . , 30% , 70% , . . Amazon : 70% 30% . , . , commodity service (), , .
, , , , . ! , , , , , . , , Kickstarter. crowdfunding-, Amazon , , ( ) . .
, , retail- , retail , , retail- , , , , - .
, , , , , , , , , , ! Thank you very much! Microsoft – , !
, ! !
We remind you that you can try Azure for free .
Minute advertising . If you want to try new technologies in your projects, but do not reach the hands, leave the application in the program Tech Acceleration from Microsoft. Its main feature is that together with you we will select the required stack, we will help to realize the pilot and, if successful, we will spend maximum efforts so that the whole market will know about you.
Source: https://habr.com/ru/post/342596/
All Articles