David Chappell about the clouds and the future development
At DevDay * Kyiv 2014, I was able to talk with the author of the key report, David Chappell. David is a renowned speaker, consultant and author of books, many of which are used to study at MIT and ETH. In his work, he seeks to help people in IT understand, use and make decisions about new technologies.
David is a specialist in cloud technologies in the corporate sector, and therefore the answers are interesting, like looking from the corporate world to modern technologies and their development:
Today, when we hear cloud computing, we usually think of large public clouds, for example, Amazon Web Services, Microsoft Azure, or Google Cloud Computing. On the other hand, there is a private cloud market: OpenStack and Microsoft System Center. Will we store all data in public clouds, or will everyone have their own private cloud?
')
Let's start this extensive topic with a definition of what a private cloud is. Many IT executives began to call their data centers private clouds because a private cloud sounds cool. But, just starting to call your data center cloud is not quite honest ( cheating ). I would suggest calling IaaS a private cloud in your data center ( Infrastructure-as-a-Service, infrastructure as a service ) or in other words on-demand virtual machines in your data center.
IaaS makes it easy to deploy identical virtual machines. You have a set of images from which you can choose, and IaaS is responsible for quickly creating virtual machines. In organizations that create many identical virtual machines, IaaS is really valuable. But, what is this organization? These are large companies. Small and medium-sized organizations do not deploy virtual machines so often, and even when they do, they create different virtual machines. And therefore, I doubt the value of private IaaS in most companies. Large companies can really benefit from a private cloud, but I'm not sure about the value of a private cloud, as I define it, in medium and small companies.
Secondly, even for large companies, where IaaS in its own data center makes sense, the private cloud cannot completely replace the public one. Again, a private cloud simplifies the allocation of virtual machines in your data center. Great, but not related to public clouds. You use public clouds because you:
Private and public clouds solve various problems. And, if you have a private cloud, it does not mean that you can do without public clouds. Private clouds make your data center more efficient, public clouds give you access to scale, low cost and pay per use. These clouds are not interchangeable.
As I understand it, private clouds will remain only in large companies. Will these companies be able to provide public cloud services - turn their private into public ones?
Practically for any company it is now impossible to enter the public cloud computing business. The reason for this is in the size of the main players - Amazon, Google, Microsoft, and the scale with which they work. You may have heard that IBM announced this year that they are investing a billion dollars in their own cloud-based data centers. But when you think about it, it turns out that they will distribute this amount to approximately 25 data centers around the world, that is, a billion dollars to 25 data centers. For comparison, Microsoft spends a billion dollars just to build one data center. In a cloud environment, they chuckle at IBM, their announcement sounded good, but, in the real scale, Amazon, Google and Microsoft clouds will not be visible.
The investment in this market is huge. What company can now decide to compete in IaaS immediately with Amazon, Google, Microsoft. Rackspace, one of the founders of OpenStack, has long been the second largest IaaS supplier in the world. And so, Rackspace announced that they were leaving the public cloud business because they could not compete with large companies. Therefore, I would be surprised if someone would try to enter the business of public clouds, perhaps, apart from facebook.
If we talk about the distribution of files, then the peering networks in general are superior to other technologies here, even cloud ones. Approximately 60% of all file distribution is provided by peer-to-peer technologies. You can comment on BitTorrent Sync, which is positioned as a peer-to-peer competitor to Dropbox, where data is not stored in the cloud, but on clients.
Unfortunately, I do not follow this market.
Is cloud technology worth something to respond to? If peer-to-peer networks turn out in personal data storage, is it possible that this technology will be able to compete in other areas, such as video distribution?
Money comes to public cloud platforms not from consumers ( consumer ), but from corporate users. And corporations will not use peer-to-peer for data storage. They will not be happy that their data is stored on different clients. I am very focused on clouds and corporations, so I have no firm conviction, but I am skeptical that peering technologies can be a threat to cloud storage.
But were the recent sensations in the IT market targeted specifically at consumers?
I do not agree. In terms of publicity - yes, but in terms of profit - this is not true. The corporate IT market is a huge business. Even venture capitalists invest much more in projects focused on corporations, but media attention is always focused on the consumer sector. If you read the press, there is always a big sensation in the consumer sector because most people are consumers and read about what interests them.
Is there a fundamental difference between the PC + era, where many devices are connected to the Internet, and the Internet of things ( IoT, Internet of Things ), where all the devices that surround us will be connected to the Internet?
The era of PC + is computers, laptops and mobile devices (phones and tablets). IoT is broader, but when we say IoT, we do not mean laptops and computers in it. There will be solutions specifically for PC + and specialized for IoT. It will not be two separate markets, they will have a lot in common, but on the other hand, there will be things that are needed only in a PC or only in IoT.
Will the developers continue to specialize? For example, will there be a specialty kitchen developer, as there is a mobile developer now, or will it be broader specialists?
I perceive IoT as an extension of modern embedded solutions. Now there are developers who specialize in developing applications that can work in limited resources of embedded systems. And for me, IoT is the transfer of all these devices (refrigerators, microwaves and cars) to the Internet. I do not think that this is such a big leap, it's just more software in more devices. My microwave already has software inside, just now it will have access to the Internet. A few months ago, I bought a new BMW, and the new car was completely IoT. My previous BMW also contained a lot of software, but it worked autonomously, but now the software is accessing the Internet.
You probably know about Tesla machines. I constantly see them, never hear them, because they are quiet, but I always see. Tesla initially behaved like a machine with a manual gearbox, if you do not press the brake on a slope, the car will start to roll back. With an automatic transmission, the car will stand still. And the car, which rolls back, surprised the American drivers, who are accustomed to the machine. Tesla fixed this behavior by updating the software via the Internet - this is the real IoT.
In the PC era, you had complete control of the device, and you shared control with the manufacturer (for example, you enabled Windows Update). In the era of PC + and IoT devices have become more closed, that is, the manufacturer has full control, and, if I may say so, you should ask the manufacturer for permission to act. Do you think this closeness of manufacturers helps or does it hinder the development of PC + and IoT?
I'm not sure that we can talk about complete closeness because the level of different manufacturers is different. Apple is Apple, Android is very open, Windows Phone is somewhere in the middle, for example, corporate users can download their applications bypassing the market if they want. But the fact remains that manufacturers control users more. It helps to be a more flawless system, for example, I would not want any developer to have full access to my new car, its brakes or clutch, it would make me worry. My personal opinion is that I don’t mind manufacturers having more control. The more system error can hurt me, the more limited access I would like to get third-party developers. The more devices penetrate into our lives, the more manufacturers close access to them.
Manufacturers have their own interests, this is a business. Does this create problems?
A little scary that manufacturers will have so much control that they can turn off my microwave. And it is true that the level of trust that manufacturers demand from us has greatly increased lately. It is possible that new technologies require new legislation.
Mobile phones are updated on the same day or week when updates are published by the developer. In SaaS (s oftware-as-a-service, software as a service ), updates are available instantly. Many developers have switched to small and frequent updates, and rolling releases ( Rolling release ). Do you think major developers will switch to such a development model?
Microsoft is now offering a subscription office suite, I pay a one-year subscription and receive periodic updates, this is very close to the SaaS model, although updates come to my computer. I would not be surprised if Windows eventually switches to this model.
On the technical side, it is possible to discuss whether major versions are necessary now, but, in marketing terms, they give the user a sense of permanence. My friend works at a medical school and complains about every change, even a change in a single pixel, in the Windows interface. She says that she was satisfied with everything as it was, she doesn’t need new functions, that don’t change anything, I’m a doctor, let me just use it. And I understand her.
But if you are Microsoft, you need to develop your products, move on. And if Microsoft will constantly change here and there, every week, every two weeks, every month, the user will be shocked every time. But, if they have major versions, where many changes occur at once, then the transition is more manageable for the user, they have time to think and prepare for changes. So versions, as a large set of changes, may no longer depend on technologies, because we can release changes to change at least every week, but they are more attractive from customers, who are allowed to keep changes under control.
Do users and marketing keep companies from floating releases?
This is a difficult problem, because technology already allows for floating releases, and it makes sense, because new things are constantly needed. But on the other hand, users have a limited ability to accept changes, for them, more often, what is already there is better, and new technologies that appear too quickly can make their lives worse. Just because they have to constantly adjust.
David is a specialist in cloud technologies in the corporate sector, and therefore the answers are interesting, like looking from the corporate world to modern technologies and their development:
- What company can now enter the market of public clouds?
- Will computers be part of the Internet of Things?
- How are floating releases related to marketing?
Cloud computing
Today, when we hear cloud computing, we usually think of large public clouds, for example, Amazon Web Services, Microsoft Azure, or Google Cloud Computing. On the other hand, there is a private cloud market: OpenStack and Microsoft System Center. Will we store all data in public clouds, or will everyone have their own private cloud?
')
Let's start this extensive topic with a definition of what a private cloud is. Many IT executives began to call their data centers private clouds because a private cloud sounds cool. But, just starting to call your data center cloud is not quite honest ( cheating ). I would suggest calling IaaS a private cloud in your data center ( Infrastructure-as-a-Service, infrastructure as a service ) or in other words on-demand virtual machines in your data center.
IaaS makes it easy to deploy identical virtual machines. You have a set of images from which you can choose, and IaaS is responsible for quickly creating virtual machines. In organizations that create many identical virtual machines, IaaS is really valuable. But, what is this organization? These are large companies. Small and medium-sized organizations do not deploy virtual machines so often, and even when they do, they create different virtual machines. And therefore, I doubt the value of private IaaS in most companies. Large companies can really benefit from a private cloud, but I'm not sure about the value of a private cloud, as I define it, in medium and small companies.
Secondly, even for large companies, where IaaS in its own data center makes sense, the private cloud cannot completely replace the public one. Again, a private cloud simplifies the allocation of virtual machines in your data center. Great, but not related to public clouds. You use public clouds because you:
- you want to pay only what you use (hourly payment of a virtual machine per hour);
- you need huge scalability;
- You want to bring IT infrastructure beyond your company.
Private and public clouds solve various problems. And, if you have a private cloud, it does not mean that you can do without public clouds. Private clouds make your data center more efficient, public clouds give you access to scale, low cost and pay per use. These clouds are not interchangeable.
As I understand it, private clouds will remain only in large companies. Will these companies be able to provide public cloud services - turn their private into public ones?
Practically for any company it is now impossible to enter the public cloud computing business. The reason for this is in the size of the main players - Amazon, Google, Microsoft, and the scale with which they work. You may have heard that IBM announced this year that they are investing a billion dollars in their own cloud-based data centers. But when you think about it, it turns out that they will distribute this amount to approximately 25 data centers around the world, that is, a billion dollars to 25 data centers. For comparison, Microsoft spends a billion dollars just to build one data center. In a cloud environment, they chuckle at IBM, their announcement sounded good, but, in the real scale, Amazon, Google and Microsoft clouds will not be visible.
The investment in this market is huge. What company can now decide to compete in IaaS immediately with Amazon, Google, Microsoft. Rackspace, one of the founders of OpenStack, has long been the second largest IaaS supplier in the world. And so, Rackspace announced that they were leaving the public cloud business because they could not compete with large companies. Therefore, I would be surprised if someone would try to enter the business of public clouds, perhaps, apart from facebook.
If we talk about the distribution of files, then the peering networks in general are superior to other technologies here, even cloud ones. Approximately 60% of all file distribution is provided by peer-to-peer technologies. You can comment on BitTorrent Sync, which is positioned as a peer-to-peer competitor to Dropbox, where data is not stored in the cloud, but on clients.
Unfortunately, I do not follow this market.
Is cloud technology worth something to respond to? If peer-to-peer networks turn out in personal data storage, is it possible that this technology will be able to compete in other areas, such as video distribution?
Money comes to public cloud platforms not from consumers ( consumer ), but from corporate users. And corporations will not use peer-to-peer for data storage. They will not be happy that their data is stored on different clients. I am very focused on clouds and corporations, so I have no firm conviction, but I am skeptical that peering technologies can be a threat to cloud storage.
But were the recent sensations in the IT market targeted specifically at consumers?
I do not agree. In terms of publicity - yes, but in terms of profit - this is not true. The corporate IT market is a huge business. Even venture capitalists invest much more in projects focused on corporations, but media attention is always focused on the consumer sector. If you read the press, there is always a big sensation in the consumer sector because most people are consumers and read about what interests them.
Internet of things
Is there a fundamental difference between the PC + era, where many devices are connected to the Internet, and the Internet of things ( IoT, Internet of Things ), where all the devices that surround us will be connected to the Internet?
The era of PC + is computers, laptops and mobile devices (phones and tablets). IoT is broader, but when we say IoT, we do not mean laptops and computers in it. There will be solutions specifically for PC + and specialized for IoT. It will not be two separate markets, they will have a lot in common, but on the other hand, there will be things that are needed only in a PC or only in IoT.
Will the developers continue to specialize? For example, will there be a specialty kitchen developer, as there is a mobile developer now, or will it be broader specialists?
I perceive IoT as an extension of modern embedded solutions. Now there are developers who specialize in developing applications that can work in limited resources of embedded systems. And for me, IoT is the transfer of all these devices (refrigerators, microwaves and cars) to the Internet. I do not think that this is such a big leap, it's just more software in more devices. My microwave already has software inside, just now it will have access to the Internet. A few months ago, I bought a new BMW, and the new car was completely IoT. My previous BMW also contained a lot of software, but it worked autonomously, but now the software is accessing the Internet.
You probably know about Tesla machines. I constantly see them, never hear them, because they are quiet, but I always see. Tesla initially behaved like a machine with a manual gearbox, if you do not press the brake on a slope, the car will start to roll back. With an automatic transmission, the car will stand still. And the car, which rolls back, surprised the American drivers, who are accustomed to the machine. Tesla fixed this behavior by updating the software via the Internet - this is the real IoT.
In the PC era, you had complete control of the device, and you shared control with the manufacturer (for example, you enabled Windows Update). In the era of PC + and IoT devices have become more closed, that is, the manufacturer has full control, and, if I may say so, you should ask the manufacturer for permission to act. Do you think this closeness of manufacturers helps or does it hinder the development of PC + and IoT?
I'm not sure that we can talk about complete closeness because the level of different manufacturers is different. Apple is Apple, Android is very open, Windows Phone is somewhere in the middle, for example, corporate users can download their applications bypassing the market if they want. But the fact remains that manufacturers control users more. It helps to be a more flawless system, for example, I would not want any developer to have full access to my new car, its brakes or clutch, it would make me worry. My personal opinion is that I don’t mind manufacturers having more control. The more system error can hurt me, the more limited access I would like to get third-party developers. The more devices penetrate into our lives, the more manufacturers close access to them.
Manufacturers have their own interests, this is a business. Does this create problems?
A little scary that manufacturers will have so much control that they can turn off my microwave. And it is true that the level of trust that manufacturers demand from us has greatly increased lately. It is possible that new technologies require new legislation.
Future development
Mobile phones are updated on the same day or week when updates are published by the developer. In SaaS (s oftware-as-a-service, software as a service ), updates are available instantly. Many developers have switched to small and frequent updates, and rolling releases ( Rolling release ). Do you think major developers will switch to such a development model?
Microsoft is now offering a subscription office suite, I pay a one-year subscription and receive periodic updates, this is very close to the SaaS model, although updates come to my computer. I would not be surprised if Windows eventually switches to this model.
On the technical side, it is possible to discuss whether major versions are necessary now, but, in marketing terms, they give the user a sense of permanence. My friend works at a medical school and complains about every change, even a change in a single pixel, in the Windows interface. She says that she was satisfied with everything as it was, she doesn’t need new functions, that don’t change anything, I’m a doctor, let me just use it. And I understand her.
But if you are Microsoft, you need to develop your products, move on. And if Microsoft will constantly change here and there, every week, every two weeks, every month, the user will be shocked every time. But, if they have major versions, where many changes occur at once, then the transition is more manageable for the user, they have time to think and prepare for changes. So versions, as a large set of changes, may no longer depend on technologies, because we can release changes to change at least every week, but they are more attractive from customers, who are allowed to keep changes under control.
Do users and marketing keep companies from floating releases?
This is a difficult problem, because technology already allows for floating releases, and it makes sense, because new things are constantly needed. But on the other hand, users have a limited ability to accept changes, for them, more often, what is already there is better, and new technologies that appear too quickly can make their lives worse. Just because they have to constantly adjust.
Source: https://habr.com/ru/post/242539/
All Articles