OpenStack PaaS Strategy
Posted by: Alex Friedland
From the translator: this article discusses two opposing views on the question of whether PaaS industry’s development of OpenStack threatens or not.
The new release of OpenStack Havana sends powerful signals to the industry that platform providers as services (PaaS vendors) may suddenly be "irrelevant." No matter how perfect or significant your current business may be, to prove its value within the dynamically developing OpenStack ecosystem is no easy task. At the moment, such authoritative PaaS vendors as CloudFoundry and OpenShift feel uncomfortable. A fun game - guess who among the recognized players will be destroyed next.
')
One of the first things that made us revisit the Havana release was the definition of OpenStack as an infrastructure-only service (IaaS) provider. Yes, the infrastructure is still in the spotlight, but at the same time there is a serious strengthening of the role of the Neutron project (formerly Quantum), as well as the addition of new core functionality to the Nova and Cinder projects, which indicates that the infrastructure update is stronger than ever . However, OpenStack is on the path of innovation, surprisingly similar to AWS. Remember that AWS first launched low-level infrastructure services, such as EC2 and S3, and then quickly upgraded the package with PaaS services such as SimpleDB, EMR and SQS. Well, OpenStack follows the path of providing a platform as a service through projects such as Heat, Trove, Savanna, Marconi, and others.
Below is a development path for Heat and Trove, two leading platform-level services within OpenStack. According to statistics , 63 engineers from 20 different companies contributed over 90 thousand lines of code to the Heat component of the Havana release. This is quite a big step from the starting point of the project, when 97% of the code for the Grizzly release was written by RedHat .
A similar picture can be observed in the project Trove , which was launched by just two companies as part of the Grizzly release (Rackspace and HP). In the Havana release, the community is already much bigger with participants like Mirantis and Red Hat, eBay SUSE.
Obviously, OpenStack is improving the stack to the level of services as a platform and the community is confidently moving towards this goal.
The importance of such progress is that it is an early sign that OpenStack will control not only the infrastructure at a low level, but the entire stack on top of it. The market seems to require it (as evidenced by the success of AWS), and OpenStack has the community driving force and resources to launch a wave that raises demand to a higher stack level.
Have you heard about the successful implementation of the Cloud Foundry or OpenShift platform over OpenStack? And I have not heard, and my prediction is that such an introduction does not threaten us. Too often, we see our customers experimenting for some time with OpenShift or Cloud Foundry, then choose instead the combination Heat as the orchestration system of cloud applications, Trove as a database, the LBaaS balancer to give elasticity, then they stick it all together. using skipt and Python code and get their own functioning solution for proving their applications. This is a clear indication that, at a minimum, for these cases, the value provided by PaaS solutions recognized in the market is insufficient to justify their use within the OpenStack ecosystem.
This does not mean that any third-party solution encountered on the path of the commoditization wave (transition of goods / services from the category of unique to the rank of rank and file) of OpenStack will end. On the contrary, solutions that have a sufficient level of additional capabilities will thrive, and the OpenStack community will do a lot to adapt them. One such example in the field of application infrastructure is a publicly available technology called Hadoop. The Hadoop community is different from OpenStack, and this technology has an undeniable and fairly differentiated value. Not surprisingly, instead of trying to copy and commodize Hadoop, the OpenStack community is working on the possibility of adapting it as part of the Savanna project.
As for third-party PaaS technologies, the differentiation of their values is not as obvious as in the case of Hadoop, and in the long term may not be sufficient to survive after commoditization. In fact, I personally think that PaaS vendors lost the game to OpenStack, without even starting it.
In the Icehouse release, OpenStack will continue to expand the available functionality, as well as open new horizons in areas that were not included in its original concept. So what's next? Cloud management and data centers?
When the OpenStack project was launched, Nimbula (RIP), CloudStack and Eucalyptus were faced with the need to answer the frightening question: “What is your difference from OpenStack?”. Apparently, today the list of vendors who will have to answer this question is growing rapidly.
BMC and CA, you have been warned !!!
How will the further development of IaaS technologies and the advancement of the stack to the level of PaaS affect the existing PaaS environment? Barb Darrow from GigaOM has especially well revealed this question in her article OpenStack - PaaS killer? A week in the cloud ( OpenStack: it is a PaaS killer? The week in cloud ). And Jesse Proudman of BlueBox posted one of the best articles on the topic of discussion. He suggested that OpenStack and Cloud Foundry are destined to live happily together.
Some readers may perceive the beginning of this post as an attack on the significance of PaaS and, in particular, on Cloud Foundry. Perhaps I used the wrong words to express my thoughts and want to apologize to everyone they insulted.
Officially declare that neither I nor Mirantis can say anything negative about PaaS in general and about Cloud Foundry in particular. On the contrary, I argue that it is success among application developers that makes the PaaS model so interesting for the IaaS community.
Let's lift the veil of what is happening inside OpenStack in the area of application support and orchestration.
Let's start with Heat , one of the fastest growing projects in OpenStack, which consumers of our products find especially interesting. Heat is the main project of the OpenStack initiative for cloud application orchestration. His goal is to “create an affordable service for humans and machines to manage the entire lifecycle of infrastructure and applications inside OpenStack clouds.” A few lines below we read that the Heat project implements a “control mechanism for running many complex cloud-based applications based on templates”, and further: “Also in the implementation of the Heat project, activities aimed at supporting compatibility with the AWS CloudFormation template format are conducted .” Here you can read about the strategy of Heat / Software, which is designed to raise the orchestration through Heat to an even higher level.
It seems to me a serious step towards PaaS, including thanks to the component that provides interoperability.
2009-2013 sonofamortician, source: sonofamortician.deviantart.com/art/history-repeats-itself-114964637 . All rights reserved.
Companies controlling a portion of the stack usually struggle to expand their control. Such a struggle, as a rule, begins with the fact that a market player declares his control over the fundamental elements of the stack, and then a transition to a higher level takes place. The story of Microsoft’s rise in the 1980s and 1990s is a great example. Starting with control over the operating system based on the IBM PC platform, Microsoft quickly managed to win over application developers, capturing the development tools segment, integrating them better and faster into its OS. On their way, they bypassed very worthy and popular players in the market, for example, Borland, which at that time owned more powerful and, undoubtedly, more popular development tools (everyone who is older should remember Turbo C / C ++ and almost religious commitment to him in the community of developers, including the author of this blog), but could not compete in the field of OS integration. And since the developed applications had to work efficiently on MS-DOS and MS Windows, such integration was vital for developers and application owners.
This story does not seem familiar to you?
Although we are still far from Microsoft, OpenStack is no exception to this rule. In the end, as Jesse Proudman correctly noted, many companies behind OpenStack are technology providers (IBM, HP, Red Hat, and others) and are very aware of the importance of controlling the entire stack. And, if you agree with the user @ulander Peder Ulander (Peder Ulander) wrote to me in his tweet, “the lower your position in this model is, the less important you are”, then OpenStack doesn’t have any other choice in the further development, except to make serious attempts to get closer to developers and application owners.
Despite all these obvious things, some PaaS vendors pointed out that the PaaS functionality in the Havana release of the IaaS community is minimal and cannot be compared with the innovations taking place in the PaaS community.
Although these experts may be right about the current state of affairs, it seems to me that they do not understand a more important point. I agree that the PaaS-functional of OpenStack is not close to the fact that they are offered by products of experienced players in the PaaS market. Anyway, targeting customers who love and buy the Cloud Foundry platform is an old song.
Our target segment should be potential customers who like it, but who do not buy it.
I talked about some of our customers who, after evaluating Cloud Foundry, decided to opt for the minimum functionality built into OpenStack to meet their application development needs. Perhaps this is pure luck, but perhaps this is the beginning of a real competitive trend, which should be recognized and to which attention should be paid at an early stage?
I want to say that this is a threat. And the proof of this is not the functionality that can be found in OpenStack Havana today, but the direction that is being developed and the players who lead it there.
We recently witnessed the announcement of a new OpenStack project called Solum , which will be implemented with the support of RedHat, Rackspace, eBay, dotCloud and Cumulogic. Sponsors describe it as “a project related to OpenStack, the purpose of which is to simplify the use of cloud services and their integration into your application development process”. Solum was originally designed for the OpenStack clouds and improves its various projects, including Heat, Keystone, Nova, Trove and many others.
Now we can argue about how much of this mission has already been accomplished, but it is obvious that this mission is very similar to the mission of any commercial PaaS, including its aspect related to ensuring interoperability, as well as the fact that it is part of OpenStack and closely integrated with other infrastructure projects of OpenStack.
The announcement of the release of Solum caused a significant stir in Twitter. Even @cloudpundit (also known as Lydia Leong from Gartner) suggested that Solum will be a direct competitor to Cloud Foundry based on OpenStack.
Now I would like to turn to the point of view of Jesse Proudman that vendors who are personally interested in PaaS solutions like CloudFoudry and OpenShift will not reduce sales of their products at the expense of OpenStack.
Facts suggest otherwise. Despite the huge investment in OpenShift, Red Hat is at the forefront of accelerating the delivery of OpenStack to a full-fledged PaaS, as evidenced by the projects Heat and Solum.
Red Hat is a veteran of open source software development and is well-versed in market trends, and I invite others to learn from its example. I want to draw your attention to what Red Hat does with OpenShift, a position that one of the developers of OpenShift eloquently explained in his blog (http://mattoncloud.org/).
"Despite the fact that we can perfectly use OpenShift based on OpenStack , and even perform more efficient integration through projects such as Heat and Neutron, we had the feeling that today our platform has a more substantial set of features that could be an integral part of OpenStack itself. Thus, we could dramatically improve the work with OpenStack. ”
To conclude, I can say the following: instead of being skeptical about pushing OpenStack to a higher level in the stack, open source PaaS providers should work with the OpenStack community to see how they can be useful. If you cannot stop the revolution, join it!
Original article in English .
From the translator: this article discusses two opposing views on the question of whether PaaS industry’s development of OpenStack threatens or not.
Strict warning from OpenStack Havana: provide unique value or die
The new release of OpenStack Havana sends powerful signals to the industry that platform providers as services (PaaS vendors) may suddenly be "irrelevant." No matter how perfect or significant your current business may be, to prove its value within the dynamically developing OpenStack ecosystem is no easy task. At the moment, such authoritative PaaS vendors as CloudFoundry and OpenShift feel uncomfortable. A fun game - guess who among the recognized players will be destroyed next.
')
Is OpenStack really just infrastructure as a service (IaaS)?
One of the first things that made us revisit the Havana release was the definition of OpenStack as an infrastructure-only service (IaaS) provider. Yes, the infrastructure is still in the spotlight, but at the same time there is a serious strengthening of the role of the Neutron project (formerly Quantum), as well as the addition of new core functionality to the Nova and Cinder projects, which indicates that the infrastructure update is stronger than ever . However, OpenStack is on the path of innovation, surprisingly similar to AWS. Remember that AWS first launched low-level infrastructure services, such as EC2 and S3, and then quickly upgraded the package with PaaS services such as SimpleDB, EMR and SQS. Well, OpenStack follows the path of providing a platform as a service through projects such as Heat, Trove, Savanna, Marconi, and others.
Evolution to the level of PaaS - statistics
Below is a development path for Heat and Trove, two leading platform-level services within OpenStack. According to statistics , 63 engineers from 20 different companies contributed over 90 thousand lines of code to the Heat component of the Havana release. This is quite a big step from the starting point of the project, when 97% of the code for the Grizzly release was written by RedHat .
A similar picture can be observed in the project Trove , which was launched by just two companies as part of the Grizzly release (Rackspace and HP). In the Havana release, the community is already much bigger with participants like Mirantis and Red Hat, eBay SUSE.
Obviously, OpenStack is improving the stack to the level of services as a platform and the community is confidently moving towards this goal.
Why should you care?
The importance of such progress is that it is an early sign that OpenStack will control not only the infrastructure at a low level, but the entire stack on top of it. The market seems to require it (as evidenced by the success of AWS), and OpenStack has the community driving force and resources to launch a wave that raises demand to a higher stack level.
Have you heard about the successful implementation of the Cloud Foundry or OpenShift platform over OpenStack? And I have not heard, and my prediction is that such an introduction does not threaten us. Too often, we see our customers experimenting for some time with OpenShift or Cloud Foundry, then choose instead the combination Heat as the orchestration system of cloud applications, Trove as a database, the LBaaS balancer to give elasticity, then they stick it all together. using skipt and Python code and get their own functioning solution for proving their applications. This is a clear indication that, at a minimum, for these cases, the value provided by PaaS solutions recognized in the market is insufficient to justify their use within the OpenStack ecosystem.
This does not mean that any third-party solution encountered on the path of the commoditization wave (transition of goods / services from the category of unique to the rank of rank and file) of OpenStack will end. On the contrary, solutions that have a sufficient level of additional capabilities will thrive, and the OpenStack community will do a lot to adapt them. One such example in the field of application infrastructure is a publicly available technology called Hadoop. The Hadoop community is different from OpenStack, and this technology has an undeniable and fairly differentiated value. Not surprisingly, instead of trying to copy and commodize Hadoop, the OpenStack community is working on the possibility of adapting it as part of the Savanna project.
As for third-party PaaS technologies, the differentiation of their values is not as obvious as in the case of Hadoop, and in the long term may not be sufficient to survive after commoditization. In fact, I personally think that PaaS vendors lost the game to OpenStack, without even starting it.
In the Icehouse release, OpenStack will continue to expand the available functionality, as well as open new horizons in areas that were not included in its original concept. So what's next? Cloud management and data centers?
When the OpenStack project was launched, Nimbula (RIP), CloudStack and Eucalyptus were faced with the need to answer the frightening question: “What is your difference from OpenStack?”. Apparently, today the list of vendors who will have to answer this question is growing rapidly.
BMC and CA, you have been warned !!!
OpenStack and its PaaS strategy: a closer look behind the scenes
How will the further development of IaaS technologies and the advancement of the stack to the level of PaaS affect the existing PaaS environment? Barb Darrow from GigaOM has especially well revealed this question in her article OpenStack - PaaS killer? A week in the cloud ( OpenStack: it is a PaaS killer? The week in cloud ). And Jesse Proudman of BlueBox posted one of the best articles on the topic of discussion. He suggested that OpenStack and Cloud Foundry are destined to live happily together.
The PaaS model is important. Cloud Foundry platform is reliable. I apologize!
Some readers may perceive the beginning of this post as an attack on the significance of PaaS and, in particular, on Cloud Foundry. Perhaps I used the wrong words to express my thoughts and want to apologize to everyone they insulted.
Officially declare that neither I nor Mirantis can say anything negative about PaaS in general and about Cloud Foundry in particular. On the contrary, I argue that it is success among application developers that makes the PaaS model so interesting for the IaaS community.
... but OpenStack MOVES along the path to PaaS. Sorry.
Let's lift the veil of what is happening inside OpenStack in the area of application support and orchestration.
Let's start with Heat , one of the fastest growing projects in OpenStack, which consumers of our products find especially interesting. Heat is the main project of the OpenStack initiative for cloud application orchestration. His goal is to “create an affordable service for humans and machines to manage the entire lifecycle of infrastructure and applications inside OpenStack clouds.” A few lines below we read that the Heat project implements a “control mechanism for running many complex cloud-based applications based on templates”, and further: “Also in the implementation of the Heat project, activities aimed at supporting compatibility with the AWS CloudFormation template format are conducted .” Here you can read about the strategy of Heat / Software, which is designed to raise the orchestration through Heat to an even higher level.
It seems to me a serious step towards PaaS, including thanks to the component that provides interoperability.
History shows that PaaS vendors should be alert
2009-2013 sonofamortician, source: sonofamortician.deviantart.com/art/history-repeats-itself-114964637 . All rights reserved.
Companies controlling a portion of the stack usually struggle to expand their control. Such a struggle, as a rule, begins with the fact that a market player declares his control over the fundamental elements of the stack, and then a transition to a higher level takes place. The story of Microsoft’s rise in the 1980s and 1990s is a great example. Starting with control over the operating system based on the IBM PC platform, Microsoft quickly managed to win over application developers, capturing the development tools segment, integrating them better and faster into its OS. On their way, they bypassed very worthy and popular players in the market, for example, Borland, which at that time owned more powerful and, undoubtedly, more popular development tools (everyone who is older should remember Turbo C / C ++ and almost religious commitment to him in the community of developers, including the author of this blog), but could not compete in the field of OS integration. And since the developed applications had to work efficiently on MS-DOS and MS Windows, such integration was vital for developers and application owners.
This story does not seem familiar to you?
Although we are still far from Microsoft, OpenStack is no exception to this rule. In the end, as Jesse Proudman correctly noted, many companies behind OpenStack are technology providers (IBM, HP, Red Hat, and others) and are very aware of the importance of controlling the entire stack. And, if you agree with the user @ulander Peder Ulander (Peder Ulander) wrote to me in his tweet, “the lower your position in this model is, the less important you are”, then OpenStack doesn’t have any other choice in the further development, except to make serious attempts to get closer to developers and application owners.
PaaS vendors don't take the threat seriously
Despite all these obvious things, some PaaS vendors pointed out that the PaaS functionality in the Havana release of the IaaS community is minimal and cannot be compared with the innovations taking place in the PaaS community.
Although these experts may be right about the current state of affairs, it seems to me that they do not understand a more important point. I agree that the PaaS-functional of OpenStack is not close to the fact that they are offered by products of experienced players in the PaaS market. Anyway, targeting customers who love and buy the Cloud Foundry platform is an old song.
Our target segment should be potential customers who like it, but who do not buy it.
I talked about some of our customers who, after evaluating Cloud Foundry, decided to opt for the minimum functionality built into OpenStack to meet their application development needs. Perhaps this is pure luck, but perhaps this is the beginning of a real competitive trend, which should be recognized and to which attention should be paid at an early stage?
I want to say that this is a threat. And the proof of this is not the functionality that can be found in OpenStack Havana today, but the direction that is being developed and the players who lead it there.
We recently witnessed the announcement of a new OpenStack project called Solum , which will be implemented with the support of RedHat, Rackspace, eBay, dotCloud and Cumulogic. Sponsors describe it as “a project related to OpenStack, the purpose of which is to simplify the use of cloud services and their integration into your application development process”. Solum was originally designed for the OpenStack clouds and improves its various projects, including Heat, Keystone, Nova, Trove and many others.
Now we can argue about how much of this mission has already been accomplished, but it is obvious that this mission is very similar to the mission of any commercial PaaS, including its aspect related to ensuring interoperability, as well as the fact that it is part of OpenStack and closely integrated with other infrastructure projects of OpenStack.
The announcement of the release of Solum caused a significant stir in Twitter. Even @cloudpundit (also known as Lydia Leong from Gartner) suggested that Solum will be a direct competitor to Cloud Foundry based on OpenStack.
Do not be skeptical about OpenStack: Join the project! Become the owner of the platform!
Now I would like to turn to the point of view of Jesse Proudman that vendors who are personally interested in PaaS solutions like CloudFoudry and OpenShift will not reduce sales of their products at the expense of OpenStack.
Facts suggest otherwise. Despite the huge investment in OpenShift, Red Hat is at the forefront of accelerating the delivery of OpenStack to a full-fledged PaaS, as evidenced by the projects Heat and Solum.
Red Hat is a veteran of open source software development and is well-versed in market trends, and I invite others to learn from its example. I want to draw your attention to what Red Hat does with OpenShift, a position that one of the developers of OpenShift eloquently explained in his blog (http://mattoncloud.org/).
"Despite the fact that we can perfectly use OpenShift based on OpenStack , and even perform more efficient integration through projects such as Heat and Neutron, we had the feeling that today our platform has a more substantial set of features that could be an integral part of OpenStack itself. Thus, we could dramatically improve the work with OpenStack. ”
To conclude, I can say the following: instead of being skeptical about pushing OpenStack to a higher level in the stack, open source PaaS providers should work with the OpenStack community to see how they can be useful. If you cannot stop the revolution, join it!
Original article in English .
Source: https://habr.com/ru/post/217049/
All Articles