Interview with Devananda van der Veen, OpenStack Ironic Technical Project Manager
We present the ninth of a series of interviews with the technical leaders of the OpenStack project in the Mirantis blog. Our goal is to educate the wider technical community and help people understand how they can contribute to and benefit from the OpenStack project. Naturally, below is the point of view of the interviewee, not of Mirantis.
The following is an interview with Devananda van der Veen, technical director of the OpenStack Ironic project .
Mirantis: Tell us about yourself.
')
Devananda van der Win: I’m a Senior Systems Engineer at HP Cloud. I have been working here for more than 1.5 years, currently I am heading the Ironic project. Before that, I worked as a consultant and MySQL database administrator.
Q: What is your relationship history with OpenStack? Why do you participate in the project?
Answer: For many years I worked with Monty Taylor and when he spoke to me about OpenStack, my reaction was: “Oh, uh, great! It sounds cool! At that time, I already wanted to switch from MySQL to something new and it was clear to me that OpenStack would be the next stage in the development of technologies. MySQL has shown a really rapid growth since about 2005. Therefore, I then turned my career in that direction, and I see that now the same thing is happening with OpenStack. I believe that the process in whose creation we are participating here will have an impact on the entire IT field, and I want to be a part of it.
Q: What are your responsibilities as a technical project manager for Ironic?
Answer: Determining the direction and coordination of the activities of the developers are in many respects the same responsibilities as other technical projects of OpenStack. I am responsible for the development of the concept, manage the verification of the code, guide the developers, but I am away from the development itself. I inspire the community to develop around Ironic.
Question: What is the most essential thing you need to do to keep the project active?
Answer: I want to create a project that would help a large number of different hardware vendors feel comfortable and not feel that they are competing for proprietary bits or splitting the project into parts. Actually Ironic is a service for provisioning hardware or “bare iron” in OpenStack. All hardware vendors, such as HP, Dell and IBM, have in-house tools for managing hardware, such as the HP iLO server remote control or the Dell DRAC remote access controller. All of them add additional functionality to the standard IPMI interface specification, and some of them implement it a little differently.
Q: Can you explain the role of the Ironic project in OpenStack? Why is he important?
Answer: This can be explained in two different ways, I will try to choose one.
So far in the development of OpenStack, we have had to use other tools for deploying OpenStack, whether it's a couple of servers in your server cabinet or rack, or a whole data center. The hardware was necessarily external to OpenStack.
Ironic's role is to provide a hardware provisioning level that was previously not available in OpenStack. Based on this role, one of the goals of Ironic is to make TripleO (OpenStack-on-OpenStack) possible. All the tools you use to deploy a complete application in the cloud can be reused to deploy the cloud, which is ultimately another integrated application.
Question: What is the difference between Ironic and TripleO?
Answer: Ironic is a service that controls the power of a physical machine and writes an image on a machine. In addition, we have plans to make other management tasks possible. TripleO is located at a higher level in the stack and uses not only Ironic, but also many other services for deploying and managing the OpenStack cloud.
Q: Tell us about the Ironic community - who are contributing to the project?
Answer: HP, Red Hat, Mirantis, IBM and others. From the hardware vendors, both HP and IBM work with us. Until now, IBM has largely participated in a related project - creating an IPMI driver entirely in Python, which was moved from xCAT to the OpenStack community. It will be used by Ironic as a more scalable replacement of the ipmitool library inherited from the Baremetal code of the Nova component.
Q: Would you like to see greater involvement from hardware vendors?
Answer: Yes, I would like to. Sure, I'd like to see Dell. So far, they are not involved in our project. Ironic is a very young project, and perhaps some hardware vendors have not yet seriously participated in it, simply because the code was not ready. I think we are rapidly approaching the level of full code availability, when people can quickly join us and add their own hardware drivers.
Question: What has the Ironic community achieved at the moment?
Answer: Providing bare metal inherited a large number of restrictions, since it was originally an attempt to launch it as part of the Nova Compute process. For the last 4 months or so, we “pulled out” Ironic from Nova, turning it into a standalone OpenStack service. It has its own API service that can be scaled. It has its own message queue and server-side database, as well as the Conductor service - something in between the Nova Conductor and Nova Compute services. Most recently, we added support for DevStack and a disk image designer, as well as a python client; Tempest tests are under development. All this is necessary for any OpenStack project, but it seems to me that this is quite a great achievement for a small team over a period of just 4 months!
Question: What features does Ironic implement in the next release of OpenStack?
Answer: I can’t say exactly what Ironic functionality will be available in the Icehouse.
Question: If you could wave a magic wand right now and turn Ironic into exactly what you want it to be, what would it be at the moment?
Answer: I would like to see it scalable and fault tolerant. My definition of high availability in this context is its resilience to failure, and not the complete lack thereof. Hardware failure is inevitable, but Ironic must be able to recover from failures.
I would like to see in its structure different drivers from different hardware vendors. I would like the drivers themselves to be publicly available, and not proprietary. I would like Ironic to manage all kinds of hardware, starting with ARM devices and ending with a “big iron” - with all the existing variety of equipment that I personally may not be aware of. I would like other people to contribute in the form of drivers in order to make it all realizable and feasible.
Question: Are there any misconceptions regarding Ironic?
Answer: One of them is that Ironic is not a configuration database and that it will not store the history of states. I was asked: "You will monitor the fan speed, the temperature of the CPU, and then take any action if something overheats?". My answer is no!".
But the biggest misconception at the moment is that you can use Nova Compute or Ironic to realize unprotected multiplayer clouds based on bare metal. This is our goal, but for today we have not yet reached it. There are a number of significant hidden issues related to the security of unprotected cloud users operating on bare metal base. If I could wave a magic wand again, I would like all these problems to be solved. This will require a huge collaboration of hardware vendors to develop reliable hardware, as well as addressing security issues at the firmware level, and not just with respect to network load and user isolation. Some work on this is already underway, but I believe that we are still far from solving these problems. In the meantime, I want people to know that today you cannot run clouds under unprotected users on bare metal.
Question: Another question from the “I would like” series. Who would you like to see in the number of people involved in the Ironic project? Who is your perfect member at the moment?
Answer: Developers who know how to work with the open source community, because they are not always easy to find. People with good systems administration experience in addition to programming, since most of what we do at Ironic is low. Yes, it is written in Python, but we do a lot to integrate with processes such as DHCP, IPMI and PXE, and I will need people who understand the security issues of embedded software. The need for such team members, it seems to me, is currently not fully satisfied.
Question: Thank you for your time.
Answer: Thank you.
The following is an interview with Devananda van der Veen, technical director of the OpenStack Ironic project .
Mirantis: Tell us about yourself.
')
Devananda van der Win: I’m a Senior Systems Engineer at HP Cloud. I have been working here for more than 1.5 years, currently I am heading the Ironic project. Before that, I worked as a consultant and MySQL database administrator.
Q: What is your relationship history with OpenStack? Why do you participate in the project?
Answer: For many years I worked with Monty Taylor and when he spoke to me about OpenStack, my reaction was: “Oh, uh, great! It sounds cool! At that time, I already wanted to switch from MySQL to something new and it was clear to me that OpenStack would be the next stage in the development of technologies. MySQL has shown a really rapid growth since about 2005. Therefore, I then turned my career in that direction, and I see that now the same thing is happening with OpenStack. I believe that the process in whose creation we are participating here will have an impact on the entire IT field, and I want to be a part of it.
Q: What are your responsibilities as a technical project manager for Ironic?
Answer: Determining the direction and coordination of the activities of the developers are in many respects the same responsibilities as other technical projects of OpenStack. I am responsible for the development of the concept, manage the verification of the code, guide the developers, but I am away from the development itself. I inspire the community to develop around Ironic.
Question: What is the most essential thing you need to do to keep the project active?
Answer: I want to create a project that would help a large number of different hardware vendors feel comfortable and not feel that they are competing for proprietary bits or splitting the project into parts. Actually Ironic is a service for provisioning hardware or “bare iron” in OpenStack. All hardware vendors, such as HP, Dell and IBM, have in-house tools for managing hardware, such as the HP iLO server remote control or the Dell DRAC remote access controller. All of them add additional functionality to the standard IPMI interface specification, and some of them implement it a little differently.
Q: Can you explain the role of the Ironic project in OpenStack? Why is he important?
Answer: This can be explained in two different ways, I will try to choose one.
So far in the development of OpenStack, we have had to use other tools for deploying OpenStack, whether it's a couple of servers in your server cabinet or rack, or a whole data center. The hardware was necessarily external to OpenStack.
Ironic's role is to provide a hardware provisioning level that was previously not available in OpenStack. Based on this role, one of the goals of Ironic is to make TripleO (OpenStack-on-OpenStack) possible. All the tools you use to deploy a complete application in the cloud can be reused to deploy the cloud, which is ultimately another integrated application.
Question: What is the difference between Ironic and TripleO?
Answer: Ironic is a service that controls the power of a physical machine and writes an image on a machine. In addition, we have plans to make other management tasks possible. TripleO is located at a higher level in the stack and uses not only Ironic, but also many other services for deploying and managing the OpenStack cloud.
Q: Tell us about the Ironic community - who are contributing to the project?
Answer: HP, Red Hat, Mirantis, IBM and others. From the hardware vendors, both HP and IBM work with us. Until now, IBM has largely participated in a related project - creating an IPMI driver entirely in Python, which was moved from xCAT to the OpenStack community. It will be used by Ironic as a more scalable replacement of the ipmitool library inherited from the Baremetal code of the Nova component.
Q: Would you like to see greater involvement from hardware vendors?
Answer: Yes, I would like to. Sure, I'd like to see Dell. So far, they are not involved in our project. Ironic is a very young project, and perhaps some hardware vendors have not yet seriously participated in it, simply because the code was not ready. I think we are rapidly approaching the level of full code availability, when people can quickly join us and add their own hardware drivers.
Question: What has the Ironic community achieved at the moment?
Answer: Providing bare metal inherited a large number of restrictions, since it was originally an attempt to launch it as part of the Nova Compute process. For the last 4 months or so, we “pulled out” Ironic from Nova, turning it into a standalone OpenStack service. It has its own API service that can be scaled. It has its own message queue and server-side database, as well as the Conductor service - something in between the Nova Conductor and Nova Compute services. Most recently, we added support for DevStack and a disk image designer, as well as a python client; Tempest tests are under development. All this is necessary for any OpenStack project, but it seems to me that this is quite a great achievement for a small team over a period of just 4 months!
Question: What features does Ironic implement in the next release of OpenStack?
Answer: I can’t say exactly what Ironic functionality will be available in the Icehouse.
Question: If you could wave a magic wand right now and turn Ironic into exactly what you want it to be, what would it be at the moment?
Answer: I would like to see it scalable and fault tolerant. My definition of high availability in this context is its resilience to failure, and not the complete lack thereof. Hardware failure is inevitable, but Ironic must be able to recover from failures.
I would like to see in its structure different drivers from different hardware vendors. I would like the drivers themselves to be publicly available, and not proprietary. I would like Ironic to manage all kinds of hardware, starting with ARM devices and ending with a “big iron” - with all the existing variety of equipment that I personally may not be aware of. I would like other people to contribute in the form of drivers in order to make it all realizable and feasible.
Question: Are there any misconceptions regarding Ironic?
Answer: One of them is that Ironic is not a configuration database and that it will not store the history of states. I was asked: "You will monitor the fan speed, the temperature of the CPU, and then take any action if something overheats?". My answer is no!".
But the biggest misconception at the moment is that you can use Nova Compute or Ironic to realize unprotected multiplayer clouds based on bare metal. This is our goal, but for today we have not yet reached it. There are a number of significant hidden issues related to the security of unprotected cloud users operating on bare metal base. If I could wave a magic wand again, I would like all these problems to be solved. This will require a huge collaboration of hardware vendors to develop reliable hardware, as well as addressing security issues at the firmware level, and not just with respect to network load and user isolation. Some work on this is already underway, but I believe that we are still far from solving these problems. In the meantime, I want people to know that today you cannot run clouds under unprotected users on bare metal.
Question: Another question from the “I would like” series. Who would you like to see in the number of people involved in the Ironic project? Who is your perfect member at the moment?
Answer: Developers who know how to work with the open source community, because they are not always easy to find. People with good systems administration experience in addition to programming, since most of what we do at Ironic is low. Yes, it is written in Python, but we do a lot to integrate with processes such as DHCP, IPMI and PXE, and I will need people who understand the security issues of embedded software. The need for such team members, it seems to me, is currently not fully satisfied.
Question: Thank you for your time.
Answer: Thank you.
Source: https://habr.com/ru/post/204712/
All Articles