Interview with Michael Basnight, OpenStack Trove Technical Project Manager

Original author: Rafael Knuth
  • Transfer
We present the eighth of a series of interviews with technical managers of the OpenStack project on the Mirantis blog. Our goal is to educate the wider community of technical experts and help people understand how they can contribute to and benefit from the OpenStack project. Naturally, the following is the viewpoint of the interviewee, not Mirantis.

The following is an interview with Michael Basnight, OpenStack Trove Technical Project Manager.

Mirantis: Tell us about yourself.

Michael Busnight: I am a lead engineer at Rackspace, working on the OpenStack project 24 hours a day, 7 days a week. I have been with Rackspace for over 7 years. He was responsible for data analytics, cloud website hosting, large-scale provisioning systems, and now I'm working on Trove. Thanks to Rackspace, I can now give 100% of my energy to an open source database. I live in Berkeley, California, working either on the veranda of my house with a cup of coffee, or next to my wife and 10-month-old son Alexander.

Question: What is your history of relationships with OpenStack? Why are you participating in the project?

Answer: When we created Trove at Rackspace, neither I nor other developers thought about OpenStack. We started to manipulate the full stack of Java, OSGI, Zookeeper, etc. ... standard Java programs. At that time, Rackspace decided to bet on OpenStack. Seeing that our future is connected with OpenStack, I decided (at that time the leading developer and leader) to curtail the tests we started to test the concept and start using the Nova controller as a provisioning mechanism. We encountered some difficulties along the way, as the OpenStack initiative was only just beginning to develop. But I had the opportunity to introduce changes to almost all OpenStack projects in the process of their development, and now we have a serious project on top of OpenStack,

Question: What are your responsibilities as technical manager of the Trove project?

Answer: Heh, good question. This is almost a recipe. I start with the basics of product ownership and management so that the project concept evolves in the direction that is best for the community. I add a pinch of project management, developing a detailed plan and tracking who is working on what. Shuffle and slowly add people management to make sure that they truly fulfill their commitments in accordance with their promises. I add a generous portion of code verification, a pinch of code “-1” and, if LUCKY, when everything is ready, I glaze it all a bit with what I develop myself.

Question: What is the role of Trove in OpenStack? Why is the Trove project important?

Answer: The Trove project is important because data storage is important. You cannot name a single project that does not need to store or cache data. Trove is a one-stop storage solution. Whether it's Redis, a real-time data warehouse with a high degree of availability, or MySQL to store all the information about your pet store, or MongoDB to store Marconi's telegraph messages. The concept of Trove is to provide all this, to ensure that the clusters are configured and work online. Trove also offers backup and recovery of services implemented with its help, and functions such as automatic backup / restore at a certain point in time are currently under development.

Question: What is truly unique and breakthrough in the Trove project?

Answer: A huge number of experts in the field of data storage systems work on Trove. We attracted key people from the community to implement many projects and services. These include developers who administer MySQL since version 3.23. We are recruiting people who have written projects for implementing clusters in MySQL. We consult with Redis experts during its implementation. We are experts in the field of development, as well as in the field of data storage and processing. If we do not know how to do this, we know those who know it.

Question: Tell us about the Trove community - who is cooperating with you?

Answer: Several high-class developers from Rackspace stood at the origins, and then the developers of HP cloud services picked up the idea. They really helped give an impetus to development. Since we have already moved into the “incubation period”, we have involved Mirantis in the development. Also involved in the project are UnitedStack and several developers from New Zealand. In addition, there are people who are interested in the intricacies of clustering. I also had some great conversations with people who are very interested in Galera Cluster and Tungsten Replicator. Starting from 2 time zones, in just a few months we have swept the whole world!

Question: What are the Trove community achievements today?

Answer: We have a service that has been functioning in Rackspace for some time. We have turned database provisioning into science! We know everything about the operation of the MySQL service. In addition, we defined our role for the community as a mechanism for provisioning relational and non-relational databases. We have done a lot in establishing relationships with companies and promoting the product. I'm curious to see what will happen to Trove six months after the release of the Icehouse release. I expect to see in it such functionality as clustering, the ability to change the configuration, as well as the embedded Redis data warehouse.

Question: What features will Trove provide in the OpenStack Havana release?

Answer: The Trove component works with MySQL. It provides provisioning instances, security groups, as well as the creation and modification of database users. It supports resizing, and we control that the configuration and service work after that in normal mode. It allows you to back up and restore. Actually, it has everything that you would like to see in the MySQL service, managed from one instance.

We spent a lot of time for Trove to become more flexible in setting up for this iteration, and our efforts were concentrated on doing everything necessary for the possibility of integration. We eliminated a large number of accumulated technical debts, including adding the ability to integrate with RHEL and create a template configuration file. Last but not least, we have added the ability to integrate with Heat, which will be optional in the Havana release.

Question: What would you like people to know about this project?

Answer: I want people to know that we are more than just MySQL as a service. In the Icehouse release, we plan to fully support clustering for Cassandra and MongoDB, as well as MySQL clustering. This includes master-follower replication and Galera / Tungsten symmetric replication and clustering.

Question: Are there any common misconceptions regarding Trove?

Answer: Again, I can repeat about the misconception that we are simply “MySQL as a service” or “SQL as a service”. We are more than that. We want to provide the provisioning and maintenance of SQL and NoSQL data warehouses. We don’t want us to be labeled as “highly specialized project”.

Question: You mentioned the concept of Trove. Can you tell us more about this?

Answer: Clusters. Clusters Clusters The most promising development direction for Trove I see is its transformation into an API for provisioning / clustering of data warehouses. People do not want to use their own clusters. They just want their product to work. Trove wants to provide ease of installation, configuration and maintenance of these clusters.

Question: Who is your target group?

Answer: I think you can guess that I will answer ... people who need clusters! The future is data backup. The same can be said about clustering. Clients need their data and guarantees that they will be available when they are needed. This will help support clustering.

Question: What are the necessary prerequisites for the launch and proper operation of Trove at the moment (in the future)?

Answer: The most serious change in the next six months is the Heat project. We hope that Heat will support the installation of clusters. This may mean joining the Heat team, and they are very good guys, so there will be no difficulties! Other than that, all you need to do is follow the standard OpenStack requirements. Once we are better integrated into DevStack, which will happen very soon, to get started, you only need to launch DevStack and use Trove.

Question: Who would you like to see among the participants in the Trove project?

Answer: I would like to see two types of people in the project. People passionate about data. Those who know storage technologies such as Galera Cluster, or Cassandra, or some other. And, more importantly, people who want to control the system. Operators have the best understanding of the system as a whole, because they need to know all of its "running gears". I had a huge amount of intellectual conversations with our engineers and operators, which allowed me to improve the Trove product.

Question: What functionality needs to be improved and tested?

Answer: I think it is necessary to improve the support of the Heat component due to its novelty. I see that Trove can do a lot with Heat in the future. I would like to see more people who would test our support for RHEL / CentOS. The Mirantis team provides tremendous help in setting up RHEL / CentOS in full, for which I am grateful. I would like other companies to join in this, which will allow us to remain honest in matters of supporting RHEL / CentOS.

Question: How exactly can people get started?

Answer: Funny. We are just working to simplify the conclusion of Trove to working mode. We use a number of non-standard methods, and one of my responsibilities as a technical project manager is standardization. We have just finished work on providing support for DevStack, and it is already on the way! This, along with other features that will be available soon in DevStack, will allow you to use Trove without any additional configuration. But for now, you need to expand diskimage-builder, create a disk image and upload it to Glance. After that, everything is as simple as using our Trove client application for provisioning an instance. In addition, some auxiliary scripts are available in the repository for integration with Trove, which I will be updating in the coming days to ensure the possibility of working with the new integration with DevStack. After a few weeks, when passions subside, it will be ridiculously easy to start working with Trove. And if someone has questions, we # openstack-trove have an army of smart people and one or two robots.

Question: Thank you very much, Michael!

Answer: Not at all!

Also popular now: