Category: Editorial

  • Real World DevOps

    One of the more interesting aspects of my job is talking about building software with customers. As I do more of this over time, the discussion have changed from “what is DevOps?” to “how do we get started in DevOps” to “how do we improve our DevOps process?” These days so many customers have bought into some aspects of DevOps that the last one is quite common. There are still a number of companies trying to get started with DevOps in some way, but very few people I speak with have no idea of what DevOps is or how why it can improve your company.

    Of course, I’m not sure there’s a great standard definition of DevOps. I often lean towards Gene Kim’s Three Ways or Microsoft’s broad view of the principles. Ultimately, most of the practical things that I find myself discussing are how to smooth the process of integrating code from developers and getting it transferred to production. Everyone wants this to go smoother, with mistakes caught earlier, and with compliance with whatever internal rules, regulatory mandates, and customer demands exist.

    Grant had an interesting conversation with a few customers recently and wrote a blog about the conversation. There are a few items that resonate with me, and things that I like to emphasize to customers. The experiences from Stuart and Chris are common in any journey to improve your software process.

    People need to test and measure. While testing has gained traction in the application development world, it’s still fairly new to database people. However, as people try to reduce mistakes and problems, often from simple issues, they adopt more testing. They also start to realize that having instrumentation to help measure the impact of changes is important to learn what to test and watch out for.

    Starting small is important as well. While there are many, many commonalities in all my customers, the way they solve these with their teammates, and the order in which they solve them, varies. Even if you were to end up with the same VCS, build process and deployment script as someone else, the journey will be different. The challenges from internal people and processes, the evolution of habits in your team with vary. You need to find your own path.

    So start small, with a few people, let them learn and then teach others. Let them build up knowledge, habits, and skills that others can adopt. That’s how most things work in the world. A few people discover something and then many follow their footsteps. If you want to build a better software process, find someone to start doing it. When they find something that works, the rest of the organization can stand on their shoulders.

    Steve Jones

    Listen to the podcast at Libsyn, Stitcher or iTunes.

  • Confidential VMs

    Ever since we started to offload workloads to hardware that we didn’t physically control, there have been security concerns. I remember when this started with application service providers and web workloads. This has continued to be an issue as more and more types of workloads have moved to cloud vendors and other hosting providers.

    Google is trying to ameliorate some of the concerns by offering customers confidential VMs. These are special types of VMs, using encryption and hardware capabilities to protect the workloads from any unauthorized access. I don’t know to what extent this practically protects a workload compared to a non-confidential VM, as the details are a bit confusing. I’m sure there is some extra protection, but the weak point in most cases here is still likely the humans that use credentials to access the VM. I’d suspect a determined attacker would try to hack the sysadmin and their laptop rather than the VM itself.

    In any case, Google is trying to ensure the added encryption doesn’t cause any workload degradation. Hardware can likely help her, but I’m not sure that you can perform encryption and decryption without using more resources. There might be minimal impact, but there has to be some resource impact. At least compared to a non-confidential VM.

    I’m glad there is research and work still happening to find ways to improve security for systems that we might no longer control. I think that’s increasingly the trend. Whether you go with a cloud vendor like AWS, Azure, GCP, etc., or you look to host with a Rackspace like provider, more and more of our infrastructure is being outsourced, and I don’t know that the trend will reverse itself anytime soon. Even if it does, the more we can provide security hurdles against unauthorized access, the better.

    Steve Jones
    Listen to the podcast at Libsyn, Stitcher or iTunes.

  • Research on the Changing Workplace

    One of the things that I have admired about Microsoft is their desire to invest in research and try to learn more about subjects that relate to their core business. Microsoft Research has a number of fascinating projects underway all the time, and I enjoy browsing the site once in awhile.

    However, they don’t just limit research to this group. They are a data driven culture, and it seems they are constantly using metrics and instrumentation to measure how the entire business works and to glean insights into how they might work better. Recently I saw a report on Microsoft’s new work-from-home workforce, driven by a group that helps companies better work with their own employees.

    Work has changed for many people in the world in 2020, and very dramatically for some. At the same time, many companies have had to drastically alter they continue to do business, recognizing that many technical employees could work from home and be as effective as they are in an office. Microsoft has analyzed their own data, and continue to do so as they seek to understand how the pandemic has changed things for employees. They do find some longer work days, more networking, and more social events.

    Interestingly, meetings are slightly shorter, which is something I’ve noticed as well. It’s almost as if the effort to walk to a conference room means people have invested more in spending an hour there. With remote meetings, I find more people are willing to cut short a meeting and end it early when they can. The downside is that there are more meetings, at least for this group.

    The importance of management is something they noticed, as well as a larger burden on managers that they must deal with as they have more meetings with employees. I’ve actually gone to every other week meetings with my manager for a time, though I’m not sure I like that. I’ll give it another month and then see what I think. I appreciate the extra work for managers, but I also know that managers should be enabling employees, which should be a big part of their job. If that means a bit of hand holding, I think that’s important.

    The two most interesting things in the piece to me were that some of the authors’ clients are planning on a two year work from home, and that work-life boundaries are blurring, especially on weekends. For the latter, I think we sometimes find work to be an anchor, and we may need to spread our 40-ish hours across 7 days to deal with the challenges of family during the week. All the challenges of children, maybe other adults and meetings, etc. can impact when we work. The one thing we certainly need to avoid is having 40-ish hours become 50-ish.

    For the former, if you plan to be at home for two years, why not just make that forever? I’d think the adaptations employees, departments, and workflows make after that amount of time might not be worth undoing.

    Steve Jones

    Listen to the podcast at Libsyn, Stitcher or iTunes.

  • A Broken Streak

    SQL Saturday #1 was in Orlando, as the first actual event that has gotten us to nearly 1000 in just over a decade. I didn’t go to the first one, but I did go to SQL Saturday #8 and many more since that time. I’ve been luck to get back to Orlando a few times, and I look forward to more in the future.

    Andy Warren announced that this year’s SQL Saturday in Orlando is cancelled. This breaks the longest streak so far, and I’m sad to see that. I wasn’t likely to go to Orlando this year, though this city is always on my list. I was hoping the event would take place in some form, but I understand the desire to step back and examine how to move forward.

    A virtual SQL Saturday isn’t quite the same for me. I’ve presented at one, and submitted to a couple more to support them, but I don’t love this format, and I think it becomes hard to make it a special event for your city. I know a few more are planned for 2020, and I hope someone runs a SQL Saturday with only local speakers for that area. That might make it feel like a local event.

    Since the pandemic hit, and since SQL Saturday #950 in Victoria, I’m not sure anyone has held a live event. I’m not sure when the next live SQL Saturday will be, but I’m hoping that it happens in 2021 sometime. I don’t think anyone can plan for this year, but I am hopeful we will find a way to get back to some live interaction next year.

    We have until 14 Mar, 2021 to avoid losing a whole year of SQL Saturdays and breaking a streak of having them every year since 2007. I’m confident that if it’s safe, someone will organize and run a SQL Saturday as soon as they can. I’m also sure it will be well attended as I’m guessing many of you are looking forward to a live event as much as I am. I just hope I get the chance to speak, because I will certainly be ready for a live audience.

    Steve Jones

    Listen to the podcast at Libsyn, Stitcher or iTunes.