Top 10 DevOps Tools You MUST Use in 2025!

00:44:44
https://www.youtube.com/watch?v=4N_MWAQEzWg

Resumen

TLDRThe video provides a comprehensive overview of tools recommended for adoption in 2025 due to their significant influence and development in 2024. These include Fabric for AI enhancements, Devbox for managing CI/CD workflows, and Port for developer portals. Chain guard images are highlighted for their security in container management, and Celium stands out for Kubernetes networking, as chosen by the user community. Crossplane is recommended for resource management with its ability to build custom APIs, and New Shell is favored for terminal operations. The video suggests Argo CD as the preferred GitOps tool, emphasizing its reliability and support. These selections are based on the innovations these tools have brought and their potential impact on workflows moving forward.

Para llevar

  • 🔧 Adopt impactful tools from 2024 for 2025.
  • 🤖 Fabric chosen for enhancing AI prompts.
  • 📦 Devbox streamlines CI/CD dependency management.
  • 🌐 Chain guard images for secure container bases.
  • 📊 Port recommended for developer portals visualization.
  • 🖥️ Celium awarded for Kubernetes network management.
  • 📜 Crossplane excels in custom API resource management.
  • 🖱️ User preference highlights Celium.
  • 💻 New Shell redefines terminal scripting.
  • 🚀 Argo CD identified as leading GitOps tool.

Cronología

  • 00:00:00 - 00:05:00

    In 2025, the focus isn't necessarily on the 'best' tools but those that proved impactful in 2024, showing maturity and significant shifts in how we work. The presenter categorizes various tools mostly used in their focus areas over the past year, highlighting 10 key categories. The aim is to spotlight tools that have evolved and deserve consideration in the tool arsenal moving forward.

  • 00:05:00 - 00:10:00

    The speaker discusses tools in the Artificial Intelligence (AI) and continuous integration/continuous delivery (CI/CD) categories. They note that while many popular AI tools exist, their choice is 'Fabric,' a more obscure tool that significantly aids in creating effective prompts. For CI/CD, traditional workflow tools haven't innovated much, relying on execution of tasks, but they emphasize the importance of tools like 'Earthly' that integrate Dockerfile and Makefile, and 'Devbox,' a package management solution that also helps with installing project tools locally or in workflows, as essential advancements in the CI/CD area.

  • 00:10:00 - 00:15:00

    In the containers category, the speaker describes the mature state of containers, with little differentiation between options like Docker Desktop and alternatives such as Rancher Desktop, Podman, etc. The more critical development is the emphasis on using slimmer, safer images which provide better performance and security. The winner in this category is 'Chainguard Images' for their focus on slim and secure base images, setting a new standard for building containers.

  • 00:15:00 - 00:20:00

    The discussion on developer portals reveals that while many companies have attempted to build platforms in various ways over the years, Kubernetes provides a unified technological base that promises success in this realm. The category winner, 'Port,' is recognized for efficient data modeling and visualization while integrating with existing Kubernetes resources. Despite some noted shortcomings, particularly around its API and Kubernetes integration, 'Port' stands out in a field with many suboptimal choices.

  • 00:20:00 - 00:25:00

    For GitOps, there's been a consolidation with Argo CD emerging as the dominant choice due to its robust community support and development. Although both Argo CD and Flux had strong followings, Flux's backing dwindled, making Argo CD the safer choice. Argo CD has proven itself in the field, solidifying its status as the preferred tool.

  • 00:25:00 - 00:30:00

    When managing infrastructure resources, the discussion highlights a shift from older configuration management tools to the newer infrastructure as code tools, with Terraform leading previously. However, now the focus is shifting towards control plane tools like 'Crossplane,' which allows for defining APIs and managing any type of resource declaratively, leveraging Kubernetes' capabilities. Crossplane is chosen as the winner for its adaptability and maturity, despite emerging competitors like CueVela.

  • 00:30:00 - 00:35:00

    The conversation about state management formats and tools that assist in defining and transforming data structures focuses on languages like Helm, Customize, Cue, and KCL. The speaker favors KCL for its simplicity and functionality, noting that it has become their go-to for handling complex data definitions internally, distinguishing it as a preferred choice over more traditional solutions like Helm for custom applications.

  • 00:35:00 - 00:44:44

    Terminals and scripting tools have seen innovations with tools like Starship for dynamic prompts and Nushell, which serves as a powerful shell and scripting language, becoming indispensable for their scripting needs. Nushell is highlighted as a critical tool due to its versatility in handling task automation and data processing within terminal environments.

Ver más

Mapa mental

Vídeo de preguntas y respuestas

  • What is the purpose of the video?

    The video aims to recommend a list of impactful tools from 2024 that are advisable to adopt in 2025.

  • Why are AI tools emphasized in this video?

    The speaker highlights their personal preference for AI tools that assist in prompt generation, like Fabric, due to its impact on improving AI interactions.

  • What is the recommended AI tool for 2025?

    The recommended AI tool is Fabric, particularly for its ability to enhance prompt generation.

  • Which tool was highlighted for continuous integration and delivery in 2025?

    Devbox, utilizing Nix packages, was highlighted for its role in simplifying installation and management of project dependencies.

  • What are developer portals and which one was recommended?

    Developer portals are interfaces for developers to interact with platforms; Port was recommended for its data model and visualization capabilities.

  • What tool won the user choice award?

    Celium won the user choice award for its effective handling of Kubernetes networking.

  • What is the significance of chain guard images in the container category?

    Chain guard images are significant for providing slim and secure base images with zero vulnerabilities.

  • Why is crossplane recommended for resource management?

    Crossplane is recommended for its flexibility in creating custom APIs and managing resources effectively.

  • What is the trend for terminal tools usage and which one stands out?

    Terminals continue to be popular for reliability; New Shell was highlighted for its combination of simplicity and power as a shell, language, and data processor.

  • How does the video suggest approaching GitOps tools?

    The video suggests Argo CD as the most reliable choice for GitOps tools due to its standing and support.

Ver más resúmenes de vídeos

Obtén acceso instantáneo a resúmenes gratuitos de vídeos de YouTube gracias a la IA.
Subtítulos
en
Desplazamiento automático:
  • 00:00:00
    [Music]
  • 00:00:05
    today's the day I announce which tools I
  • 00:00:09
    recommend using in
  • 00:00:10
    2025 this is the list of the best of the
  • 00:00:13
    best creme de La Creme actually I'm
  • 00:00:16
    lying this is not the list of the best
  • 00:00:18
    tools but rather the list of tools that
  • 00:00:21
    made impact in 2024 the tools that
  • 00:00:24
    matured enough that I recommend this
  • 00:00:26
    permanent Tools in your tool belt it is
  • 00:00:29
    the list of tools that prove their worth
  • 00:00:31
    and should be adopted in 2025 some of
  • 00:00:34
    them are the best in their respective
  • 00:00:36
    category While others made a significant
  • 00:00:38
    shift in the way we work that I believe
  • 00:00:42
    deserve to be in the spotlight some are
  • 00:00:44
    small While others are massive some are
  • 00:00:47
    widely known While others are obscure so
  • 00:00:51
    the best of 2025 is more like the best
  • 00:00:54
    of the new tools and services or those
  • 00:00:56
    that matured enough to be used in 2025
  • 00:00:59
    that being said
  • 00:01:00
    every single tool I will mention today
  • 00:01:02
    deserves to be in the spotlight all
  • 00:01:04
    those I will present should be adopted
  • 00:01:06
    or at least considered by you
  • 00:01:09
    nevertheless I will choose a winner in
  • 00:01:11
    each category another important note is
  • 00:01:14
    that I will not cover every single
  • 00:01:16
    category that would be impossible Since
  • 00:01:19
    There is close to an infinite number of
  • 00:01:21
    ways we can categorize tools instead I
  • 00:01:24
    will focus on the categories that were
  • 00:01:26
    my focus throughout the past year there
  • 00:01:29
    are 10 categor ories presented in
  • 00:01:31
    alphabetical order we'll go through
  • 00:01:33
    artificial intelligence or AI continuous
  • 00:01:35
    integration and continuous delivery or
  • 00:01:37
    cacd containers developer portals giops
  • 00:01:40
    infrastructure as code and control
  • 00:01:42
    planes manifest management terminals
  • 00:01:45
    miscellaneous it contains the tools I
  • 00:01:48
    believe are valuable but I couldn't
  • 00:01:50
    place in any specific category and
  • 00:01:52
    finally there is users Choice which
  • 00:01:55
    contains the tools and services you
  • 00:01:58
    chose finally before we jump right into
  • 00:02:01
    it let me know that in case I covered a
  • 00:02:04
    tool mentioned here in some other video
  • 00:02:06
    there will be a link to it in the
  • 00:02:08
    description in case you want to explore
  • 00:02:10
    it further okay that's enough chitchat
  • 00:02:12
    I'm sure you prefer that we jump right
  • 00:02:14
    into it so let's do just that starting
  • 00:02:17
    with
  • 00:02:20
    ai ai is everywhere right it's a hype as
  • 00:02:24
    well everyone is jumping into it
  • 00:02:27
    everyone is investing money into it yet
  • 00:02:29
    no one is sure what it's for it's a mess
  • 00:02:32
    nevertheless this is neither the time
  • 00:02:34
    nor the place to dive into the current
  • 00:02:36
    state of AI what matters is that there
  • 00:02:38
    is a bunch of AI chats like open AI
  • 00:02:40
    better known as Chad GPT Google gini uh
  • 00:02:44
    clothe by anthropic and other Solutions
  • 00:02:46
    and then there is AMA and an infinite
  • 00:02:50
    number of other models that can be
  • 00:02:51
    self-hosted if you're a software
  • 00:02:53
    engineer which you almost certainly are
  • 00:02:56
    if you're watching this video you're
  • 00:02:57
    likely already using GitHub co-pilot or
  • 00:03:00
    a similar solution that helps you write
  • 00:03:02
    code even if that code is yo all of
  • 00:03:05
    those are great but my choice goes to
  • 00:03:08
    something much much more obscure much
  • 00:03:11
    smaller and mostly irrelevant to most
  • 00:03:14
    over the past year or so the AI related
  • 00:03:16
    tool I use the most is fabric it helps
  • 00:03:20
    with prompts now that might sound silly
  • 00:03:23
    but only until we see the differences in
  • 00:03:25
    a responses between normal and really
  • 00:03:29
    good prompts with fabric we don't have
  • 00:03:31
    to think how to use different prompts
  • 00:03:33
    how to communicate with different AIS
  • 00:03:35
    how to write prompts and so on and so
  • 00:03:37
    forth all we have to do is use one of
  • 00:03:39
    the existing patterns directly or copy
  • 00:03:42
    and modify them to serve our specific
  • 00:03:44
    needs from there on whenever we need
  • 00:03:47
    something from an AI we just pass our
  • 00:03:49
    input to one of Fabric's patterns and
  • 00:03:52
    get what we need it is truly awesome and
  • 00:03:55
    I use it all the time and I'm
  • 00:03:57
    proclaiming it the winner in this
  • 00:04:00
    category it's a small project that could
  • 00:04:02
    not even imagine competing with the big
  • 00:04:05
    guys I mentioned earlier yet it is the
  • 00:04:08
    tool I use all the time try it out the
  • 00:04:12
    next category is
  • 00:04:17
    cicd continuous integration and delivery
  • 00:04:19
    tools or as I like to call them
  • 00:04:22
    workflows haven't changed in decades
  • 00:04:25
    there is no innovation in that area but
  • 00:04:27
    only small often marginal imp
  • 00:04:29
    improvements GitHub actions Argo
  • 00:04:31
    workflows GitHub cicd tecton and other
  • 00:04:35
    workflow tools are all essentially the
  • 00:04:38
    same none of them made any significant
  • 00:04:40
    Innovation since Jenkins they all do the
  • 00:04:43
    same thing they all execute tasks one
  • 00:04:46
    after another or in parallel they might
  • 00:04:48
    be using a different syntax to Define
  • 00:04:50
    workflows they might have different
  • 00:04:52
    maintenance cost and performance
  • 00:04:54
    benefits yet they're all the same I lost
  • 00:04:57
    interest in them whichever you're using
  • 00:05:00
    right now is probably good enough I am
  • 00:05:02
    much more interested in what's happening
  • 00:05:04
    outside workflow tools we saw dagger as
  • 00:05:08
    an attempt to Define workflows that can
  • 00:05:10
    run both locally and inside traditional
  • 00:05:12
    workflow tools we can Define what we
  • 00:05:14
    need in typescript python or go and run
  • 00:05:16
    it locally or inside workflows like
  • 00:05:18
    GitHub actions the main advantage is
  • 00:05:20
    that we can Define execution of tasks
  • 00:05:23
    once and run them or run it that
  • 00:05:26
    workflow anywhere and from that
  • 00:05:28
    perspective it effectively replaces
  • 00:05:30
    shell scripts or make file dagger is
  • 00:05:33
    potentially great yet I'm not yet
  • 00:05:36
    convinced that it's the way to go we
  • 00:05:38
    also saw rise in adoption of quite a few
  • 00:05:40
    tools that essentially try to replace
  • 00:05:43
    make file Earthly is certainly
  • 00:05:45
    interesting it combines Docker file and
  • 00:05:47
    make file into a single definition
  • 00:05:49
    that's a great idea since those two are
  • 00:05:52
    one of the most commonly used formats so
  • 00:05:55
    combining them makes a lot of sense
  • 00:05:58
    probably then there are tasks and just
  • 00:06:01
    that are trying to reinvent make file
  • 00:06:03
    which frankly should have been retired
  • 00:06:05
    long time ago both task file and just
  • 00:06:08
    file are arguably better formats than
  • 00:06:11
    make file I use both mainly since I
  • 00:06:15
    cannot yet make up my mind which I like
  • 00:06:18
    more we also saw the rise of cargo which
  • 00:06:20
    tends to solve a real problem of
  • 00:06:23
    promotions of releases managed by
  • 00:06:25
    github's tools specifically Argo CD
  • 00:06:27
    promotions became kind of comp licated
  • 00:06:30
    with gibs and we definitely need a tool
  • 00:06:32
    that will help us in that area yet I
  • 00:06:35
    think that cargo even though it features
  • 00:06:37
    some very interesting New Concepts is
  • 00:06:39
    still too young and not polished enough
  • 00:06:41
    to be videly used it will probably get
  • 00:06:44
    there sometime but not today the biggest
  • 00:06:47
    Improvement in the cicd area are NYX
  • 00:06:53
    packages Nyx in general is awesome and
  • 00:06:56
    its packaging system is amazing we
  • 00:06:58
    finally got packages that work
  • 00:07:00
    everywhere be it Linux Mac or Windows we
  • 00:07:03
    finally have a way to define all the
  • 00:07:05
    tools that the project needs and the
  • 00:07:07
    capability to get those tools no matter
  • 00:07:10
    whether we need them on a laptop in
  • 00:07:12
    workflows or anywhere else now when we
  • 00:07:15
    talk about NYX I do not refer to Nyx OS
  • 00:07:17
    it came too late most of us do not care
  • 00:07:19
    about operating systems running on top
  • 00:07:21
    of servers and are not willing to change
  • 00:07:24
    Mac OS or Windows as operating system on
  • 00:07:27
    our laptops n packages are What
  • 00:07:31
    attracted me to Nyx and we get them
  • 00:07:33
    through NYX shell that can run anywhere
  • 00:07:37
    the problem however is that NYX syntax
  • 00:07:39
    is just too complicated for a user who
  • 00:07:42
    did not choose to dedicate its life to
  • 00:07:44
    Nyx it is a programming language of
  • 00:07:47
    sorts that is just too complicated and
  • 00:07:50
    obscure to learn quickly so my love for
  • 00:07:54
    NYX packages and hate for its syntax led
  • 00:07:57
    me to Dev box it's awesome it g gives us
  • 00:07:59
    all the advantages of Nyx shell and
  • 00:08:02
    packages with a very simple and
  • 00:08:04
    straightforward syntax all we have to do
  • 00:08:07
    is Define devbox Json either manually or
  • 00:08:10
    through devbox CLI and you'll notice
  • 00:08:12
    that every single repository I created
  • 00:08:15
    during the last year or two has the
  • 00:08:18
    devbox Json file devbox in it is the
  • 00:08:21
    First Command I execute when I start
  • 00:08:23
    working on a new project and devbox
  • 00:08:26
    shell is the First Command I execute
  • 00:08:28
    when I want to work on a project I use
  • 00:08:30
    it to install the tools I need to work
  • 00:08:32
    on a project no matter whether that's on
  • 00:08:34
    my laptop or inside workflows like
  • 00:08:37
    GitHub actions or anywhere else Dev box
  • 00:08:39
    is awesome I cannot recommend it enough
  • 00:08:42
    and it is a clear winner in this
  • 00:08:44
    category it is not a tool that you will
  • 00:08:46
    use to run cicd workflows but it should
  • 00:08:49
    be used inside the workflows to install
  • 00:08:51
    all the tools you'll need that alone
  • 00:08:53
    would not make it special but when we
  • 00:08:56
    combine it with the usage of those same
  • 00:08:58
    packages locally or anywhere else where
  • 00:09:00
    you might be working with a project it
  • 00:09:03
    becomes indispensable also if you need
  • 00:09:06
    those same packages as a container or
  • 00:09:08
    Dev container it can also generate
  • 00:09:10
    Docker file and Dev container Json files
  • 00:09:13
    if you haven't already I strongly
  • 00:09:15
    strongly recommend adopting Dev books if
  • 00:09:18
    you're watching my videos and following
  • 00:09:20
    along my instructions you're almost
  • 00:09:22
    certainly already using devbox since
  • 00:09:25
    it's in every one of my setup
  • 00:09:28
    instructions next let's talk about
  • 00:09:33
    containers containers are now boring no
  • 00:09:36
    matter whether we are talking about
  • 00:09:38
    running them locally or in real servers
  • 00:09:41
    most of us do not care anymore which
  • 00:09:44
    container engine is used inside
  • 00:09:46
    kubernetes we care even less which one
  • 00:09:48
    is used with manag services like Google
  • 00:09:50
    Cloud run Azure container apps Lambda or
  • 00:09:52
    whichever other you might be using it's
  • 00:09:56
    a commodity or low implementation detail
  • 00:09:59
    similarly I do not think it matters what
  • 00:10:01
    you're using on your laptop is it Docker
  • 00:10:04
    desktop or one of the Alternatives like
  • 00:10:06
    Rancher desktop or podman or nerd cutle
  • 00:10:09
    or something else now I know that some
  • 00:10:12
    of you have strong feelings and claim
  • 00:10:15
    that one is much better than the other
  • 00:10:17
    if you're one of those all I can say is
  • 00:10:19
    that I don't care anymore and I don't
  • 00:10:21
    think that many do they all work fine
  • 00:10:25
    they all run containers some of them
  • 00:10:28
    have additional Banks and whistles but I
  • 00:10:30
    don't think any of them are life
  • 00:10:32
    changing now there is something
  • 00:10:34
    important and very successful going on
  • 00:10:37
    around containers even though it might
  • 00:10:38
    not seem that important at the first
  • 00:10:40
    glance that something is Slim and safe
  • 00:10:43
    images over time most of us learned
  • 00:10:46
    about the importance of having slim
  • 00:10:48
    images we learned that containers are
  • 00:10:50
    not VMS that containers do not need a
  • 00:10:52
    whole operating system as a result we
  • 00:10:55
    stopped using Bas images like Ubuntu
  • 00:10:57
    sento SN addas we learned that
  • 00:10:59
    the less we put into container images
  • 00:11:02
    the better we learned that the best
  • 00:11:04
    operating system for images is not to
  • 00:11:07
    have an operating system or to be more
  • 00:11:10
    precise to use distroless images when
  • 00:11:12
    possible and minimal images when not we
  • 00:11:15
    learned that most of security
  • 00:11:17
    vulnerabilities come from stuff we don't
  • 00:11:20
    use with that in mind the winner this
  • 00:11:22
    year is chain guard or to be more
  • 00:11:25
    precise chain guard images they are the
  • 00:11:28
    images you should using as base images
  • 00:11:30
    for whatever you're building they're
  • 00:11:32
    slim and they're safe if you use them as
  • 00:11:35
    the base your images will be small in
  • 00:11:38
    size they will be more performant and
  • 00:11:39
    will be safe with no CV e unless the
  • 00:11:45
    code of your application introduces them
  • 00:11:47
    I strongly recommend using chain guard
  • 00:11:49
    images as your base images and that's
  • 00:11:52
    the winner next in line are Developer
  • 00:11:57
    portals last couple of years we saw
  • 00:12:00
    Resurgence of platform engineering I
  • 00:12:02
    intentionally said Resurgence since
  • 00:12:05
    we've been building developer platforms
  • 00:12:07
    for decades now every bigger company I
  • 00:12:09
    worked with had some kind of a platform
  • 00:12:12
    that helped developers be more
  • 00:12:13
    productive that could be anything from a
  • 00:12:16
    server with scripts junkins jobs some
  • 00:12:18
    kind of a UI that provides links to
  • 00:12:20
    Services observability dashboards or
  • 00:12:23
    anything else developer platforms have
  • 00:12:25
    been around for a very very long time
  • 00:12:29
    even though they might not have been
  • 00:12:30
    called like that nor people working on
  • 00:12:33
    them have been called platform Engineers
  • 00:12:35
    now the important note here is that most
  • 00:12:37
    of them failed or were only moderately
  • 00:12:41
    successful so here's the important
  • 00:12:43
    question is it going to be any different
  • 00:12:45
    this time I think it will this time we
  • 00:12:48
    have standards mainly in the form of
  • 00:12:50
    kubernetes this time we finally all
  • 00:12:52
    agreed what is the underlying technology
  • 00:12:55
    sitting below those platforms this time
  • 00:12:57
    we know how to create control planes on
  • 00:12:59
    top of kubernetes how to extend apis
  • 00:13:01
    with crds how to make all the tools
  • 00:13:04
    interoperable by forcing them to be
  • 00:13:06
    kubernetes native how to do service
  • 00:13:08
    Discovery by quaring apis and so on and
  • 00:13:11
    so forth ultimately most of us made a
  • 00:13:13
    choice to adopt kubernetes and most of
  • 00:13:16
    the vendors chose kubernetes as the base
  • 00:13:19
    on top of which their tools are running
  • 00:13:21
    kubernetes is the base on top of which
  • 00:13:24
    we should be building platforms there is
  • 00:13:26
    no I repeat no doubt about that
  • 00:13:29
    kubernetes is what might make us
  • 00:13:31
    successful in building platforms it's
  • 00:13:33
    what might make Decades of failures a
  • 00:13:36
    success kubernetes future depends on
  • 00:13:39
    that kubernetes was not designed to be
  • 00:13:40
    used directly but as a base on top of
  • 00:13:43
    which we build a final solution and that
  • 00:13:46
    final solution is a platform fine-tuned
  • 00:13:49
    to the needs of a company now this
  • 00:13:52
    category is not about developer
  • 00:13:53
    platforms mostly because having a
  • 00:13:55
    platform means combining an infinite
  • 00:13:57
    number of tools and processes this
  • 00:13:59
    category is about developer portals
  • 00:14:01
    defined as user interfaces through which
  • 00:14:04
    developers can communicate with the
  • 00:14:06
    platform within this category of portals
  • 00:14:08
    or simply graphical user interfaces
  • 00:14:11
    sitting on top of platforms typically
  • 00:14:13
    built by extending kubernetes we have
  • 00:14:16
    quite a few tools to choose from there
  • 00:14:18
    is backstage as one of the fastest
  • 00:14:19
    growing cncf projects it is widely
  • 00:14:22
    adopted both by end users and vendors
  • 00:14:25
    who are using it as the interface for
  • 00:14:28
    whichever tool they selling everyone is
  • 00:14:30
    jumping onto backstage but also
  • 00:14:32
    realizing that it is very expensive to
  • 00:14:35
    manage and not really built to leverage
  • 00:14:37
    kubernetes it is based mostly on static
  • 00:14:40
    files and has quite a few issues related
  • 00:14:43
    to the dynamic nature of what we do
  • 00:14:45
    today the shining star of Backstage is
  • 00:14:48
    its extensibility through the plug-in
  • 00:14:50
    system it has a massive number of ready
  • 00:14:52
    to go plugins which are for most part of
  • 00:14:55
    very poor quality it allows us to create
  • 00:14:58
    our own plugins which can do anything we
  • 00:15:00
    need them to do but at a significant
  • 00:15:03
    development and maintenance cost it's
  • 00:15:06
    great yet bad at the same time there are
  • 00:15:09
    also managed backstage offerings like
  • 00:15:11
    Roi that solve some of the problems
  • 00:15:13
    mainly maintenance cost if you choose
  • 00:15:15
    backstage and you can use SAS and you
  • 00:15:18
    can afford it roie is great there are
  • 00:15:21
    many alternatives to Backstage mostly As
  • 00:15:23
    commercial often SAS offerings there is
  • 00:15:26
    cortex Ops level and many others the the
  • 00:15:29
    one I ended up using the one I strongly
  • 00:15:31
    recommend and the one that is the winner
  • 00:15:33
    is Port when it was created it
  • 00:15:36
    understood that the main purpose of
  • 00:15:38
    portals is not to pretend that they are
  • 00:15:40
    platforms but rather to sit on top of
  • 00:15:42
    platforms as such it was focused on
  • 00:15:44
    providing a data model ways to visualize
  • 00:15:47
    it and to trigger events when we fill in
  • 00:15:50
    forms based on those data models it
  • 00:15:53
    hasn't been doing much more and that was
  • 00:15:55
    a good thing I liked that it had a clear
  • 00:15:58
    Focus
  • 00:15:59
    what I did not like is that it ignored
  • 00:16:01
    kubernetes just as with backstage
  • 00:16:03
    kubernetes is more of an afterthought
  • 00:16:06
    than the design choice or design
  • 00:16:08
    decision I understand why it is so Port
  • 00:16:11
    like many others focused on where we are
  • 00:16:13
    instead of focusing on where we will be
  • 00:16:16
    it focused on the present rather than
  • 00:16:19
    the future that was somehow mitigated
  • 00:16:21
    with the ability to discover kubernetes
  • 00:16:23
    crds and CRS so I chose port for two
  • 00:16:27
    main reasons first it is is a data model
  • 00:16:29
    with the ability to present data as a
  • 00:16:31
    graphical user interface rather than
  • 00:16:33
    trying to be everything to everyone that
  • 00:16:36
    fits perfectly into my idea that portals
  • 00:16:39
    are not platforms but rather graphical
  • 00:16:42
    representations of platforms the second
  • 00:16:45
    reason was the addition of
  • 00:16:47
    discoverability the moment they added
  • 00:16:48
    the option to discover crd is sitting in
  • 00:16:51
    my cluster that serves as a control
  • 00:16:53
    plane I could stop redefining the same
  • 00:16:56
    things over and over and over and over
  • 00:16:57
    and over again instead of telling Port
  • 00:17:00
    what is what it could simply discover
  • 00:17:02
    Services sitting in my platforms the
  • 00:17:05
    problem however is that integration with
  • 00:17:07
    kubernetes does not seem to get much
  • 00:17:08
    love and is still in early stages on top
  • 00:17:11
    of that ports API is bad forcing me to
  • 00:17:14
    use the UI to manage the portal while I
  • 00:17:17
    want the UI mostly for consumers not for
  • 00:17:20
    service providers or portal Builders
  • 00:17:23
    nevertheless among all the bad choices
  • 00:17:25
    Port is still in my opinion the best as
  • 00:17:28
    as long as using SAS and its price are
  • 00:17:31
    acceptable that being said I still hope
  • 00:17:34
    I really hope that one of the existing
  • 00:17:36
    Solutions will get it right and or
  • 00:17:39
    actually or that the new one will emerge
  • 00:17:42
    given that I don't want to choose the
  • 00:17:44
    less bad option but one that is really
  • 00:17:46
    good anyways today that's Port so it's a
  • 00:17:50
    winner of this category now let's talk
  • 00:17:52
    about
  • 00:17:55
    gitops there was a gitops war we had and
  • 00:17:59
    still have Argo CD and flux as dominant
  • 00:18:02
    tools and a few failed attempts like
  • 00:18:04
    runer Fleet we were spending tremendous
  • 00:18:07
    amount of time trying to deduce whether
  • 00:18:09
    Argo CD is better than flux those
  • 00:18:11
    conversations are now mostly over Argo
  • 00:18:14
    City W the main patron of flux was V
  • 00:18:19
    Works which went bankrupt leaving flux
  • 00:18:21
    with no finances to continue There are
  • 00:18:23
    still contributions flowing in but they
  • 00:18:25
    cannot be compared to the love Argo CD
  • 00:18:28
    is getting in into it the company
  • 00:18:29
    started Argo CD or to be more precise
  • 00:18:32
    acquired it continues investing in it
  • 00:18:35
    even though it never had any commercial
  • 00:18:37
    interest in it those contributions are
  • 00:18:38
    now much smaller mainly because many
  • 00:18:40
    other companies started investing in it
  • 00:18:42
    redhead was there for a long time cfres
  • 00:18:45
    joined the effort afterwards a was
  • 00:18:48
    formed by core Argo CD maintainers
  • 00:18:50
    mostly X employees of into it and quite
  • 00:18:53
    a few others even though I would argue
  • 00:18:55
    that Fluxx has a better design because
  • 00:18:57
    it does and better architecture and is
  • 00:19:00
    based on better ideas it is
  • 00:19:03
    unfortunately slowly dying living Argo
  • 00:19:06
    City is the last man standing hence the
  • 00:19:08
    winner is Argo CD if you're already
  • 00:19:11
    using flux there might not yet be a
  • 00:19:13
    compelling reason to change if you're
  • 00:19:16
    using something other than nargo CD or
  • 00:19:18
    flux you might be making a terrible
  • 00:19:20
    mistake if you're just starting Argo CD
  • 00:19:23
    is a Clear Choice don't even bother
  • 00:19:25
    looking for an alternative at least not
  • 00:19:27
    today that that might change in the
  • 00:19:29
    future but today if you're looking for a
  • 00:19:32
    GitHub tool choose Argo City next let's
  • 00:19:35
    talk about Resource
  • 00:19:39
    Management people tend to talk about
  • 00:19:41
    configuration management infrastructure
  • 00:19:42
    is code and though these days control
  • 00:19:45
    planes as very different concepts
  • 00:19:47
    they're partly right those are very
  • 00:19:50
    different yet they serve the same
  • 00:19:52
    purpose there are just different
  • 00:19:54
    approaches to Resource Management that
  • 00:19:57
    is becoming clearer since the lines
  • 00:19:59
    between configurations applications and
  • 00:20:01
    infrastructure are becoming blured it is
  • 00:20:04
    often close to impossible to separate
  • 00:20:06
    configuration from resources it's hard
  • 00:20:08
    to distinguish what is an application
  • 00:20:10
    what is infrastructure we are moving
  • 00:20:12
    towards immutability and everything
  • 00:20:14
    manage through apis no matter whether
  • 00:20:16
    that's networking or processes or
  • 00:20:18
    hyperscalers or anything it's all about
  • 00:20:21
    managing resources and what changed over
  • 00:20:24
    time is how we manage resources long
  • 00:20:27
    long time ago we would write scripts
  • 00:20:29
    that would manage resources scripts were
  • 00:20:31
    complicated to write and close to
  • 00:20:33
    impossible to maintain given the number
  • 00:20:35
    of permutations between the desired and
  • 00:20:38
    the actual state of resources as a
  • 00:20:40
    result we got configuration management
  • 00:20:42
    tools like CF engine and later on Chef
  • 00:20:44
    and puppet only for most of them to be
  • 00:20:46
    eventually replaced by Anvil they all
  • 00:20:49
    have two things in common they're based
  • 00:20:51
    on promise Theory allowing us to specify
  • 00:20:54
    what we want instead of specifying how
  • 00:20:56
    to do something it was the job of those
  • 00:20:58
    those tools to convert the actual into
  • 00:21:00
    the desired State the second thing they
  • 00:21:03
    all had in common is that they were
  • 00:21:05
    designed to work directly with servers
  • 00:21:08
    and that everything is mutable we call
  • 00:21:11
    those tools configuration management
  • 00:21:13
    tools now two things happened afterwards
  • 00:21:16
    we started adopting immutability and
  • 00:21:19
    more importantly apis that's when
  • 00:21:21
    terraform and later on pulumi were born
  • 00:21:24
    instead of trying to SSH into servers
  • 00:21:27
    and do whatever needs to be done they
  • 00:21:29
    focused on apis we would Define what we
  • 00:21:32
    want the desired State and those tools
  • 00:21:35
    would reconcile it into the actual state
  • 00:21:37
    by talking with apis be it AWS serer go
  • 00:21:41
    cloud or any other we call those
  • 00:21:44
    infrastructures code tools now to be
  • 00:21:46
    clear anible and other configuration
  • 00:21:48
    management tools could talk to apis just
  • 00:21:52
    as terraform and other infrastructures
  • 00:21:54
    code tools could SSH still the primary
  • 00:21:57
    focus of configuration a management
  • 00:21:59
    tools is SSH and infrastructures code
  • 00:22:02
    tools communication with existing apis
  • 00:22:05
    what all those tools mentioned so far
  • 00:22:08
    have in common is that they all assume
  • 00:22:10
    that there are definitions in some
  • 00:22:12
    binary sitting somewhere let's say a
  • 00:22:15
    laptop we execute that binary that takes
  • 00:22:17
    those definitions and does whatever
  • 00:22:19
    needs to be done with the resources on
  • 00:22:21
    the other end so that those resources
  • 00:22:23
    are in the same state as what's defined
  • 00:22:26
    in those manifests today we have eight
  • 00:22:28
    third wave of resource management tools
  • 00:22:31
    there is not yet a clear winner on how
  • 00:22:33
    we call them so I'll call them control
  • 00:22:35
    plane tools their primary focus is on
  • 00:22:38
    creating custom apis instead of defining
  • 00:22:40
    what we want on the same level as the
  • 00:22:42
    resources we're trying to manage those
  • 00:22:44
    tools allow us to become service
  • 00:22:46
    providers they accomplish that by
  • 00:22:48
    enabling us to create apis that
  • 00:22:50
    consumers can talk to now we can Define
  • 00:22:53
    what an application or a database or a
  • 00:22:54
    cluster or anything else is as an API
  • 00:22:58
    and point with its schema consumers do
  • 00:23:01
    not need to deal with low-level details
  • 00:23:03
    like VPC subnets or whatever else
  • 00:23:05
    providers like AWS or kubernetes give us
  • 00:23:09
    there is a clear internal separation of
  • 00:23:11
    concerns with some people in a company
  • 00:23:13
    creating services and others consuming
  • 00:23:15
    those Services consumers consume by
  • 00:23:18
    invoking apis and providers provide by
  • 00:23:22
    Building Services behind those apis
  • 00:23:25
    while infrastructure as code tools are
  • 00:23:26
    based on Lessons Learned From cons uming
  • 00:23:28
    Services creating by others control
  • 00:23:32
    plane tools are based on Lessons Learned
  • 00:23:34
    From Building Services there are many
  • 00:23:37
    Tools in this category one of the older
  • 00:23:39
    ones is open application model or om
  • 00:23:42
    which started with a focus on
  • 00:23:44
    applications but since then has been
  • 00:23:46
    extended to any type of resources the
  • 00:23:49
    most commonly used implementation of
  • 00:23:50
    open application model is Cube it is the
  • 00:23:53
    tool I used heavily in the past however
  • 00:23:56
    it never managed to get over the initial
  • 00:23:59
    open application model spec resulting in
  • 00:24:02
    a failure to extend itself beyond the
  • 00:24:04
    initial scope as a result it had a great
  • 00:24:08
    start but since then its popularity and
  • 00:24:11
    usage dropped we also got cluster API
  • 00:24:14
    which enabled us to create cluster as a
  • 00:24:16
    service cluster API is amazing but is
  • 00:24:20
    focused only on kubernetes clusters it
  • 00:24:23
    is solving only a fraction of the
  • 00:24:25
    problem at the same time it is a very
  • 00:24:28
    opin
  • 00:24:28
    way to manage kubernetes clusters which
  • 00:24:31
    might or might not fit everyone's needs
  • 00:24:33
    we also got crossplane which Justice
  • 00:24:35
    Cube VA allows us to create our own API
  • 00:24:38
    end points as well as processes that
  • 00:24:40
    listen to those end points and perform
  • 00:24:42
    the reconciliation need to convert the
  • 00:24:44
    actual into the desired state of
  • 00:24:47
    resources what makes crossplane more
  • 00:24:49
    powerful is its flexibility it allows us
  • 00:24:52
    to create any type of a definition with
  • 00:24:54
    very few if any restrictions the major
  • 00:24:57
    downside of crossplane or to be more
  • 00:24:59
    precise crossplane compositions was that
  • 00:25:02
    it tried to Define processes that do the
  • 00:25:05
    reconciliation as yaml while yaml is
  • 00:25:08
    great for defining the desire state it
  • 00:25:10
    is a very bad solution for defining the
  • 00:25:13
    reconciliation of the states that was
  • 00:25:15
    fixed later on with the addition of
  • 00:25:17
    functions that allow us to use any
  • 00:25:19
    programming language to Define
  • 00:25:20
    compositions the newest addition to the
  • 00:25:22
    family whose members build or contribute
  • 00:25:26
    towards building control planes is ra it
  • 00:25:29
    is very very similar to cross plan
  • 00:25:31
    compositions and might one day be a
  • 00:25:34
    better solution but right now it is in
  • 00:25:37
    its infancy all those tools within the
  • 00:25:39
    control plane subcategory are based on
  • 00:25:41
    kubernetes it would not make sense to
  • 00:25:43
    pick anything else since kubernetes
  • 00:25:46
    already has built in and in a way
  • 00:25:48
    standard mechanism that provides the
  • 00:25:50
    base all those tools through leverage it
  • 00:25:53
    has the machine that allows creation of
  • 00:25:56
    crds which extend the core API and it
  • 00:25:58
    has controllers that are effectively
  • 00:26:01
    processes that listen to events in the
  • 00:26:03
    API and perform whichever operations
  • 00:26:05
    that should perform to reconcile the
  • 00:26:08
    states hence we can consider control
  • 00:26:10
    planes extended versions of kubernetes
  • 00:26:13
    clusters so the winner in this category
  • 00:26:16
    must be from the control plane
  • 00:26:18
    subcategory that's where we are going we
  • 00:26:20
    cannot continue ignoring the option of
  • 00:26:22
    building our own apis and providing
  • 00:26:24
    services to others in our company we
  • 00:26:27
    could just as well call these category
  • 00:26:29
    developer platforms except that there is
  • 00:26:31
    more to platforms than managing
  • 00:26:32
    resources nevertheless we are moving
  • 00:26:35
    away from here are the files execute
  • 00:26:38
    this and something will happen to I
  • 00:26:40
    built a service for managing this and
  • 00:26:43
    that it is exposed through this API use
  • 00:26:46
    it the winner in this category is
  • 00:26:48
    obviously crossplane with a note that
  • 00:26:50
    Crow might take over if it grows from an
  • 00:26:53
    infant into something more mature now
  • 00:26:55
    before I continue let me stress that I'm
  • 00:26:57
    part of the crossplay project you can
  • 00:26:59
    interpret that in any way you like you
  • 00:27:01
    can say Victory is biased Crossman is
  • 00:27:03
    not the best choice alternatively you
  • 00:27:05
    can also say Victor would not be
  • 00:27:07
    involved with Crossman if he does not
  • 00:27:09
    think it's the best project in its
  • 00:27:11
    category it's up to you to choose which
  • 00:27:13
    interpretation you would like to take or
  • 00:27:15
    even to come up with the third one the
  • 00:27:17
    reason I'm saying all this is
  • 00:27:19
    transparency I want to be clear when
  • 00:27:21
    there might be some conflict of interest
  • 00:27:23
    or when I might be subjective all in all
  • 00:27:26
    this year's winner for the Resource
  • 00:27:28
    Management categories crossplane and now
  • 00:27:30
    we can move to State Management
  • 00:27:36
    formats we've seen explosion of tools
  • 00:27:39
    and formats that ultimately do the same
  • 00:27:42
    thing I'm referring to tools that allow
  • 00:27:44
    us to Define data in some format and
  • 00:27:47
    output that data into XML Json yaml or
  • 00:27:50
    whichever other format we need we had
  • 00:27:53
    such tools for a long time and their
  • 00:27:56
    number increased drastically with the
  • 00:27:58
    adoption of kubernetes which expects us
  • 00:28:01
    to define the desired state in yaml or
  • 00:28:04
    Json one of the first if not the first
  • 00:28:07
    such tool to emerge is no other than
  • 00:28:10
    Helm it went through some drastic
  • 00:28:12
    changes over its lifetime the current
  • 00:28:14
    model is based on go templating which is
  • 00:28:16
    essentially a templating engine that
  • 00:28:19
    allows us to produce the output in any
  • 00:28:21
    format as such it is completely
  • 00:28:24
    oblivious that we are in case of Helm
  • 00:28:26
    dealing with data and that is its
  • 00:28:29
    biggest weakness we can essentially put
  • 00:28:32
    anything as a template and go templating
  • 00:28:34
    will happily convert it to text the fact
  • 00:28:36
    that we use Helm to generate yaml is not
  • 00:28:39
    of much importance go templating creates
  • 00:28:42
    text output that's why we need some
  • 00:28:44
    silly constructs like indent to ensure
  • 00:28:47
    that text can be interpreted as yaml the
  • 00:28:50
    main strengths of Helm are its age and
  • 00:28:53
    adoption it's been around forever and
  • 00:28:55
    everyone using kubernetes is familiar
  • 00:28:56
    with it it also became an unofficial
  • 00:28:58
    standard resulting in almost all open-
  • 00:29:01
    Source projects and vendors creating
  • 00:29:04
    Helm charts as a way to distribute their
  • 00:29:06
    stuff whichever third party app we need
  • 00:29:08
    to run there is almost certainly a Helm
  • 00:29:11
    chart for it on top of all that Helm is
  • 00:29:14
    also a packaging tool it allows us to
  • 00:29:16
    package those templates those charts and
  • 00:29:19
    publish them to a registry that
  • 00:29:21
    simplifies distribution a lot and not
  • 00:29:24
    all other tools we'll discuss today have
  • 00:29:27
    it if you would be picking a winner for
  • 00:29:28
    the format that should be used for third
  • 00:29:30
    party apps we could stop here Helm is
  • 00:29:33
    the undisputable king in that area
  • 00:29:36
    however I will not make it that easy so
  • 00:29:38
    instead we'll limit the scope to your
  • 00:29:41
    apps which format should you use for
  • 00:29:44
    your applications in that scenario I
  • 00:29:46
    would argue the helm is the worst option
  • 00:29:48
    the sole idea that we should use free
  • 00:29:50
    text templating to generate data sounds
  • 00:29:53
    silly yaml is structured data and it
  • 00:29:56
    makes perfect sense to use a language
  • 00:29:58
    anguage or a format that understands
  • 00:30:00
    data not long after Helm emerged we got
  • 00:30:03
    customized as a project maintained in
  • 00:30:06
    one of kubernetes 6 or special interest
  • 00:30:09
    groups the executable was even added to
  • 00:30:11
    cube control so we do not even need to
  • 00:30:14
    install anything it's simply there
  • 00:30:16
    customize takes a different approach
  • 00:30:18
    when compared with Helm instead of
  • 00:30:20
    applying templates it applies patches
  • 00:30:23
    with customized we work with pure yaml
  • 00:30:25
    and then patch that yaml to get
  • 00:30:26
    variations we need those p could be
  • 00:30:28
    applied to everchanging TGs or to
  • 00:30:30
    differences present in various
  • 00:30:32
    environments or anything else really
  • 00:30:34
    customize is great but only a relatively
  • 00:30:36
    small projects that fits well with many
  • 00:30:39
    internal applications if you define an
  • 00:30:41
    application we are likely not going to
  • 00:30:43
    have many variations of the Manifest of
  • 00:30:46
    that app we would typically change tags
  • 00:30:48
    with each release and have some
  • 00:30:50
    differences in various environments we
  • 00:30:52
    might be changing host a number of
  • 00:30:53
    replicas and a few other pieces of data
  • 00:30:55
    when moving the application from let's
  • 00:30:58
    let's say staging to production
  • 00:31:00
    environment all in all if what you have
  • 00:31:02
    does not require great number of
  • 00:31:03
    variations or some complex logic
  • 00:31:05
    customiz might be the winner still that
  • 00:31:08
    is often not the case so let's continue
  • 00:31:10
    we also got an explosion of formats or
  • 00:31:12
    to be more precise languages that were
  • 00:31:15
    designed in some form or another to work
  • 00:31:17
    with data structures there is Carwell Y
  • 00:31:20
    which is great format but failed to see
  • 00:31:22
    adoption outside Tano users there is
  • 00:31:24
    also Q which tends to be very popular
  • 00:31:26
    among go developers since it is a
  • 00:31:28
    language built on top of go to make
  • 00:31:30
    things more interesting we got timoni
  • 00:31:33
    which introduced patterns that make
  • 00:31:35
    working with q more focused on
  • 00:31:37
    kubernetes as well as a packaging
  • 00:31:39
    mechanism that publishes its packages
  • 00:31:42
    into oci images the problem with q and
  • 00:31:45
    through it timony is that it has its
  • 00:31:47
    quirks it is not always easy to
  • 00:31:50
    understand what should be done and how
  • 00:31:51
    it should be done and it can be very
  • 00:31:53
    challenging language to be picked up by
  • 00:31:56
    people not interested in dedicating a
  • 00:31:58
    lot of time to master it we also got KCl
  • 00:32:01
    which is similar to q but easier to
  • 00:32:04
    learn and without some of the Q
  • 00:32:06
    silliness unlike Q anyone can pick it up
  • 00:32:09
    in no time then there is pickle which
  • 00:32:12
    looked promising it's great for those
  • 00:32:14
    used to Java or groovy but fail to gain
  • 00:32:17
    my trust at least there is nothing truly
  • 00:32:19
    wrong with it it's just that I liked
  • 00:32:21
    some other languages more finally we
  • 00:32:23
    also got CD Kates which allows us to use
  • 00:32:26
    general purpose languages to out put
  • 00:32:28
    data mainly yaml CD Kates can be great
  • 00:32:31
    or
  • 00:32:32
    horrible depending on which language we
  • 00:32:35
    choose for example the experience with
  • 00:32:37
    typescript is excellent while go results
  • 00:32:39
    in too much boilerplate code mostly
  • 00:32:41
    related to Transformations there are of
  • 00:32:43
    course many many other formats and
  • 00:32:45
    languages it would take too much time to
  • 00:32:47
    go through all of them those I mentioned
  • 00:32:49
    are likely best contenders to be the
  • 00:32:51
    formats you are already using or that
  • 00:32:54
    you should switch to I used all those
  • 00:32:56
    formats extensively especially
  • 00:32:57
    throughout 2024 my goal was to gain
  • 00:33:00
    enough experience with all of them to be
  • 00:33:02
    able to make a decision which one will
  • 00:33:04
    be my choice that turned out to be KCl
  • 00:33:07
    so I'm proclaiming it a winner it took
  • 00:33:10
    me a while to rewrite almost everything
  • 00:33:12
    I have but I finally finally did it I
  • 00:33:16
    still keep simple stuff as pure yl but
  • 00:33:18
    everything else is now in the KCl format
  • 00:33:21
    to clarify when I say everything I mean
  • 00:33:23
    everything that is not being distributed
  • 00:33:25
    outside my organization and excluding
  • 00:33:28
    third party apps I use those are still
  • 00:33:30
    Helm and are likely going to continue
  • 00:33:33
    being Helm charts better I like it or no
  • 00:33:36
    that's irrelevant so KCl for the win now
  • 00:33:39
    let's talk about
  • 00:33:43
    terminals it's not a secret that I spent
  • 00:33:46
    significant amount of time in terminals
  • 00:33:48
    I think that everyone should use
  • 00:33:50
    Terminals and I'm sure that many do it's
  • 00:33:53
    much easier more reliable and
  • 00:33:55
    reproducible to execute commands than to
  • 00:33:57
    click button
  • 00:33:58
    uis are great for
  • 00:34:01
    observing not necessarily for operations
  • 00:34:04
    it would be impossible to compile the
  • 00:34:06
    list of all the tools worth using inside
  • 00:34:08
    terminals so I will focus only on those
  • 00:34:11
    that made a significant impact on me and
  • 00:34:14
    my workflow lately Starship is one of
  • 00:34:17
    those that I fell in love the moment I
  • 00:34:20
    discovered it it's a very easy yet very
  • 00:34:23
    effective way to generate Dynamic
  • 00:34:25
    prompts it is now on all my machines and
  • 00:34:27
    I could not live without it unless
  • 00:34:29
    someone shows me a better one then there
  • 00:34:31
    is charm which is a group of projects
  • 00:34:33
    that do not necessarily have much in
  • 00:34:34
    common except that they are all somehow
  • 00:34:37
    related to terminals if you want to
  • 00:34:39
    convert your go code into an interesting
  • 00:34:42
    or not interesting amazing CLI there is
  • 00:34:46
    H if you would like to send emails
  • 00:34:48
    through terminal there is pop if you
  • 00:34:50
    would like to convert a shell script
  • 00:34:52
    into an interactive CLI that is gum
  • 00:34:55
    there are many many other projects and I
  • 00:34:57
    cannot recommend them enough charm is
  • 00:34:59
    amazing and you should check it out if
  • 00:35:02
    you haven't already the next in line is
  • 00:35:04
    salich that makes terminal multiplexing
  • 00:35:07
    easy I like easy my brain is not
  • 00:35:09
    developed enough to grasp tools that
  • 00:35:12
    require a lot of learning Visage I was
  • 00:35:14
    up and running with all my terminal tops
  • 00:35:16
    and paints in no time I love it and I
  • 00:35:19
    use it all the time finally there is no
  • 00:35:21
    shell which arguably made the biggest
  • 00:35:24
    impact on me when creating scripts and
  • 00:35:26
    CIS I was always storn between the
  • 00:35:28
    Simplicity of bash and the power of go
  • 00:35:31
    bash is great for very simple scripts
  • 00:35:33
    and go is amazing when working on
  • 00:35:35
    complicated CIS much of what I do is
  • 00:35:38
    somewhere in the middle much of it is
  • 00:35:40
    more than dozen or so lines when bash
  • 00:35:42
    becomes silly but less than thousands of
  • 00:35:44
    lines when go is a great choice Nell
  • 00:35:47
    changed all that it is a shell a
  • 00:35:50
    language and data processor at the same
  • 00:35:53
    time with it we effectively write a
  • 00:35:55
    script that compiles so we the detect
  • 00:35:57
    errors right away it outputs everything
  • 00:36:00
    as data instead of text so piping
  • 00:36:03
    outputs as inputs and processing those
  • 00:36:05
    inputs is a breeze it works everywhere
  • 00:36:07
    it's awesome and easily this year's pick
  • 00:36:10
    as a winner after a while I started
  • 00:36:12
    using it exclusively for all my scripts
  • 00:36:15
    if you're following along my
  • 00:36:16
    instructions from other videos you
  • 00:36:18
    likely already saw it in action even
  • 00:36:20
    though that simpler scenario where it
  • 00:36:22
    doesn't shine as much as it could try it
  • 00:36:25
    use it you'll love it now the the second
  • 00:36:27
    to last category is
  • 00:36:32
    miscellaneous this is the category with
  • 00:36:34
    all the tools that I feel are important
  • 00:36:36
    but I could not fit into any other
  • 00:36:38
    category there are three of them that
  • 00:36:41
    made a splash nuts has been around for a
  • 00:36:44
    while more often than not used by third
  • 00:36:46
    party tools it's small lean fast and
  • 00:36:49
    easy to set up and use pu sub server
  • 00:36:51
    since I think that we should all be
  • 00:36:53
    moving our applications to the publish
  • 00:36:55
    And subscribe model at least when more
  • 00:36:57
    complex systems are concerned nuts is a
  • 00:37:00
    great choice it might not be as advanced
  • 00:37:03
    as let's say Kafka and other similar
  • 00:37:04
    tools still more often than not we don't
  • 00:37:07
    need such complex tools that do much
  • 00:37:09
    more than what we need nuts is often
  • 00:37:12
    just what is needed then there is vasm
  • 00:37:14
    cloud that was moved from sandbox to
  • 00:37:16
    incubation status in cncf I'm not yet
  • 00:37:19
    convinced that vasm is ready for Prime
  • 00:37:21
    but if you're already using vasm vasm
  • 00:37:24
    Cloud might be a good choice of a tool
  • 00:37:26
    to Define and manage your apps finally
  • 00:37:29
    the last one in this somehow random
  • 00:37:31
    category is Dapper a cnca project which
  • 00:37:34
    was always always awesome what makes it
  • 00:37:37
    special is that it recently graduated
  • 00:37:40
    now it is in the group of elite projects
  • 00:37:43
    that proved itself worthy to be
  • 00:37:45
    alongside other projects cncf considers
  • 00:37:48
    mature now to be clear it's close to
  • 00:37:50
    impossible to declare a winner in this
  • 00:37:52
    category since all those are completely
  • 00:37:54
    different they are all great in their
  • 00:37:56
    own Merit still I promise to pick one
  • 00:38:00
    and that one is nuts I'm picking it
  • 00:38:03
    mostly because I feel that even though
  • 00:38:05
    it is commonly used by third party apps
  • 00:38:07
    it is not getting as much love from end
  • 00:38:10
    users as it should okay there is only
  • 00:38:12
    one category left and that one might be
  • 00:38:15
    the most interesting one let's see what
  • 00:38:16
    are your
  • 00:38:21
    choices when I asked you to send me your
  • 00:38:23
    favorite tools and services frankly I
  • 00:38:26
    did not expect such a massive response
  • 00:38:28
    you wrote quite a few public comments
  • 00:38:30
    and sent me a massive number of private
  • 00:38:32
    messages so thank you all for
  • 00:38:35
    participating I compiled a list of
  • 00:38:38
    hundreds of tools and services you're
  • 00:38:39
    using now it would take a lot of time to
  • 00:38:42
    list all of them so I'll limit it to the
  • 00:38:45
    top 10 sorted in descending order from
  • 00:38:48
    most to least mentioned so that would be
  • 00:38:51
    Argos CID kubernetes cium crossplane git
  • 00:38:54
    ESO terraform orop tofu cloud port and V
  • 00:38:58
    cluster so according to you Argo City is
  • 00:39:01
    a clear winner it's the tool that most
  • 00:39:04
    of you recommend however I will not
  • 00:39:06
    Proclaim it a winner of this category
  • 00:39:08
    simply because I already did that when
  • 00:39:10
    we talked about gitops then there is
  • 00:39:12
    kubernetes which I assume all of you are
  • 00:39:14
    using and like including all those who
  • 00:39:16
    voted for Argo CD I will ignore that one
  • 00:39:19
    as well simply because it's a base that
  • 00:39:21
    everyone uses these days if kubernetes
  • 00:39:23
    would be a contender it would win every
  • 00:39:25
    single year at least in the foreseeable
  • 00:39:28
    future the next in line is celium and
  • 00:39:30
    that's the one I will proclaim the
  • 00:39:31
    winner celium is amazing it became the
  • 00:39:35
    cni of choice for most kubernetes
  • 00:39:37
    clusters there are many people who run
  • 00:39:39
    kubernetes without even knowing that
  • 00:39:41
    it's networking is handled by celium for
  • 00:39:43
    many of us celium was the first
  • 00:39:45
    introduction to BPF many other projects
  • 00:39:48
    adopted BPF TKS in part to the success
  • 00:39:51
    of celium for some celum removed the
  • 00:39:54
    need for a service mesh for others it
  • 00:39:56
    provides valuable source of networking
  • 00:39:58
    metrics then there are those who use it
  • 00:40:00
    primarily to manage Network policies and
  • 00:40:02
    so on and so forth almost everyone is
  • 00:40:04
    using celum for some reason or another
  • 00:40:07
    some without even knowing they're using
  • 00:40:09
    it it is truly the de facto standard
  • 00:40:12
    when networking is concerned and it
  • 00:40:13
    rightfully deserves to be the winner for
  • 00:40:16
    the user Choice
  • 00:40:20
    Award there we have it folks those are
  • 00:40:22
    the winners of 2025 we got a relatively
  • 00:40:25
    significant AI tool fabric It Is by no
  • 00:40:28
    means as important or as big as what
  • 00:40:30
    we've been seeing in the AI space yet it
  • 00:40:34
    is the tool I'm using all the time it
  • 00:40:36
    shows the tremendous difference between
  • 00:40:37
    bad and good prompts with the cicd space
  • 00:40:41
    I chose NYX packages specifically devbox
  • 00:40:44
    it is not strictly speaking a cicd tool
  • 00:40:48
    we can use it to Define packages mostly
  • 00:40:50
    CLI needed to work on a project those
  • 00:40:53
    can be used when working on a project
  • 00:40:54
    locally or inside cicd workflows you can
  • 00:40:57
    also use it to build container images or
  • 00:40:59
    Dev containers if you work in remote
  • 00:41:01
    environments it's amazing and I use it
  • 00:41:04
    in every single project I work on when
  • 00:41:06
    containers are concerned I also made an
  • 00:41:08
    unusual Choice instead of choosing
  • 00:41:10
    between Docker Rancher desktop and other
  • 00:41:13
    container engines I chose chainu guard
  • 00:41:15
    images chainu guard made a significant
  • 00:41:18
    change by providing slim and zero
  • 00:41:21
    vulnerability based images developer
  • 00:41:23
    portals are one of the biggest Trends
  • 00:41:25
    right now to be more precise developer
  • 00:41:28
    platforms are the area where we see
  • 00:41:30
    companies putting a lot of investment
  • 00:41:32
    and portals and an important part of
  • 00:41:34
    them unfortunately I don't think we have
  • 00:41:37
    a great solution in that area among
  • 00:41:39
    mostly bad choices I think that Port is
  • 00:41:42
    the best choice so I proclaimed it a
  • 00:41:45
    winner in that category gups was one of
  • 00:41:47
    the areas where we had a lot of
  • 00:41:48
    discussions in the past we would often
  • 00:41:50
    enter into heated arguments where some
  • 00:41:53
    would be defending Argo CD others would
  • 00:41:55
    be advocating for flux and and some
  • 00:41:57
    would be desperately trying to introduce
  • 00:41:59
    other Solutions I think those debates
  • 00:42:01
    are now over Argo CD is the winner in
  • 00:42:04
    general as well as my choice in this
  • 00:42:06
    category that was an easy one even
  • 00:42:08
    though I think that flux is better in
  • 00:42:12
    quite a few aspects when it comes to
  • 00:42:13
    Resource Management it is now clear that
  • 00:42:16
    resources should be managed by
  • 00:42:18
    kubernetes no matter whether those
  • 00:42:19
    resources run in the same cluster or are
  • 00:42:22
    elsewhere kubernetes is becoming the
  • 00:42:24
    standard and the default choice for
  • 00:42:26
    managing any and I repeat any type of
  • 00:42:29
    resources anywhere within that category
  • 00:42:32
    there is nothing even close to
  • 00:42:34
    crossplane new projects are emerging so
  • 00:42:36
    that might change in the future but for
  • 00:42:38
    now it is an Easy Choice as well another
  • 00:42:41
    area where you saw a lot of changes are
  • 00:42:43
    State Management formats there is an
  • 00:42:46
    explosion of languages and formats that
  • 00:42:48
    ultimately all serve the same purpose
  • 00:42:51
    they all transform something into
  • 00:42:54
    structured data predominantly yl new
  • 00:42:56
    languages are being born and wrappers
  • 00:42:58
    around existing languages are being
  • 00:43:00
    created it's as if everyone is trying to
  • 00:43:04
    replace God templating used in Helm the
  • 00:43:06
    one that impressed me the most by being
  • 00:43:09
    relatively simple yet providing
  • 00:43:11
    everything I think we might need is KCl
  • 00:43:14
    it is the winner and my recommendation I
  • 00:43:17
    moved all my complex definitions to KCl
  • 00:43:20
    when it comes to working in terminals
  • 00:43:22
    New Shell is a language I ended up using
  • 00:43:25
    almost every day it is a shell and the
  • 00:43:27
    language and a data processor at the
  • 00:43:30
    same time all the scripts I'm writing be
  • 00:43:32
    it simple or complex are now Nal it's
  • 00:43:35
    awesome and I strongly recommend it the
  • 00:43:38
    miscellaneous category is a weird one
  • 00:43:40
    since it contains unrelated tools I
  • 00:43:42
    could not play anywhere else but I feel
  • 00:43:44
    deserve to be considered even though it
  • 00:43:46
    is not very new and it is widely used
  • 00:43:48
    among third party projects I think that
  • 00:43:50
    nuts does not get the attention and
  • 00:43:52
    adoption it deserves at least among end
  • 00:43:54
    users so proclaiming it a and
  • 00:43:57
    recommending it is my way to put it
  • 00:44:00
    under the spotlight finally your choice
  • 00:44:02
    the user choice is cium there's probably
  • 00:44:05
    no need to say what it is nor why it was
  • 00:44:07
    chosen it's the winner and I'm glad you
  • 00:44:10
    chose it otherwise I would need to add
  • 00:44:12
    11th category networking only to ensure
  • 00:44:15
    that it is one of the winners so That's
  • 00:44:18
    all folks that was the list of the tools
  • 00:44:20
    you should consider adopting in 2025 you
  • 00:44:23
    probably use at least some if not all of
  • 00:44:25
    those if you aren't get to it put them
  • 00:44:28
    into your toolkit thank you for watching
  • 00:44:29
    see you in the next one cheers
Etiquetas
  • AI
  • CI/CD
  • Developer Portals
  • Containers
  • Resource Management
  • GitOps
  • Kubernetes
  • Networking
  • Terminals
  • 2025 Recommendations