OpenStack needs more than one hat

    Posted by Nick Chase

    It seems like it's getting fashionable to scold OpenStack. It’s natural, every technology suffers a growth disorder when it is already advanced enough for people to use it, but not so advanced as to meet 100% of everyone’s expectations.
    OpenStack seems to be at this stage right now. Gartner complains that OpenStack lacks focus . One former member complains that the Ceilometer is ... less than it should . And now Matt Asay of ReadWrite has joined this majority, believing that OpenStack should be a leader , and nominating Red Hat for this position.

    Matt believes that OpenStack cannot evolve properly without a strong hand, and that Red Hat is best suited for the role of such a strong hand that would guide (or even control) the development direction of OpenStack.

    I in no way deny the relevance of Red Hat in the OpenStack project. Still, this is participant number 1 in terms of the amount of code entered into the database; nobody argues with that. But I would definitely disagree with some of Matt's arguments.

    “No” is not always the right answer

    Matt believes there are serious challenges to the current development path of OpenStack.

    He says that someone has to step forward and take responsibility, which sometimes implies the cessation of new developments:

    On the one hand, the damage of OpenStack is caused by chaotic product management. Top open source software projects such as Linux have a strong team of few managers who know the price of the word "No." But OpenStack strives to say “Yes” to every new feature that does not guarantee compatibility or tackle complex and tedious problems. This may be due to a lack of clear vision of what OpenStack is and is not, according to Gartner .

    Part of the problem mentioned in this statement may be due to some confusion not with what OpenStack is / is not, but rather with what it relates to and what does not apply to OpenStack. The fact that a project has been started and work is ongoing on it does not mean that it is included in OpenStack.

    In addition, weren't the same experts complaining just a few months ago who are now complaining that OpenStack is doing too much, that we are not innovating fast enough to compete? Even Matt is unsatisfied with the lack of innovation in OpenStack. Innovation is the result of the opportunity to develop the project, because it seems important. You cannot have a truly innovative environment in which one vendor can complete a project unilaterally.

    OpenStack members are not amateur developers

    In part, Matt believes, the problem is that the OpenStack developers are working on what interests them, and not on solving the aforementioned "complex and tedious problems." But he makes an assumption, which is generally incorrect with respect to OpenStack.

    But partly, the reason may be that OpenStack may be developing thanks to developers who do this out of a love of development, and not out of concern for the end user / cloud operator. This is a problem common to many open source projects in which participants are ranked higher than users . Such a philosophy is unconstructive and irrational.

    I am not hinting that OpenStack cannot better meet the needs of users and operators; this is a completely different task. But submissionOpenStack developers as people who participate in the project “for the love of development” are fundamentally wrong. Yes, almost all of them do their job with great enthusiasm. And almost all of them get paid for it.

    OpenStack is different from many other open source projects. Most of its participants are involved in the project on behalf of companies, and these companies do not pay their salaries because of the kindness of their corporate heart, but because they have a business task that needs to be solved. These companies expect OpenStack to perform a certain function for real customers, and when this does not happen, they pay developers to create the necessary functionality. This means that OpenStack developers are working on real-life tasks for real users and operators, while undoubtedly enjoying their work and therefore are not looking for another. They are not engaged in the development of sports interest.

    The winner gets the biggest jackpot

    Matt also claims that any project that starts as OpenStack will ultimately lead to a “winner gets it all” situation.

    The development of free software is usually carried out on the principle of “the winner gets everything.” At Linux, the winner (Red Hat) got it all. In Android, the winner (Samsung) got it all. Open source projects implemented by organizations with the participation of a large number of product vendors, which are to be further developed, almost always end with the “winner gets everything” scenario. Yes, this has not happened in Hadoop yet, but this is an exception, not a rule.

    You know what they say in the stock market: "past performance does not speak of future results." OpenStack is a free software development project sponsored by a nonprofit organization, that's true. But for a number of reasons, it differs from Linux or Android.

    Firstly, there are simply too many leading players in it, whose future in a certain non-trivial way depends on OpenStack. IBM, Dell, AT&T, Cisco, HP and 23 other companies (including, of course, Red Hat) have committed themselves not only to provide financial support to OpenStack, but also “to coordinate strategies with the OpenStack mission.” Both HP and IBM launched public clouds, at least partly powered by OpenStack. They are not going to concede and will not allow any other vendor to dictate the direction of the project to them.

    There are also a number of individual participants to be reckoned with. Despite the incredible estimates that several thousand people worked on the Linux kernel, in fact, after 20 years of its existence, less than 600 names are listed in the CREDITS version of Linux 3.10. In just 3 years, almost twice as many people took part in OpenStack, and this is in one development cycle.

    As for Android, it was not originally a project with ordinary participants, being launched inside Google.

    All this leads me to the conclusion regarding a comparison with Hadoop. Matt calls Hadoop an exception, but in reality Hadoop just might be a “canary in the mine” - a harbinger of the commoditization that we have seen in the IT industry that applies not only to hardware, but to such large projects. Perhaps the days when one company can suddenly break into a project and get everything in the past.

    Doing what is best

    I talked a lot about whether Red Hat can take on the role of the “leader” of OpenStack, but only time will give an answer to this question. Instead, the main question to ask is - would it be good?

    Obviously, that would be good for Red Hat. Matt is right in saying that OpenStack has the best chances (at least for now) in the private cloud industry, where we do not compete with Amazon, and, of course, Red Hat is the vendor that would benefit most in this market . And, of course, it would be nice for OpenStack to have such a propagandist with great potential who would bring her to the club of participating companies.

    But this is not control over the direction of development of the project. Is OpenStack development chaotic? Of course. Is every idea good? Of course not. But great ideas usually do not appear in the streamlined process of a “planned trip.” If they arise, then from such chaos, since this chaos is fed by the needs of real customers who need real solutions right now. If one vendor tried to influence the development of the project in the right direction, in his opinion, we would get a project that would be good enough in one thing - what one vendor wanted. But someone else’s use cases could easily be ignored and even ordered to live long.

    OpenStack is just interesting because how many companies, which under normal circumstances are competitors, work together to make OpenStack the best it can. How many of them would have left the project if they had to comply with Red Hat's desire to create a product tailored to the needs of Red Hat?

    When the question of “leadership” arose for the last time, our CEO Adrian Ionel noted that while leadership is important, it is not as important as momentum. Matt says projects that have been boosted can fail, citing OS / 2 and OpenOffice as examples. For OS / 2, it was not an open source project; two companies (Microsoft and IBM) stood behind him, one of which was also a competitor. What about OpenOffice? Yes, it’s true that having bought Sun, Oracle (also the only vendor) did not want to continue to carry this burden. Therefore, she transferred it to the Apache Foundation, which announced in October that the number of downloads of its software version exceeded 75 million in less than a year and a half.

    It is incredible what an impulse of development can do.

    Responding to complaints

    The fact that Red Hat is a well-deserved leader in OpenStack cannot be denied. But we have said this before and will say it again: the OpenStack project does not need a single vendor dictating conditions, a well-wisher, etc.

    Therefore, it is undoubtedly cool to scold OpenStack and say that “something must be done” in order to change the project. But the fact is that OpenStack will go through this “difficult teenage” period, when it is still clumsy and when the elders think that “someone should take the boy’s hand”. Can the current process be improved? Of course. But the community can do this on their own, without the "dad" (or "mom") telling him to "tidy up the room."

    Original article in English .

    Also popular now: