Questions worth asking Oracle at Percona Live …

Oracle will be at Percona Live. This is a great news. Oracle was totally absent at last year edition. Not taking part to business event organized by competitors is a frequent Oracle Corporation behavior. Percona Live is a business event own by Percona (one of Oracle competitor) and some other Oracle competitors like SkySQL/MariaDB will also take part to the event. This is really nice that Oracle make this exception this year. MySQL is an open source ecosystem and having everyone at a common even is a very positive thing (even if the event is own by one of the parties and not yet a truly community even).

Gathering so many MySQL talents at the same place is exceptional. After FOSDEM 2013 event where all components of MySQL were gathered in the same room I realized how great it was to have MariaDB / Percona / Oracle peoples at the same Place. This triggers nice discussions as we all know each others for a long time and we share some common interests.

Last year Oracle was totally absent of Percona live. Oracle preferred to send all their new announcement during the conference through blogging. The community was expecting the same process this year ;-) Pushing key technical informations without interacting directly with the community was not what an open source community expect. This is great to have people live to interact with the MySQL community. MySQL is a turbulent community were discussion is a fundamental process to move things in the right direction. We are all fans of technology and having everyone at the same place will for sure trigger nice discussions.

Oracle is sending very nice ambassadors :

Tomas Ullin managing MySQL engineering at Oracle has successfully be turning MySQL into an “Enterprise Class Database”. This is great and we can definitely thank him for that. I personally was not expecting that Oracle would push MySQL in area like scalability on 64 cores, crash safe replication, enhanced security features … which turn MySQL into a product that in many cases competes very efficiently against Oracle flagship product Oracle 11g (with a much lower cost of ownership :-) ). Oracle has not yet acknowledge this fact and its marketing still position MySQL as the database for the web. For many users MySQL now is definitely much more. Most of the new support paying customers are in the enterprise market. This is a fantastic achievement.

Dimitry Kravtchuk whose benchmarking expertise has given many people an invaluable insight into the MySQL Server behavior. I consider him as the Sherlock Holmes of MySQL with all the astounding investigation he conducted on the server behavior. Most of these investigations have shed light on very peculiar behaviors and lead to the correct interpretation and trigger big progress in the code.

Luis Soares takes care of MySQL replication. Replication is with is with scalability the second most important area of MySQL. This is an area of great innovation.

All these people have a long history of commitment with MySQL as an open source product before it belonged to SUN or Oracle. They really know what an open source ecosystem is. We will have here a real opportunity to talk technical subjects and confront approach to various problems.

But beside being a product where technical excellence matter MySQL is an ecosystem with a lively community. For me leading the community does not just mean leading by having technical excellence. This is necessary but far from enough. In an open source community there are fundamental questions that cannot be avoided. Beside being a product where technical excellence matter governance and fairness is key. We all know that open source is build on values and principles based on sharing and fairness.  Like most of you I prefer technology to politics but sometime politics is essential no to be lead by small nice steps to a place where you were not initially planing to go.

Thank to Oracle for the technical excellence and the investment made and to come.

But I think of a few non technical questions that really matter for the MySQL community (Support providers, Storage engine developers, OEM customers …) and that need to be raised. These questions are fundamental for many peoples to have a clear idea of what kind of governance Oracle envision for MySQL future.

Do Oracle believe that the current MySQL development process is adapted to an open source product with a vibrant community and many active contributors ?
Oracle approach is based on no communication / no sharing with the community except the end result. This give the community no insight on the design choices which are not open to discussion. This can have a very negative impact on the ability of the MySQL community to improve the product and keep up with the code. In the past most of the key improvements, innovations, ideas have come from community members (individual, big users like Google, Facebook, Twitter …). MariaDB in this respect has kept the principle of 100% transparency. This event will be a good opportunity for the MariaDB team to show the long term benefit you get from this 100% transparent / 100% open source / open to fair competition approach.

What is the official process to guaranty that in the future the code will be released on launchpad in a timely and fair way ?
There has been multiple issues in the past regarding releasing the code late or regarding the revision history. Stewart smith(Percona) has been a good advocate on this subject. Percona depend heavily on the good will of Oracle to be able to produce their own version of the server. Any problem can greatly impact them and many others. MariaDB that fully own its own codebase as of MariaDB 10 is more immune to these issues. But MariaDB also depend on Oracle to back port features and guaranty that the MariaDB product is truly a MySQL dropin.

Do you plan to go further into providing new features as closed source extensions. Does the community has any guaranty that MySQL will under Oracle governance remain an open source product in the long term ?
The first closed source extension for MySQL(Thread pool, authentication plugin) appeared under Oracle governance in 2011. MySQL 5.6 has brought a new closed source features with the audit plugin. Adding closed source features lead to lock-in situation. Most of the closed source feature have been redeveloped by MariaDB as open source features. Redeveloping these features is an unnecessary burden for the community. Some people even consider that in this respect you do not comply with the promise made to the European commission to keep MySQL open source.

Do you plan to have a process to keep the bugs database open and to release test cases ?
There has been recently a lots of complains by the community regarding this subject. Keeping an open bug database  is vital to guaranty that third party or linux distribution can validate non regression of their version. Valeri Kravtchuk (Perconna) has with a deep knowledge of the bug process put some light on all these issues. Do we have statistics on the amount of bug entered in the public database that are turn into private ? It seems that the proportion is increasing over time. It would be great to clarify on this point and to have public numbers and justifications. Closing the bug database even partially is a long term threat to the MySQL product. I am not speaking of the second bug database used by paying customer. I imagine that in the end both the community and people running the Enterprise should hit the same bug that should end up in both bug databases. I am of course not speaking neither of bugs related to closed source features which are the sole responsibility of Oracle and where the community cannot help ;-)

Will oracle extend the period regarding the commitments they made to European Commission regarding MySQL to get clearance when buying SUN ( MySQL included ) ?
In my personal opinion this is necessary to keep MySQL community trust in Oracle governance of MySQL. Oracle commitment made to EC for 5 years will reach an end at the end of 2014 and this is raising worries for many members of the MySQL community (customers, OEM vendors, third parties developing storage engines…). To be honest rereading these commitment and putting a negative operator in front of each of them is totally and absolutely frightening for MySQL future :
http://www.oracle.com/us/corporate/press/042364
Oracle should give guaranties to the community. They have a big responsibility in this respect.

So having Oracle at Percona Live is a really nice great news. Oracle people will be really welcome by the MySQL community. This will be a good opportunity to talk about technology and confront our vision of the future.
But it would also be nice if Oracle could bring some answers regarding their fair governance of MySQL.
Oracle has a huge responsibility in this respect and the MySQL community is very attentive on this subject.
Technical excellence of one of the actor of the MySQL ecosystem is one element among others regarding the successful future of MySQL as an open source product.

———————

Based on Matt Keep valuable twit on : “For those who engage in swearing & general profanity at work, a handy guide of suggested alternative phrases” I have tried to expunge my initial prose from any bad words. :-)

16 comments to Questions worth asking Oracle at Percona Live …

  • “Oracle approach is based on no communication / no sharing with the community except the end result.”

    That’s not true at all. Through their virtual developer days in March (this year and last year) they help share the roadmap, and they do lots of roadmap stuff during MySQL Connect, which is not only happening this year, it’s been extended by a day to include more tutorials.

    You say “In the past most of the key improvements, innovations, ideas have come from community members” and that’s true – of the past. When Oracle didn’t own MySQL. Oracle takes some stuff from community, and does some of its own engineering. In this regard it’s better than either MySQL AB or Sun was. I’m not sure what the complaint is, it sounds like a compliment to Oracle.

    Do you mean “Please open the bugs database again?” Because that IS important, but completely different from the complaint you’re making (that the roadmap isn’t shared) which isn’t valid.

    • Serge

      I agree that Oracle communicates a lot on the new features once they are complete. In that phase they can even enter into very detailed discussions. This is really nice and I thank them for that.

      I was more talking about continuous communication process during the architectural choice, the design phase, the implementation.
      A good example is the innoDB full text search. This development has been going on for 5 years without any communication. Communication started once it was finished ;-) I confess that on innoDB not so many people could have helped. But maybe a discussion could have help and I am sure the issue related to too much rights necessary to be granted to access index metadata ( http://goo.gl/rWceY ) would have emerged during a techies discussion and the right solution might have been proposed.
      “The requirement that a user have SUPER and now PROCESS privileges in order simply to see information about full text indexes on their own tables is very disappointing.”

      MariaDB in this respect is acting totally transparently. Anyone is welcome on their irc channel with a continuous communication/discussion.

      5.7 might prove that you are right. We will we see either fully baked feature popping out or the opening of discussion with the community.

      I do acknowledge the fact that Oracle has done a great job integrating the past contributions of the MySQL community. I recognize that to Oracle and thank them for that. In that respect we can effectively say that Oracle has been better than what MySQL AB was doing in the past.
      My worries are more for the future. If Oracle goes on the community will progressively lose grip of the code. This will make it extremely difficult for the community to contribute. It will also make it difficult for fork providers to be able to add their own features. This will impact the ability to have a fair competition on the support. Fair competition is what users want.

  • “But MariaDB also depend on Oracle to back port features and guaranty that the MariaDB product is truly a MySQL dropin.”

    I don’t see how it’s Oracle’s responsibility to make sure MariaDB is a MySQL drop-in replacement.

    • Serge

      I agree that Oracle is totally free to do what they want with MySQL which they own. I am just concerned about the fact that they should release the code revisions to the community in a fair and timely way. If they do not this is a very bad hit to the MySQL community.
      If Oracle was to go that way users should better understand what is happening and how it will impact their long term possibility to chose.
      This might trigger move from user to distribution were there is this guaranty. This is something we already see today. Oracle attitude has raised suspicion regarding Oracle intentions. There are a lot of MySQL users moving to other version for confidence reasons.
      This process of making life more difficult for fork providers is not perceived by end user but it is important that they know about it if it happens.
      Users do not want lockin. They want fair competition on support. Having multiple version is key in this respect. This is the only long term guaranty of low cost of ownership which in many cases is the first reason to chose MySQL

  • “Some people even consider that in this respect you do not comply with the promise made to the European commission to keep MySQL open source.”

    Threatening Oracle isn’t going to help here. They put in a ton of new features, and you’re complaining about one or 2 that are closed. It’s not like GTIDs are closed, or that Oracle is closing the source of everything. MySQL Enterprise Monitor has been closed from the beginning of its lifespan, under MySQL AB. You’re choosing specific things to make Oracle look bad, on purpose, and ignoring the fact that Oracle is mostly continuing the path that MySQL and Sun have set. You obviously have an agenda.

    • Serge

      Sheeri I am not threatening Oracle. Oracle has made the choice to commit to the European Commission.
      Of course deciding if adding closed source feature is in compliance is a very arguable point. I was just pointing a worry on the proportion and category of closed source features in the future.
      To be honest I think that this addition of closed source feature has served the 100% version like MariaDB. MariaDB has produced the same features in a 100% open source philosophy and it seems that it is something very appreciated by customers. Being 100% open source leads to much confidence as their is no lockin involved and the choice for support remains open.
      Percona support MariaDB. They have even back ported the MariaDB thread pool feature which is closed source for Oracle and reserved to paying customers. This is great. This is a recognition for excellence . Open source is based on meritocracy not on lockin. In the same way MariaDB can decide to implement bug fixes produced by Percona.
      I fully agree that currently the closed source features represent a very tiny part of the code produced by Oracle. I recognize that they put a lot of effort to produce fantastic features.
      But most of users interpret these closed source features as a tendency that illustrates the future of Oracle MySQL. I have no agenda and I am just expressing a concern.

  • Read up on facts, don’t believe the FUD. Release test cases are available – to distributions. It’s under a different license. It’s surprising how many people don’t bother actually looking this stuff up.

    • Serge

      Sheeri could you clarify ? What do you mean under a different license ?
      Do not take me wrong I am not spreading FUD, just expressing my current understanding of the issue. It seems I am not alone to have this understanding. Percona Live can be a good opportunity for Oracle to clarify on this point.

      This is what the wikipedia says about it (MySQL article) :
      In August 2012, TechCrunch’s Alex Williams reported that Oracle was holding back MySQL Server test cases, a move that he concluded indicated that Oracle is attempting to kill the product.[74] Percona also reported that Oracle is no longer synchronizing their changes with the public source repositories.[75] Widenius called this a breach of the agreement that Oracle entered into with the EU as a condition of their acquisition of Sun.[76]
      Aren’t these facts ?

      • Those are indeed all facts. It’s a fact that a journalist said someone was holding back test cases. Percona reported that test cases are separate from the code. That’s a fact too. But separation and one person saying “they’re holding back” doesn’t make it true. And of course Monty’s saying it’s a breach of the agreement. That’s a fact too.

        I can say MariaDB doesn’t work, but that doesn’t make it true. Even though it’s true that I said it. (Except I didn’t say that :D )

        • Serge

          I perfectly accept the semantic fact that if it is true that you have said something does not mean that what you have said is true. I was just mentioning this as something that need to be clarified. The community and users are very sensitive on these issues.
          Users perfectly understand that undermining the capability of third party to develop and support MySQL will create a long term dependency to Oracle. Customer believe it could put them in a lockin position in their relation to Oracle. They have not chosen open source to face this risk.
          You lost me with you MariDB example ;-) but what is also true is that if you face a bug with MariaDB you can enter it in a 100% public database, you can openly talk about it on MariDB irc and the fix if it is ever produced will be available with its associated test case to everyone.

          you said :Release test cases are available – to distributions. It’s under a different license. ”
          What do you mean ? Does “distributions” include Percona and MariaDB ?

  • Why would Oracle kill MySQL in 2014?

    When Oracle bought Innobase in 2006, Percona was just barely in existence as a company. There was no Percona patchset, there was no MariaDB. If Oracle wanted to kill MySQL, they would have killed InnoDB in 2006.

    Why would Oracle kill MySQL now, when it makes money for them? If they’re killing MySQL, why do they have tons of trainings for it, and travel to 6 continents to do so? They do about 15-20 different cities *each month* (I know because I research it for the podcast). Why would Oracle spend so much money on a product they’re killing?

    They’re not killing MySQL, that’s why. They have no intention of it, it’s a product that makes money for them. Killing MySQL now, when there are viable forks, means that they lose all their business. Larry Ellison may not understand open source, but he wouldn’t kill a product that’s making money for him.

  • Serge

    Sheeri I am an optimistic guy so I am sure that there is a very nice future for MySQL / MariaDB / Percona after the end of 2014. I agree that oracle could have killed innoDB and then MySQL many times. I agree that they currently invest a lot in promoting MySQL and they make money with it. This for sure benefit globally to the whole ecosystem. We have a lot of customer that are big fan of Oracle event : webinar, training… I am too as I am for the excellent Percona events.

    But this is the fact that at the end of 2014 Oracle will be freed from its commitment to the European Commission. This potentially means a lot. I am sure nothing wrong will never happen. But potentially Oracle has the right to close down documentation, not to renew third party agreement for storage engines, to make support mandatory for commercial license… I am of course not saying that they will dot it. The community, third parties, customer, users all the one that are investing on MySQL simply want things to be clear. This is necessary to keep confidence. If Oracle has no intent to break any of the promises they made and I am sure they do not then they just have to say it. This is not an obligation but it will just be fair for people investing on the MySQL technology.

    I believe this worry comes mainly from the following perception. Currently MySQL has remained totally independent inside Oracle. After the 2014 deadline is gone they is a fear that the Oracle “Classic” management will take over and align MySQL with other business practices.
    Oracle management can decide to arbitrate base on more pragmatic, business oriented rationale. The fact that MySQL has become an strong internal competitor to Oracle 11g they might trigger changes like changes on pricing. These are speculations and user would just prefer Oracle to clarify on this subject and give some guaranty on future pricing. This would help customers to make their long term investment choices. Business is based on confidence.

    I personally have none of these worries. I am more worried about Oracle making life difficult for competition. They will then make even more money than what they are making today. But this will also once fair competition is gone create a big danger for customers. To avoid this to happen we all have to be very attentive on Oracle respect of the MySQL community. The point of this argument was mainly to raise the attention of users further than just around technical excellence.

    The argument “we improve the product a lot so we deserve to lead the community” is a little bit short. You have to play fair and I sincerely hope Oracle will.

    • According to you, Oracle’s not even following the EU agreement. So why ask them to promise more time, if you don’t even think they’re following it when they promise it? You won’t trust their agreement anyway, and they can’t promise forever, and you’ll be suspicious about that too.

      • Serge

        I did not say that Oracle was not following the EU agreement. I was just highlighting the fact that many people wanted to be reassured about Oracle good will after the 2014 deadline. I trust Oracle as most members of the community do when they commit to something. And that is exactly why it would be nice if they could commit to respect their previous engagement for a longer period. This uncertainty after 2014 is deserving everyone in the MySQL community. We all know that adoption is based on confidence. Oracle positive position on this subject would have a positive effect for all the MySQL ecosystem.

  • “The fact that MySQL has become an strong internal competitor to Oracle 11g they might trigger changes like changes on pricing.”

    Can you show me where that is actually a fact? Oracle’s db and MySQL don’t really compete. There are plenty of people who have switched from one to the other, but I haven’t seen anything that shows that Oracle’s specifically losing money to MySQL (as opposed to sqlite, mongodb, cassandra, etc).

    • Serge

      I agree that Oracle always present Oracle as the database for the web. They of course never present it as a competitor of Oracle 11g. I do not expect that from them ;-)
      But there are more and more use case were both can fit. So let us call MySQL a competitor of Oracle in these areas.
      My feedback is more based on what I face every day at customer sites. In the US it might be slightly different as you have almost all the big web properties and it skew the picture.
      In Europe we do not have that and our market is mainly normal companies, big Oracle shop included. These company are looking for cost reductions in IT. A few years ago they were very reluctant to use MySQL and they just consider us as kids. Now they come by themselves to ask for advices on cost reduction. In that case of course we never position MySQL as replacement of Oracle. This is not true and most migration project are complex, costly and uncertain.

      Many of the big account we met are operating both Oracle and MySQL. In the past the one in charge of operations were very reluctant to operate MySQL. They now are more familiar with MySQL and they have realized that the the mean time between failure for MySQL is lower that with Oracle, that the resolution time for bugs is shorter. I was personally quite surprise when I heard that from customer but they had numbers to prove it and I now consider it as a fact.
      Now that MySQL can scale better on bigger hardware, offer a crash safe replication, has more security feature and can execute a much wider range of queries many users are considering extension of its use cases.

      If you look at SkySQL references many of them are in the enterprise market. http://www.skysql.com/customers. I cannot speak for others but I do not think it is much different. MySQL sales force inside Oracle might even be doing better addressing these big account as they have a better knowledge and access to them. Of course on a business perspective the database market is still a growing market so the impact on other sales is marginal. Having MySQL replace Oracle in a wider area just makes it grow at slower rate, not a decrease I agree.

Leave a Reply

  

  

  


4 × seven =

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong> <pre lang="" line="" escaped="" highlight="">