Colin Charles Agenda

I have never been to a DrupalCon before so my first was DrupalCon Portland 2013 (with some 3,500+ attendees). My first DrupalCon happened to also be one that I spoke at, and I hope to return to Austin in 2014 (added bonus: I’ve never been to Texas before).

SkySQL had decided to get a booth at DrupalCon since I was speaking and I have to say that the booth was very successful. You may ask why and the simple reasons are:

1. Everyone at DrupalCon was a user of MySQL. Once they hit scale, they may need professional services or even just switch to MariaDB. This is a captive audience.
2. Everyone running Drupal at scale with many modules and nodes generally faced issues with database slowdowns. Hello remote DBA, consulting, or 24/7 support.
3. We never had a bit of spare time to relax at the booth because people came over in droves to find out how MariaDB could help them, how SkySQL might make sense for them, or just poke us with general Q&A about MySQL. This happened even thru lunch, which we consumed at the booth. Let me add that lunch was great: Mexican, BBQ, etc.
4. My talk was at an odd slot: competing with beer at 5pm. The room was full. There were many interesting questions. People were clearly interested in improving their database for Drupal. Lots of positive/immediate feedback on Twitter.
5. Affable characters manned the booth. SkySQL sent Rod Allen and Marc Sherwood, and I enjoyed managing the booth with them. Conversation was great. The day I had to give a talk, I spoke all day at the booth then gave my talk!

It was also great to catch up with Dave Stokes, MySQL Community Manager, as he had a booth too. I hope we sent him some traffic! He’s also the writer of the Keeping your Drupal from Drooping series, so read part 1 & part 2. I reckon that I’ll blog more about Drupal & MariaDB soon as well.

I met some old friends., some of whom are working for the Whitehouse now. I also met some new friends who are all very gung ho about MariaDB at scale for Drupal hosting.

Here’s to happy Drupal & MariaDB/MySQL users!

Lately more people ask me for comparisons between Percona Server & MariaDB. There isn’t a definitive blow-by-blow feature comparison yet, but it’ll come soon.

All that said, its great to see new features from MariaDB make it into Percona Server. The features that I’ve managed to track: group commit for the binary log, threadpool and atomic write support for Fusion-io devices.

It started with the group commit for the binary log feature in Percona Server 5.5.18-23.0 (docs). 19 December 2011 is a long time ago, and before we made MariaDB 5.3 or 5.5 GA as well. 

The next feature that made it into Percona Server was the threadpool (different implementation compared to MySQL Enterprise). We had a threadpool since MariaDB 5.1 (libevent based), but improved it drastically in MariaDB 5.5. It made its debut in Percona Server 5.5.29-30.0 (26 February 2013 – so about 10 months after we included the feature), but was further improved in 5.5.30-30.2 to include priority connection scheduling.

The latest feature to be included? Fusion-io DirectFS atomic write support. It made its debut in Percona Server 5.5.31-30.3 (24 May 2013 – so about a month since we included the feature). I like this because Laurynas Biveinis provided some feedback on the maria-developers list.

Great opensource at work here. I look forward to more features making their way back & forth between the branches.

I just got back from a vacation to see articles about Business Source Licensing. I’ve divided my thoughts into four parts here: Opensource and its merits, Is unpaid opensource usage bad?, MariaDB’s “Problem”, Business Source Licensing. If you haven’t read them yet, here’s some mandatory reading:

1. Open source: Its true cost and where it’s going awry by Monty Widenius
2. MySQL Co-Founder Wants You To Pay Up For Open Source

There is much abuzz on Twitter as well. From the likes of Mike Olson (who is right, MariaDB may have issues that are different to other OSS products – no two OSS projects/products are alike), to a lengthy conversation between Jim Jagielski & Matt Asay, as well as another conversation spurred by Matt Asay.

Now for some of my own commentary.

Opensource and its merits

Companies have been heavily using opensource and the reason they like this is because it is open. They don’t pay for licenses like proprietary software. They use opensource because they don’t have to pay for support, services, or anything around it. Countries have pro-opensource policies so that they can empower local citizens and further strengthen their sovereignty. This is what makes opensource popular: the fact that the software comes to you with many freedoms.

Is this bad for companies building businesses around opensource software products? Well, kind of. It means you have to provide real value before someone decides to pay you. And if for some reason you price yourself out of the market, companies choose to hire resources internally. This is the beauty of opensource. Many companies I know have started to use RHEL licenses from Red Hat; once they decide they see less value from the updates or the knowledgebase, they switch to CentOS at their next cycle. No problem there.

Is unpaid opensource usage bad?

I’m going to say that I disagree with Monty and think that he is wrong here:

“The more people are using it and, in these cases, abusing the whole idea of open source by not paying back either with development or money to help projects, it is actually destroying open source.”

I really don’t think opensource is destroyed by having many users and lacking corporate sponsors. This is the way of opensource and has been for a long time. Apple makes use of CUPS to ensure printing works – they did so long before they hired their main developer. We all benefitted from Samba which is how we talk to Windows printers/shares/etc. which had no real commercial company around it (Linuxcare, then IBM, then other providers funded the work). LibreOffice has always existed with lots of work by various distributors of OpenOffice.org (via the ooo-build system), which is why the project took off so fast.

MariaDB’s “problem”

When there is commercial need for opensource, the corporate sponsors will arise. It takes a long time to get to a stage where you are going to get profitable in an opensource services or infrastructure company. Red Hat didn’t get to a billion dollars overnight. Neither did MySQL.

I will not comment on the financials of Monty Program, SkySQL or how tough it has been to bootstrap the MariaDB project because I clearly am privy to information there. I am particularly proud of how we’ve done a relatively great job at getting MariaDB users and distribution, all on a bootstrap marketing/PR budget with no professional help :-) However, I will reminisce another day.

Simply put: if Oracle stopped producing opensource MySQL or decided that they would shut it down, there would be immediate need for MariaDB and the corporate sponsors would come in throngs. The truth is that Oracle continues to produce MySQL as an opensource product. It may not be a full opensource project (internal trees, delayed public pushes, private bugs database, internal mailing lists, etc.) that follows “the architecture of participation”, but it is still an opensource product. This is what has enabled people to take MySQL and extend it further. Look at the Facebook 5.6 tree, or the Twitter 5.5 tree.

There is talk about the dual-licenses that MySQL chose to use. I remember a time when the connectors were LGPL. They were then relicensed as GPL. They still are. But I think we effectively nipped this with the: MariaDB LGPL Java client, MariaDB C Client Library, and the BSD drizzle stuff.

Business Source Licensing

Now for the bits on business source:

“The whole idea with business source is actually very trivial. It is a commercial licence that is time-based and which will become open source after a given time, usually three years. But you can get access to all the source. You can use it in any way but the source has a comment that says you can use it freely except in these circumstances when you have to pay,” Widenius said.

“You’re forcing a small part of your user base to pay for the restrictions, which can be if you’re making money from [the software], if you have more than 100 employees, or you’re a big company or something like that. So you’re forcing one portion of your users to pay. But because it’s time-based, everybody knows that you can still contribute to the project,” he said.

“Because you have the code, you know that if the vendor does something stupid, somebody else can give you the support for it. So you get all the benefits of open source except that a small portion of users has to pay. As long as you continue to develop the project, each version still gets a new timeline of three years.”

Hmm. I see many people commenting that MariaDB might become business source licensed. I am here to tell you that MariaDB is GPLv2 software. It will stay GPLv2 software.

Reading the definition of business source licensing, it is nothing like what Matt Asay portrays it to be:

“Business source is simply proprietary software released under a Microsoft-esque shared source license that magically becomes fully open source after a period of time.”

I’m sorry but the description above is pretty clear. This is nothing like Microsoft shared source. It is code that becomes licensed under an OSI-friendly license after a time-period; however everyone using the software gets the code. How does one enforce payments? I don’t know. What are the conditions requiring you to pay? I don’t know.

At this stage, I am open to thoughts on such a licensing model but I have no firm thoughts on this myself. The best description of how this works is given above by Monty.

Update: Sun 2 Jun 2013 17:33:53 MYT Monty has an update on business source licensing in a comment on Matt Asay’s column.

The Chromebooks have arrived in Malaysia: The World’s First Samsung 4G Chromebook. They come with WiFi and a WiMaX chip so you connect to YES4G (up to 20Mbps speed on this network). They retail for RM1,299 (USD$419) or RM988 (USD$318) with a 24-month contract that costs RM88 with 3.5GB of data transfer (exceed the quota and you get free data usage at a reduced speed).

In the USA, the Samsung Chromebook with 3G retails for USD$329 while the Samsung Chromebook WiFi retails for USD$249.

I’m not sure why YTL has jacked up the price so much. There is no way that a WiMaX chip costs so much more compared to the 3G chip. There is already a $80 premium in the USA to get 3G (something you can use on any network – Maxis, DiGi, Celcom, etc.). Why does the WiMaX chip cost $90 more than the 3G model (total for the privilege of WiMaX: USD$170 – RM527!). Keep in mind that this extra means you only are mobile where the YES4G network is available (which is not everywhere, compared to the 3G networks in Malaysia).

Next up: LTE is here. Speeds exceed 20Mbps easily. Maxis and Celcom are already providing devices, including portable WiFi hotspots (MiFi devices). You can also tether your devices easily.

To add: there is likely a new Samsung Chrombook in the works. If it can keep the same price point and get a lot more power, it could be very interesting. The current Chromebook went on sale at Amazon in October 2012 (so some 7+ months ago). A refresh is definitely around the corner, though that shouldn’t stop anyone from buying one (it certainly won’t stop me).

My verdict? Buy a Chromebook with WiFi only. WiFi is everywhere these days. I don’t think the premium for connecting to the YES network makes sense. Real problem? These devices aren’t sold outright in Malaysia in any official capacity which I see as a problem.

Update: With 100GB free “cloud storage”, and only 16GB local storage, one would presume that you would use the cloud a lot for things. Music and videos will have to be streamed. Can you live with 3.5GB of data transfer (this is uploads and downloads)? Will you ever really get to use your Google Drive to its potential?

I was always a Posterous user over a Tumblr user.

How did I use Posterous? I would email photos, take random quotes from websites, etc. and ensure it made it to my “outpost”. I had stopped using Posterous a little before the Twitter acquisition, and didn’t care much for it when it shutdown. I did take a backup of my stuff though, because it never was federated to my own blog or anything like that. I did like the fact that it would post links to Twitter automatically and would save my images to Flickr as a backup bucket for my photos. 

How did I use Tumblr? It claims I have 11,766 items, which you can say is a mirror of my tweets for a period of time. It also contained a mirror of my Posterous posts. In fact it shows that it might have stopped not long after the iPhone 4 came to Malaysia and I visited Phnom Penh for a barcamp. So it does bring back memories. I just found the mass post editor – last Tumblr entry was October 2010. Apparently I started using it in June 2007. It might have even fed my last.fm feed to it as I can see music being popped in during the early days. It also states my timezone was GMT+10, so definitely legit – I was living in Melbourne then.

I decided not to delete this archive (just to rename the subdomain with the -old tag – very handy). However the new blog will never be my primary blog due to Tumblr’s architecture.

So I guess that puts an end to the re-Tumblr experiment. I even had the bookmarklet installed in the browser raring to go.

WordPress has aside support. I don’t use it here but its almost similar to a service like Tumblr, no? I guess it goes back to owning your content – I’m a fan of keeping things here on a site that I control (note that I don’t use wordpress.com either).

That said, Posterous sold for a lot less to Twitter (I speculate). Tumblr sold for $1.1 billion to Yahoo!. Let’s hope all is well, failing which users will be a migrating. 

“The market is so small you have to think international” – Mattias Miksche, CEO, Stardoll (TWIST#343).

Sweden has like 9 million people. They don’t think they need to “go global”. They start thinking global. The local market is too small. It’s not like the USA.

I’ve said the same thing about Malaysia (and Singapore) before: why are you building locally if you’re a web-shop? Build, test, prove yourself, but make sure you’re global. 28 million people separated by different language needs will ensure you don’t have much larger than the population of Sweden when it comes to launching a product. 

So don’t think about going global after a year. Think about being global from day one.