Colin Charles Agenda

Just a few days ago, I presented at the MariaDB Roadshow in London, and I had a lot of fun. While I had canned slides, I did know the topic intimately well, so it was good to get further in-depth. In addition, we had these MasterMind sessions, basically the place to get one-on-one time with Anders/Luisa/or me, I noticed that pretty much everyone said they were buying services afterwards (which more or less must mean the event was rather successful from that standpoint!).

In addition to that, I was happy to see that from attendee feedback, I did have the highest averages – thank you!

So here’s to repeating this in Paris next week — Gestion des données pour les applications vitales — MariaDB Roadshow Paris. I look forward to seeing you there, and I know we are repeating the MasterMind sessions. To fine-tune it, try to bring as much information as you possibly can so our time can be extremely effective.

I’m happy to be back in London in May 2016, to talk at two events:

1. The London MySQL Meetup Group – Monday May 9 – 6.30 PM – 9.00 PM – options for High Availability in the ecosystem that one would consider today. This is a cut down from my Percona Live tutorial, which had about 88 registered attendees and about that amount showed up and asked questions even through the break and after. I had a lot of fun, and I expect I will have similar fun in London talking about this area that has changed a lot in recent times.
2. Data for the Enterprise — MariaDB Roadshow in London – Wednesday, May 11, 2016 from 9:30 AM to 2:00 PM – this should be a fun gathering, and you would think its just for MariaDB Corporation customers, but it isn’t – all are welcome and you should register!

Looking forward to seeing you at these events. And if you want to chat about MariaDB Server, MySQL or anything related in the opensource world, don’t hesitate to drop me a line via email or Twitter @bytebot.

• Planning for May MariaDB Corporation roadshows in London and Paris. The Paris edition is also one where Kentoku Shiba from Spiral Arms will make an appearance; we’re also likely to plan several meetings around there too.
• Planning the MariaDB Berlin Meetup. Getting the speakers from our side. Ensuring that other user groups know.
• Publishing mariadb-mysql-tutorials on Github. Consider this to be a growing repository of how to get stuff done in the various servers in the ecosystem.
• Attended/participated in the MariaDB Corporation company meeting, and the follow-up developer meeting.
• Finishing the decks for rootconf.in & #perconalive Santa Clara. And OSDC.de.
• Delivered keynote for rootconf.in. Had a panel. And also delivered a tutorial.
• Delivered a tutorial for Percona Live – 88 registered, about that showed up, if not a little more. It was a full room, and there was no chance for doing practicals. Talked for 3 hours with a 10 minute break in-between (where I was still just talking…). Delivered a talk about MariaDB Server 10.1 and a sneak peak of 10.2. Lots of meetings, and lots of good conversation here.
• Help where I need to with the upcoming Gartner work.
• MariaDB Corporation had a website redesign; are the download resources going to the correct place? Check with marketing.
• Filed: MDEV-9882: Lack of config defaults in 10.1.13 because I think the out-of-the-box user experience needs to be improved. MDEV-9889
  create ubuntu xenial 16.04 repository. mysql#81051.
• Admin: filed 3 expense reports
• Customer/prospect email bumped up to us count: 7. Alliances related emails/chats count: 2. Management level emails: 9
• Wrote posts for the MariaDB Corporation blog – AliSQL and some features that have made it into MariaDB Server, Automated Server Management with opensource tools, [Modelling databases in a web browser], MariaDB Server growth bolstered by OpenStack User Survey. Also now all this is available in a Github repository: mariadb.com-blog.
• Submit idea to MariaDB Foundation about individual supports and having mariadb.org aliases, something like what the Linux Foundation Individual Supporter program does. After all, we are seeing a List of Donors appear on the MariaDB Foundation site now.
• Blogged:
  1. MariaDB Berlin Meetup
  2. (TWEET) SUMMARY OF PERCONA LIVE 2015
  3. FOSDEM 2016 NOTES
  4. Trip Report: Bulgarian Web Summit
  5. MAJOR POST-GA FEATURES IN THE 5.7 RELEASE!
  6. MARIADB BERLIN MEETUP NOTES & SLIDES

We had the first MariaDB Berlin Meetup on Tuesday 12.04.2016 at the Wikimedia Berlin offices at 7pm. More or less there were over 54 people that attended the event, a mix of MariaDB Corporation employees and community members. We competed with the entertainment at the AWS Summit Berlin which was apparently about 400m away! Food and drink were enjoyed by all, and most importantly there were many, many lightning talks (minimum 5 minutes, maximum 10 minutes – most were about 6-7 minutes long).

The bonus of all of this? Lots and lots of slides for you to see. Grab them from the Google Drive folder MariaDB Berlin meetup April 2016.

1. Monty talked about improving the speed of connections to MariaDB Server, some work he’s just pushed fairly recently to the 10.2 tree.
2. Dipti spoke about MariaDB ColumnStore and it is now clear we’ll see some source/binary drop by the end of May 2016.
3. Sergei Petrunia and Vicentiu Ciorbaru spoke about the upcoming window functions that MariaDB Server 10.2.0 will see (yes, the alpha should be out real soon now).
4. Jan spoke about InnoDB in 10.2.
5. Lixun Peng spoke about a fairly interesting feature, the idea to flashback via mysqlbinlog and how you can have a “Time Machine”. I can’t wait for flashback/time machine to appear in 10.2. The demo for this is extremely good.
6. Kolbe spoke about data at rest encryption using the MariaDB Amazon AWS KMS plugin.
7. Sanja and Georg went up together to speak about 10.2 protocol enhancements as well as what you’ll see in Connector/C 3.0.
8. Wlad gave us a good rundown on authenticating with GSSAPI, something you will notice is also available in MariaDB Server 10.1’s later releases.
9. Johan Wikman gave us an introduction to MariaDB MaxScale, which started off the talks on MaxScale.
10. Markus talked about the readwritesplit plugin.
11. Massimiliano went into the Binlog server.
12. Martin didn’t use slides but gave us an amazing talk titled “Rival concepts of SQL Proxy”; it was very well given and I’ve encouraged him to write a blog post about it.
13. Community member Ben Kochie, an SRE at SoundCloud gave us a quick talk on Monitoring MySQL with Prometheus and how much they depend on the PERFORMANCE_SCHEMA.
14. Diego Dupin spoke a little about the MariaDB Java Connector, and the idea was to do a demo but the projector via HDMI seemed to be a bit wonky (this was also true of using my Mac; the VGA output however worked fine). So it was just a quick talk without any deck.

We ended with a quick Q&A session with Monty dominating it. Lots of interesting questions around why the name Maria, licensing thoughts, ensuring all the software we have are in distributions, etc. Some ended up going for pizza while others ended up in a hotel bar at the Crowne Plaza Potsdamer Platz — and the chatter went on till at least 11pm.

Thanks again to Georg Richter who found us the venue and also did a lot of the legwork with Wikimedia Foundation.

Interesting developments in the MySQL world – it can now be used as a document store and you can query the database using JavaScript instead of SQL (via the MySQL Shell). There is also a new X Plugin (see: mysql-5.7.12/rapid/) (which now makes use of protocol buffers (see: mysql-5.7.12/extra/protobuf/)). I will agree, this is more than just a maintenance release.

Do get started playing with MySQL Shell. If you’re using the yum repository, remember to ensure you have enabled the mysql-tools-preview in /etc/yum.repos.d/mysql-community.repo. And don’t forget to load the X Plugin in the server! I can’t wait for the rest of the blog posts in the series, and today just took a cursory look at all of this — kudos Team MySQL @ Oracle.

However, I’m concerned that the GA is getting what you would think of as more than just a maintenance release. We saw 5.7.11 get at rest data encryption for InnoDB, and now 5.7.12 getting even more changes. This is going to for example, ship in the next Ubuntu LTS, Xenial Xerus. Today it has 5.7.11, but presumably after release it will be upgrade to 5.7.12. I am not a huge fan of surprises in LTS releases (predictability over 5 years is a nice thing; this probably explains why I still have a 5.0.95 server running), but I guess this small band-aid is what we need to ensure this doesn’t happen going forward?

As for the other question I’ve seen via email from several folk so far: will MariaDB Server support this? I don’t see why not in the future, so why not file a Jira?

Config Management Camp is becoming a great addition to the post-FOSDEM crowd. Short train ride away from Brussels to Ghent, lots of good content for an overnighter in a hotel (2d/1n). It routinely sells out (the cost is unbeatable, free ;)), so go early. Some quick notes:

Jonathan Boulle – CoreOS – etcd @baronboulle

• /etc but distributed with a clustered key-value store with GET/SET operations
• Support SkyDNS, Kubernetes, etc.
• CoreOS mission is to “secure the internet” and they like to provide automatic updates. Updates = rebooting. They didn’t like ZooKeeper so much.
• etcd: strong consistency guarantees, simple/fast HTTP API, OSS
• Uses the Raft consensus algorithm (replicated log to model a state machine; append only log). Concepts: leaders, elections, terms
• Written in Go, statically linked – etcd daemon and etcdctl cli

Service discovery using SmartStack – John Billings – billings@yelp.com

• http://nerds.airbnb.com/smartstack-service-discovery-cloud/
• Load balancing was done via something called create-lb-config and it was a lot of manual stuff which was deployed via puppet. Very labour intensive.
• They created something called PaasTa as a Paas (https://github.com/Yelp/paasta). And they found out about AirBnB’s SmartStack
• HAProxy allows you to re-dispatch on connection failures and easy to add connection logging as well. Each HAProxy instance also has a status page
• http://engineeringblog.yelp.com/2015/04/true-zero-downtime-haproxy-reloads.html
• Originally Nerve & Synapse made a Zookeper connection per service so it would make Zookeeper slow. Modify Nerve/Synapse to multiplex all operations over a single Zookeeper connection
• Use Zookeeper + AWS without issue for service discovery
• etcd would probably also work and supported by SmartStack but they already know Zookeper. SmartStack is conceptually similar to consul/consul-template

Platform Orchestration with Kubernetes and Docker

• Julian Strobl – julian@endocode.com – experts in linux/OSS dev, trainers, consultants located in berlin
• They have a 4-node Raspberry Pi cluster that one of them made
• There used to be a Master/Minion relationship, but now its a Master/Worker relationship when you are dealing in the Kubernetes world
• https://github.com/saturnism/gcp-live-k8s-visualizer

How CoreOS is built, modified and updated – Brian “redbeard” Harrington, Principal Architect, CoreOS

• @brianredbeard / brianredbeard.com
• repo sync – allows for managing many git repos at once
• cros sdk is used to build Chromium OS, but its truly an SDK (lots of bash scripts that do things the way you want them)
• OMAHA protocol created by Google to handle updates to desktop applications (eg. ChromeOS, Browser, etc.). There are OSS bindings provided by CoreOS in Go.