It have been a long time since my last post, it’s a really busy year at work with enjoyable things to do!, but of course nothing is an excuse to stop sharing updates.
This year I have been working in several areas which includes software development, community strategies and spreading the word at several conferences. I think it have been the perfect mix of things to learn and improve. All of this consumes a lot of time, hopefully my wife and kids are very supportive on all of this, the balance is very important to keep everyone healthy.
Below you will find a short update for:
- Fluent Bit
- Conferences and Fluentd
- Monkey & Duda I/O
Fluent Bit
As I always say, data collection is hard and when dealing with small devices there are not too many good options around. As I introduced previously, I continue working on the Fluent Bit project, which is a data collection tool for Embedded Linux. In the last months many improvements and additions have been made such as co-routines support, abstracted network I/O layer, fully support for Fluentd and Treasure Data cloud service, library mode (C, C++), packaging, etc.
The project is growing and it have reach a stability point where it can be used safety, as of today a general overview of the architecture is the following:
Now I am focusing into add fully internal routing capabilities and the option to use file system buffers to provide reliable mechanisms when network operations are not available for cases where the system faces some power outage.
This is a really interesting project, if it got your attention you are welcome to join us!.
Conferences and Fluentd
Spread the word about what do we do at Treasure Data is very important, even more why we do it. Collecting data is pretty hard when it comes from multiples sources, performance is a consistent pain in the processing and that’s the reason why we need specialized tools to solve that problem: Fluentd.
During this year, I have been lucky enough to receive enough trust from the organization of several Conferences where I was given the opportunity to present different topics about Data collection, which include to describe generic problems in production environments, workarounds and possible solutions like Fluent as an enterprise open source to address this problematic.
After apply to several conferences as a speaker and with the valuable support of Treasure Data, I was able to represent our mission at LinuxCon/CloudOpen Tokyo, FISL16 Brazil, LinuxCon/CloudOpen Dublin, EmbeddedLinux Conf Dublin, LinuxCon/CloudOpen Seattle and StartechConf Chile.
Learning is a continuous cycle of every life experience and I can tell you that attending conferences have been very challenging for me. As most of you know my native language is Spanish so speaking in English in public required to pay attention to improve how to express ideas, concepts and use the right vocabulary, I made many mistakes, my English is far to be perfect but I am improving. Effectively I am removing the word basically from my dictionary, I found that I used that word too many times!.
I would say that giving a talk, it’s just a fraction of what you can take of the full conference experience, I try to understand why the attendees are there, I have been able to learn about current market needs and what other people is doing and what common problems they have. In my personal life I do not often talk to strange people, I had to fix that, attending a conference requires to approach people, introduce your self, know them and learn about their experiences. When doing Networking at a conference and get back home, it’s not just about to count the number of business cards you get, more important is to grab others experiences and how that knowledge can help to set right directions to your company or personal life, and this works on two-ways, your other end will learn a lot from you, share experiences is the key.
What I have learn
- Fluentd is widely used!, I met several users in different countries and I have get very important feedback about their use case.
- Traveling to a different country with a serious different time zone can affect you a lot. What worked for me was to travel with enough time to adjust my internal clock. Since I start traveling I adjust my food schedule, always start thinking as you are in your destiny. If in the target country it’s already breakfast time, go for it, drink a lot of water, force your self to behave internally as you live there.
- If your target country time is opposite to where you come from, go out: walk, visit some place, stay awake, don’t stay at the hotel, don’t take a nap.
- I discovered that when I start talking at a conference, it takes me between 8-10 minutes to get relaxed and be more fluent in English, of course that time it’s far to be ideal, I am working on that.
- If your presentation have demos and you have to open your Terminal, NEVER EVER use a terminal with dark background, projectors and general illumination will not help your attendees to read what you are writing. Instead try to use a white background with big black letters, forget about your fancy text colors, nobody cares, what really matters is that people can read and understand what you are doing.
To keep in mind
A conference is good or bad based on the speakers and it message, if a conference have a good list of speakers and topics associated, they will attract more attendees. As a speaker, always make sure to keep in mind that attendees go to your session because of your message, not for your background, let’s put our ego’s behind. If we focus 100% on the message, our content and session flow will be better. People love to learn from our experience and stories, just make sure that most of what you share will re-enforce the message.
Monkey & Duda I/O
I continue working on Monkey HTTP Server, during this year I have slowly migrated the core architecture to prepare it for the incoming HTTP2 protocol support, the previous design was not flexible enough for that, now all pieces are in place.
I am happy that Treasure Data fully support this, Monkey it’s a fundamental piece of Fluent Bit, some months ago they supported a contest for the new logo we have:
Yeah, it changed a lot, some people complained but now Monkey is more robust, fully armed with features and more extensible. I think this one represent better it reality.
Monkey adoption continue growing in the Embedded Linux world, mostly for ARM and MIPS based architectures.
About Duda I/O which is a web services framework written on top of Monkey is used in production a lot, and in the last months after the new Monkey architecture was in place, I started migrating the code base to the new model, it’s still work in process so I expect to have some news before the end of the year.
