59th Monthly Technical Session

Laurence Armstrong
henngeblog
Published in
7 min readJul 12, 2019

Monthly Technical Session (MTS) is HENNGE’s mini-conference. As the name implies, MTS is held monthly and its talks are (mostly) about technology.

The 59th MTS was held on June 21st, 2019.

“Reporting JJUG GCC 2019 Spring” by Fukutomi

In this talk, Fukutomi told us about his experiences going to JJUG GCC — a Java community event held twice a year in Japan.

The JVM is not the most well suited runtime for the cloud environments of modern computing. Fukutomi attended a talk at the conference where they introduced a new alternative adept in cloud computing environments called GraalVM.

Apparently it starts up faster, takes 1/10th of the memory and can run Java programs up to 100x faster!

“Grafana Dashboard as code” by Doi

Doi has been exploring service monitoring tools for several years now. He has noticed that Prometheus has been a trend for a while. Prometheus is a leading open source toolkit for data storage, retrieval, and analysis. Doi has also noticed that now when people talk about Prometheus, they also bring up Grafana. Grafana is focussed on the visualisation side of data analysis. It’s robust, easy to use and easy to maintain.

Doi told us about how one of the major problems with dashboards is that they usually have to be manually configured. This is tedious, error-prone and very hard to scale.

With Grafana, everything can be configured with JSON. To make it even easier, Doi uses Grafonnet, a JSON superset which allows you to program your Grafana dashboard JSON making it easier to both read and write.

The configuration files can then be shared as templates or examples throughout the community.

Doi reckons that with its continuous updates and community support, Grafana has a bright future ahead of it!

“Sharing customer information in TCP” by Iwano and Tokuda

In this talk, Iwano and Tokuda told us all about TCP.

TCP is an internal web app used at Hennge to share customer-related information throughout the company. This includes information from their setup, service usage as well as feedback and requests.

Internally, HENNGE’s primary language is English, but most of our customers interact with us in Japanese. What we’ve started doing recently is automatically translating their survey responses internally so that we can see English versions of their responses quickly and easily.

When survey responses are imported into TCP they text is automatically translated using Google Translate API and then stored alongside the original. Now when viewing results using TCP they are also displayed in English, saving us a lot of time compared to if we did it each time ourselves!

Tokuda also lets us know about the TCP dashboard and how it works. He showed us how they take a lot of complicated information and display it concisely to the user in a dashboard that makes it easy to understand.

“From zero to first purchasing order” by Xudong

This is a continuation from Xudong’s previous talk by the same title in last month’s MTS.

Xudong told us of their experience trying to get their prototype ready for production. At first, things did not work very well and a lot of fiddling and small changes were required. Building a visualiser helped a lot with testing and tuning.

Xudong extracted from this experience a golden nugget of wisdom:

Building hardware is similar to building software — the initial design is not that good…

The hardware had issues with light leakage and a misconfigured LED.

After getting feedback and tidying these things up it was time to start manufacturing the first batch.

The team went to the factory where they were going to manufacture the devices. There were a lot of things to figure out and the first day was spent going back and forth refining the manufacturing process. There was a lot of time put into the design and configuration but the manufacturing itself was fairly quick. Our first trial customer will be one of our sales offices.

We’re excited to see how this project will work out in the wild!

“Yubikey” by Kodama

Kodama holding up a Yubikey

A Yubikey is essentially a hardware authentication device. There are several types and they support NFC, USB-A, and USB-C.

One great feature they have is FIDO U2F / FIDO 2. This is an open universal second factor authentication protocol which allows internet users to securely access online services with a single key.

Kodama explains that normally when you log in you have to type in your 6-digit authentication code from your authentication app. Now you can replace doing this with just touching your finger to your Yubikey and it handles all of that for you!

It has various other features varying in usefulness. Kodama told us about features from static password storage to storage of ssh keypairs.

Kodama observed one major drawback using Yubikey for static passwords is that with the touch of a finger the Yubikey will type out and then enter your password. This is great when you’re on the login page, but if you have Slack open and you accidentally touch your Yubikey with your finger then you will also send your password to everyone in the channel you have open.
Not very secure….

“How to save millions with spot instances” by Michael

In this talk, Michael told us about his experience of how he managed to save millions of yen by using spot instances.

With EC2 you can start up an instance on demand and then you pay for how long you’re using it (you can stop at any time). AWS also allows you to purchase reserved instances where you can get 10% off if you pre-pay for this time up-front for a set period (e.g. a year). This is good if you know you’re going to need a number of instances to be always running for the next year.

Another thing you can do it use spot instances. These instances represent currently unused EC2 capacity that AWS doesn’t need at that moment. These are very cheap (almost 70% off) because AWS reserve the right to shut down your instance and reclaim this capacity any time they like.

Michael managed to save us almost 80% from previous months by using spot instances for one of our services!

Your application has to be very flexible and fault-tolerant, but if you can manage that then spot instances are very worthwhile!

“Past, present and future — 6 weeks in HENNGE” by Billy

When Billy was a student, most of what he had done everyday was just print patterns to the console. He was never allowed to use external libraries in his assignments. Hence, Billy never really got the chance to use the actual powerful frameworks used in today’s modern computing.

The first thing he had to do when he got to HENNGE was building a Twitter clone. This was a huge learning curve for Billy. He had to learn so many technologies that were new to him from Docker to AWS to Google Auth. After this, he moved into a new team who used a completely different set of tools. This time Billy had to learn technologies from Vue.js to Firebase to Heroku.

But this time Billy could pick things up quicker. Importantly, Billy had learned how to learn!

For the second part of his talk, Billy told us about his experiences as a lover of travel. Billy used to travel a lot but has slowed down quite a bit since then. However, recently something reminded Billy of why he did all the traveling he used to do and inspired him to try to pick it up again. To motivate himself Billy has bought a stamp book of a 100 Japanese castles. For every one of these castles he visits, Billy gets a stamp in his book. Billy’s goal is to use this book to motivate himself to travel through more of Japan over the next 5–10 years.

He’s very excited to be off and motivated again! We wish you the best of luck Billy!

As usual, we had a party afterwards :)

--

--