Meet ThermoGenomika : Our IOT solution for temperature monitoring in server rooms
Temperature Monitoring is critical for datacenters
At Genomika Diagnósticos, a genomics medicine laboratory located at Recife, we work with several lab management systems and equipments that are working fully 24x7 dedicated to support our biologists, physicians, analysts and technicians on registering, receiving, processing of human samples coming from several places around Brazil.
Some of those systems are running on cloud providers and others such as firewalls, LDAP, specific systems and data storages are running locally. We have two small rooms dedicated for our datacenter in order to store all those physical servers running non-stop.
There are several factors that we need to monitor in a data center, such as battery no-breaks, network structured cable , temperature and humidity of the room, etc.
One of our main concerns is the temperature monitoring. Increasing temperature in a datacenter is a risk for malfunctioning or a beginning of a fire because of the hot air exchange in a closed room between the local air and the machines. We really need to maintain all the hardware working in a adequate temperature and humidity.
So let's define our problem statement here.
Our first solution
Our fist solution was a simple physical sensor. The cost was little and we could buy several units to monitor the temperature and log the data temperature in the sensor. The big problem of course is that we needed to enter in the room every time to collect and check the data. It is not reactive, so it does not work if we need to go to the room to check the temperature.
Our second solution
Out second solution was to find a solution in the market that could:
- Monitor the local temperature
- Alert us when the temperature goes up or down above the threshold
- Historical data for analysis to review the efficiency of our air-conditioning equipments
- Based on the current temperature decisions are made such as turn off all the servers or turn on the backup air-conditioning
We found several solutions, but for each room it was costly. Furthermore, it made us dependent on a third provider in case of any problems.
Our third solution, meet TermoGenomika.
At Genomika, I encourage all my IT team, specially the trainees during their experience time with us , as they are getting used to our problems, to point out hidden problems and solutions. Some of those solutions can be a simples change in the process or even a new product for our lab.
We use the problem-based learning training with our team. The idea behind it that our trainees are familiar with the problem and it will try to solve it, During the process, the trainee must understand the problem and try to find the best solution.
In the journey, they:
- Develop a strong learning foundation
- Develop, build, and apply newly-acquired skills
- Experience both success and failure
- Learn to think in more ways than one
- Find innovative ways to counteract existing problems
One of our trainees familiar with electronics and programming skills decided to build a tool in-house to monitor the server rooms temperatures, alert when they start to goes up or down over a specific threshold, and a live dashboard on our tv so we could be alerted in real time when the things start to going down.
I will not detail here tech specs, but one of our trainees (2017.1 team) brought a NODEMCU data logger connected to a temperature sensor. The ESP8266 is connected to wi-fi network and sends the temperature data to the cloud using webservice API that stores the temperature and datetime of the collection into a data store in order to show the data.
All the code for the data collection and sending to the cloud was written in Lua Scripting. The Web-Server was written using the tech stack Django + JQuery + Python 3 + BootStrap.
The project took about one month to go live. For about two years it helped us on the monitoring of two server rooms at Genomika. It saved us sometimes when the air-conditioner had broken.
Here a collection of screenshots of the dashboard screen and the sensors.
The project grew up, and we decided to study the expansion of the use of ESP8266 in other monitoring applications such as: WI-FI signal monitoring of our access points, emergency energy plugs and even our telephone lines when it started to become mute.
The applications are endless! It was a perfect example of joining internet of things with web programming and laboratory management problems!
I have to congratulate to the team that worked in this project: João Victor Vaz (founder of the project and our alumni trainee), Dyego Carlos (WebDesigner) and Lucas Eduardo (our current trainee helping us in the maintenance).
References
- ESP8266 data logger
- Django + Python 3
- Twitter BootStrap
