How Sentiance decreased the deployment time in AWS China
Case Study Summary
Sentiance, a contextual intelligence service, is using China Telecom Upstream by Datapath.io to deploy new versions of the software in China. The results of this collaboration are truly impressive:
- 70X faster deployment
- operational readiness
- significant performance improvement in China
Sentiance is a data science company turning IOT sensor data into rich insights about people’s behavior and real-time context. These insights enable companies to understand how customers go through their everyday lives, discover and anticipate the moments that matter most, and adapt their engagement to real-world behavior and real-time context. Our deep learning based solution learns to reason about human behavior, thereby moving from traditional activity detection to cognitive intelligence.
As the Internet of Things gains momentum consumers are rapidly adopting smart objects in their everyday lives. The sensors in these IOT devices generate a constant stream of data about how they behave and why they behave the way they do. Sentiance helps companies make sense of this new and very powerful data. Some of the world’s most innovative companies are adopting the Sentiance platform to help them provide more tailored, personalized and contextual experiences to their customers.
David Damen, VP Engineering at Sentiance, described the main idea of the service: “The mission of Sentiance is to become a standard of contextual intelligence based on sensor data. Contextual intelligence basically means that you want to understand human behavior, find out what people are doing, why they are doing the things they do and predict next actions. Sensor data means that we base ourselves on data collected from sensors, on observed data – meaning location subsystem, gyroscope, accelerometer and other connected devices. The technology we built around it helps us understand the places users visit, how they move from one place to another one, which transportation mode they use like walking, cycling, or taking the tram. Our tech provides more accurate results than, for instance, iOS or Android on the phone itself. We also go deeper, we won’t just say you drove a car, but actually figure out how smooth or aggressive the ride was”.
Sentiance is hosted on the AWS Cloud, in Europe and China as well:
“We have a SDK that can be embedded into any mobile app. This SDK does the data collection – it decides when to turn on or off, when to collect data, at what frequency, when to send it to our backend. The processing is done on our backend, a cloud-based data processing platform, currently hosted on AWS, in Ireland and China. The China case is the reason why we’re using Datapath.io,” – said David Damen.
Sentiance is developing in China now, which is pretty reasonable, taking into attention the size of this market and opportunities that it offers. As David Damen highlighted, “there is a big market there. And the timing is about right for jumping into it now”.
However, many obstacles need to be overcome on the way to the successful development of international web services in China. Thus, Sentiance also faced several challenges there:
“About 8 months ago we decided that we will build out an equivalent platform in Amazon China region. We have various regulatory reasons for that. For example, we need the information which we collect in China to stay in China. The main problem at this point was that we needed to transfer hundreds of gigabytes of machine images into AWS China region which was an incredibly slow process,” – explained Christoph Liekens, Senior Data Engineer at Sentiance.
How the Datapath.io solution helped
Facing the problem of very slow deployment to China, engineers at Sentiance started to look for an optimal solution to solve this issue:
“To find the right solution, we tried some other things like split our artifacts into a lot of small pieces and transfer them in parallel. It works to some extent but it is not a very elegant process. There are several tools which we tried before Datapath.io and it gave us some improvement but we haven’t found any satisfactory solution. The process was still very slow”, – said Christoph Liekens.
Eventually, Datapath.io’s solution turns out to be the best one for Sentiance:
“When we got an opportunity to try out the Datapath.io solution, we did it. It gave us a much faster connection to get our artifacts into the AWS China region. It finally enables us to run our normal centralized data processing platform”, – stated Christoph Liekens.
Currently, Sentiance transfers the data to AWS Frankfurt and then to AWS China, using the Datapath.io China Telecom upstream. It leads to more than 70x improvement:
“On a high level, previously, before we started using the Datapath.io solution, the deployment process from Europe to China took up to one day. With Datapath.io, we brought this time down to 20 minutes,” – revealed David Damen.
Christoph Liekens added: “It is also important that now we can react much quicker when we have some problems to solve in the AWS China region. If you have an emergency and your transfer takes 5 hours, it is not an emergency anymore. Datapath.io’s solution allows us to deploy new versions of our software much quicker and improves our operational readiness since it became a lot faster to see if something is going wrong and fix it in time”.
For the future business development, Datapath.io’s Network Performance Platform would also be valuable for Sentiance:
“In the long term, we’re evolving our platform more and more towards a reactive microservices environment where we want to be able to deploy new versions of the individual services faster and more often than we do right now – more releases in the same time frame. We can only do that if we’re able to get those releases out to our different environments quickly enough. Additionally, we want to expand not only in China but also in the U.S.,” – said David Damen.
Christoph Liekens added: “We’re building out our platform in China even more than what we have right now. For the rest of the world, we will continue using Datapath.io’s connection to bring new versions of our software, extending the platform with new infrastructure components, and, obviously, we’re planning to continue with the operational monitoring”.