Grab
to rewrite the system every two years.
One
example of innovations made by utilizing big data is the presence of GrabShare
and GrabNow. GrabShare is a ride sharing service with others, with a point of
unidirectional direction. While GrabNow is a way to get the fastest driver by
directly approaching the nearest driver that is not in a booking status. At
GrabNow, this becomes a solution with observations in Jakarta that users will
find it easier to order the Grab directly in front of them.
Grab
invested in building a research and development center (R&D center) in
various locations.
Collaborating
with Microsoft covers many things such as big data, artificial intelligence and
machine learning. One of the implementations is planned to use mobile facial
recognition with AI built-in for drivers and passengers.
Big
data analytics not only helps to understand the information contained in the
data but also helps to identify the data that is most important for current and
future business decisions.
Grab
is one company that combines services with technology. With the use of this
technology, Grab utilizes Big Data to support its services in operating its
activities. Big data is an important element for Grab to know the behavior of
its users, as well as its driver partners. With Big Data, Grab can find out
which regions travel the most orders and destinations of the trip. This is used
by Grab to support its business and services.
"A
lot of data collected means that there is a lot of insight for us. In one day
we received 10 TB of data. When totaling the same as multi-petabytes of data.
This is what makes us the most requested online transportation service in
Southeast Asia, "said Head of Engineering Grab, Ditesh Gathani, on October
25, 2017.
His
party utilizes Big Data from passenger and driver data track records to be
processed even better. He called it as "Data Demand". Data processing
can also set the flow of orders at rush hour in certain areas. Take for
example, the results of the data that is processed will optimize the process of
booking passengers and taking bookings by drivers.
One
example of innovations made by utilizing big data is the presence of GrabShare
and GrabNow. GrabShare is a ride sharing service with others, with a point of
unidirectional direction. While GrabNow is a way to get the fastest driver by
directly approaching the nearest driver that is not in a booking status.
He
said that to solve the problem, his team applied the hyperlocal approach. For
example, it sent 15 Grab teams to spend six months in Jakarta. They finally
found that Jakarta citizens would find it easier to order Grab in front of
them.
Ditesh
explained, the purpose of Grab is to want to change the transportation system
in the cities where it operates within the next 10 years. This is what Grab
wants to achieve by combining the functions of big data and machine learning,
which can predict consumer demand. Not to forget, Grab also plans to
collaborate with the local government. Currently, Grab claimed to have shared
data in real-time with the Singapore government, which contains data related to
location, direction of travel, speed, to analyze traffic flow.
Quoted
from Tech Wire Asia, Monday (12/17/2018), data collected by Big Data Grab
reaches 4 petabytes or about 4,000 terabytes. The amount of data is equivalent
to 2 trillion pages of writing or a 53-year-long video with high resolution.
Ditesh
also said, the abundant data on one side forced Grab to rewrite the system
every two years. Therefore, the Grab engineer team works only to provide
solutions that are valid for a period of two years.
Abundant
data, making companies willing to invest heavily to build research and
development centers (R&D centers) in various locations. The total R&D
Grab has six points, Seattle (US), Ho Chi Minh (Vietnam), Singapore, Beijing
(China), Bangalore (India), and Jakarta (Indonesia). The choice of location is
also not arbitrary. It considers the availability of qualified local engineers
to help the Grab business. For locations that do not exist in the Grab business
area, such as Seattle, Beijing and Bangalore, it was chosen because in that
country it has good talented engineers due to the presence of various triple
class A technology companies.
Of
all the data processing results mentioned, Grab invests more in the use of big
data. Unmitigated, it has also pocketed capital disbursement from Softbank with
a value of US $ 750 million or equivalent to Rp 9.8 trillion.
Meanwhile,
cooperation with Microsoft will cover many things such as big data, artificial
intelligence and machine learning. One of the implementations is planned to use
mobile facial recognition with AI built-in for drivers and passengers.
After
the ETL process is completed, other services to perform analytics such as
holistics, tableau and Spark will access the data in the data warehouse. The
obstacle experienced by using architecture like this is that the analyzed data
is not real time because it is yesterday's data and the burden on Redshift as a
data warehouse is very high along with the amount of data analytics needed.
At
the end of 2016, the company took the decision to change the existing
architecture with the consideration that the old architecture was not able to
provide data analytics in real time and server implementation became difficult
given the large number of requests so that if the server was upgraded it would
cost a lot. Grab made the decision to move the server to the Cloud service on
Amazon and move to use Data Lake by utilizing the Helios service from Amazon.
Every existing MySQL database will be combined with the PyroisOrchestrator
service which will automatically perform ETL every hour and the ETL results
will be stored directly into Data Lake. In Data Lake, data is stored as Parquet
and partitioned according to time. According to Grab, partitions must be in
accordance with company requirements, time-based partitions are considered
appropriate for Grab because the relevance of the data most needed to be
analyzed in this company is limited by the time domain. Grab creates a Data
Gateway that is integrated with Google authentication to restrict access to
existing Data Lake. With the addition of this security layer, Grab can limit
the people who can access data and limit the access and queries performed.
References
Firdaus,
S. (2018, December 15). IMPLEMENTASI BIG DATA ANALYTICS PADA APLIKASI GRAB.
Retrieved from satrianifirdaus.my.id:
http://satrianifirdaus.my.id/2018/12/15/implementasi-big-data-analytics-pada-aplikasi-grab/
KumparanTECH.
(2017, Oktober 25). Cara Grab Pakai Big Data untuk Memahami Penumpang dan
Pengemudi. Retrieved from Kumparan:
https://kumparan.com/@kumparantech/cara-grab-pakai-big-data-untuk-pahami-penumpang-dan-pengemudi
Nabila,
M. (2017, Oktober 26). Mengintip Strategi Grab Optimalkan Big Data dalam
Operasional. Retrieved from DailySocialId:
https://dailysocial.id/post/mengintip-strategi-grab-optimalkan-big-data-dalam-operasional
R.,
J. I. (2017, October 25). Grab Sebut Big Data Jadi Strategi Penunjang Layanan.
Retrieved from Liputan6: https://www.liputan6.com/tekno/read/3140751/grab-sebut-big-data-jadi-strategi-penunjang-layanan?utm_expid=.9Z4i5ypGQeGiS7w9arwTvQ.0&utm_referrer=http%3A%2F%2Fsatrianifirdaus.my.id%2F2018%2F12%2F15%2Fimplementasi-big-data-analytics-pada-aplikasi-grab%2F
Setiawan,
R. (2018, December 17). Grab Punya Big Data 4.000 TB, Setara Video HD Durasi
53 Tahun. Retrieved from DetikInet:
https://inet.detik.com/cyberlife/d-4347750/grab-punya-big-data-4000-tb-setara-video-hd-durasi-53-tahun
Komentar
Posting Komentar