London’s number one taxi app has come a long way in two years thanks to real time data analytics.
The taxi application Hailo has grown at an incredible rate. Since launching two years ago, Hailo is now present in 15 cities. It was the seventh taxi app in London to launch, but has become the most popular with over half a million downloads from the Android and Apple stores.
The application uses Cassandra data bases for storage and Acunu analytics to run analytic queries on the big data in real time.
"When we started down the Cassandra road, the first thing we did was put customer records into Cassandra. The long term view was to be able to operate solidly through Cassandra to be able to run AWS," says Dave Gardner, architect at Hailo. "The interesting thing about Cassandra is it doesn't have the same features as MySQL. We would have done search features, analytics queries, but we were planning to transition so knew we would lose this ability to do analytic queries.
"I saw this Acunu product and as soon as I saw it I knew we should have it. So it was great to be able to try it out and we saw it would be able to fill the gap. So we have Cass for pure storage, elastic search and Acunu analytics for adding stuff up and analytical queries."
"Acunu is all about real time big data," says Dai Clegg, VP marketing at Acunu. "What we do is we provide a platform. People out there who want to deal with high velocity data where they have thousands of events per second arriving, and they need to cope with that. They don't want to just land it in the database, so that they can run a big analytics programme over it later; they want to know what is happening now."
Tim Moreton, founder and CTO at Acunu says: "We sell software to people who are looking to get value out of big data. In particular, big data where the problem is velocity, rather than variety. We help people get value out of high velocity data which is applicable across a wide range of verticals."
Clegg adds: "Acunu is used by organisations who have a lot of data and need immediate insight into that data. Business that are driven by the data they process, whether through products or services and we give them an insight into what happens."
Moreton explains how whenever anyone uses Hailo, Acunu responds within seconds: "Every time you move the little pin, data is going into Acunu analytics. Every five seconds, every taxi in London using the app is sending back its position to Acunu, so you can imagine that's an awful lot of data as we're collection information which you can use to look at traffic conditions.
"Monitoring is a word you find associated with auto analytics a lot, but we are monitoring and helping people understand what is happening in their environment. Are people hailing taxis in a pattern, what is the supply and demand in London?
"The interesting thing about Acunu is we can pump that info back out to the end user. Hailo have started doing this thing where they can show the information to taxi drivers and show them where they can pick up a fare in less time from a certain spot. This was very hard to do previously and it now increases driver engagement and reduces customer waiting time. So it is valuable on both sides."
Clegg highlights the importance of immediate data feedback for the taxi app as Hailo don't own any taxis. "All they have is an app for the customer and for the cabbie and a database in the middle. So they compete on their ability to be smarter with data in real time than their competitors," says Clegg.
"The first app we did for Hailo was monitoring their network and making sure that when people were asking for taxis they were being responded to quickly and the whole process was running quickly so their service level agreement was high. It was important for them as a start up to rage engagement. We weren't providing data to the cabbies at first."
Clegg continues: "We want the business users to be able to give it to the data team and they can do it themselves. The IT guys can the get on with the difficult stuff and the business guys can get involved."
Moreton says: "It's about that empowerment: anybody who is able to use excel can be empowered to get insight from an entirely new class of data that is arriving fast. That's something that traditional business tools just can't tackle. So I think its' a really exciting first."
Gardner stressed the importance of using Cassandra and Acunu for Hailo because the company always had a desire for greater resilience and knew it would be scaling the business out worldwide. "We wanted servers to be located near people using Hailo, so it was important to have locality of service," he said.
"It was quite difficult to do with data before we used Cassandra, but that's relatively easy to do in term of running a database app that has data on multiple continents and will shuffle that data automatically for you. The main driving force was that moving onto more than one continent we wanted zero downtime, so we picked a database that would address those needs.
Gardner continues: "We are early adopters of Acunu, but that gives us some advantages and disadvantages. Advantages are we get our own way in that we can suggest our own features and they'll add it. A lot of our information is geographic, so we suggested that it would be useful if we could draw it on a map, so with the next edition we got that. The turnaround was quite incredible.
"The disadvantage is planning capacity: Acunu will happily count up what it asks you to count up, but if you ask it to do too much it has a fixed capacity using Cassandra, so we have a fixed number of requests per second we can do with Cassandra. We could just add boxes and get more capacity, and it is tempting to add do anything, but that's not right. We don't need that. We're learning to be more disciplined and create the tables where we need them."
The take-up of Hailo has been very impressive. It is becoming more commonplace to see London's black cabs with the Hailo sticker on the back, or even yellow cabs, Hailo's trademark colour, with the black Hailo logo emblazoned on the side. The network of taxis using Hailo is rapidly expanding.
"It's not without its challenges because Hailo is all about the network and supply and demand: you have to balance it. If you have too many drivers they're not happy because they're not getting enough jobs and if you have too many customers they're not happy because there's not enough taxis," says Gardner. "But the company is phenomenally successful. We need to make the technology not inhibit growth, but use the tech to be supportive of the growth and almost drive the growth.
"We want to take Hailo to as many places as possible so if you go somewhere you can get a cab easily and if you're a taxi river you spend less time driving around empty by finding more people who are off the beaten track who wouldn't usually take a taxi," says Gardner.
"Then in the more distant future there will be more coming from Hailo, we are basically a network. We have more than half million downloads, so we have this enormous network and the future is to build more services and leverage that network to try and solve other problems for people with as much slickness and efficiency as we have with helping people trying to get a taxi."
get a cbr Big Data weekly update
Unable to register now