When you think of travel, the new-age version of it i.e. – the one which is almost opposite to how our grandparents used to travel, you cannot not think of pioneers like MakeMyTrip.
It’s been a trailblazer in this space and has been shown to slice some 50 per cent of the country’s online travel sales, besides being a top-of-the-mind name for many young, yuppie travelers.
But when you think of the blistering speed and those thin patience levels that the digital world is famous (or notorious) for, you cannot just wish away slow page load times, costly downtime, holiday season avalanches and other issues that volatile spikes in traffic trigger.
Sometimes, the problem is in the application, and curiously enough, sometimes it’s much under the hood, let’s say- the database. We get an opportunity to understand this side of the digital road with Sanjay Mohan, CTO at MakeMyTrip, as he paints a journey that navigates areas like uptime, provisioning, scale, caching, nodes, clusters, and more.
Can you start with some peek into how has the experience been like in the post MySQL scenario so far?
With Clustrix, we have seen better performance, and better uptime. The major advantage we get from Clustrix is that load is distributed among all nodes since data is shared and with increasing traffic, adding more nodes to same cluster is very simple and helps to scale easily.
So that helps in terms of cart traffic spikes, activity and number of servers?
When it comes to accommodating a certain level of cart activity, ClustrixDB was originally designed around a cluster of smaller servers. Today, customers use a simple software-based version of the product to easily add or subtract capacity with just a few clicks according to their traffic demands, so that way, there are no complex workarounds required.
Now, there’s no limit to how much traffic MakeMyTrip can handle, and customers see zero drop-off in site performance. We also save immensely in development, operations and administration every year.
Why do sites and digital enterprises get challenged with traffic-peaks, demand fluctuations etc.?
Basically, it’s a matter of handling scale and how you provision to handle the scale. Consumer internet companies typically handle scale with redundancy of hardware capacity, or with smarter alternatives on the software side – typically done using caching on the web front-end layer, and by using performance data-stores on the backend.
What about alternatives like PostgreSQL, NoSQL and new-age database trends like Cassandra, Mapreduce etc.?
Well that depends on the use-case. At MMT we’re using a mix of relational, non-relational (NoSQL), batch, and streaming systems. Some of the use cases also employ persistent message queues.
So the role of technology at the company continues to stay strong?
Yes, MakeMyTrip is a technology-led, online travel company. Technology is central to everything that we do here.