Pinup: Qubole – The Growth Of Big Data as a Service

Qubole-BDaaS

In late 2013, the growth of ‘Big Data as a Service’ (BDaaS) was being tipped as one of 2014’s standout trends. It was speculated that BDaaS was even the natural replacement of Software as a Service (SaaS) – the SaaS approach to acquiring specialised enterprise applications had proved itself to be a valid technology model, so it made sense to market analysts that big data applications would follow a similar path.

The three principal benefits of BDaaS are similar to those of SaaS. Firstly, BDaaS solutions are highly scalable; it’s an important benefit because big data can become incredibly resource hungry in a very short space of time – more so than any other current technology can. As data repositories grow in size, so does the demand for storage space – the upshot being that hosting big data applications in-house can quickly lead to scalability issues.

Secondly, BDaaS offers improved security; the nature of big data is hard to monitor and audit and small companies would face massive resource overheads of trying to secure repositories. BDaaS transfers the responsibility for security the host.

Finally, quality and level of service. Big data and related technologies are still emerging, therefore, there is a significant skills shortage in the market. It means that developing big data applications in-house is not something most companies can consider. BDaaS offers the client company a way to procure ready made big data services that will be maintained and extended by the service host.

(Chart Image Source: BigData 50)

Big-Data-Investments

On startup that has embraced BDaaS is Qubole, and the quality of their service has already earned them an excellent reputation in the market place. The company was features in Jeff Vance’s ‘Big Data 50 – The Hottest Big Data Startup’s of 2014’ and won a DataWeek award for their ‘Presto as a Service’ offering. After receiving $7 million in funding from Lightspeed Ventures and Charles River Venture their growth has been explosive; they’ve already secured clients as varied as Pinterest, MediaMath, Nextdoor and Saavn, with more big tech industry names undoubtedly set to follow.

Pinterest data engineer Mohammed Shahangian recently enthused about the product in an interview, saying “Qubole has been a huge win for us. Qubole has proven to be stable at petabyte scale and has given us 30%-60% higher throughput than Amazon EMR. It has also made it extremely easy to onboard non-technical users”. It’s a glowing endorsement.

Their feature-rich big data platform was designed by the creators of Facebook’s and Apache Hive’s big data infrastructure, meaning their cloud-based big data service offers the same advanced capabilities as those used by large big data organisations. Its features include Hadoop as a Service, an intuitive GUI, optimised Hive, improved S3 performance and managed clusters – though arguably its biggest benefit to small businesses is the auto-scaling. The feature means that Qubole actively saves clients money by “spinning up users’ clusters when a job is started, automatically scaling or contracting them based on the workload, and spinning the servers back down once the job is done”.

For more information about Qubole’s pricing structure, head to their website to find out more.

By Daniel Price

If you are an exciting new startup in the Cloud, Big data, IoT, Wearable tech space and looking to be covered in the fantastic CloudTweaks community. Drop us a line and you may just be featured in our Pinup series.  

For Enterprise level exposure opportunities, please review our new Whitepaper and Premium Newsletter offerings.

Follow Me!

Daniel Price

Daniel is a Manchester-born UK native who has abandoned cold and wet Northern Europe and currently lives on the Caribbean coast of Mexico. A former Financial Consultant, he now balances his time between writing articles for several industry-leading tech (CloudTweaks.com & MakeUseOf.com), sports, and travel sites and looking after his three dogs.
Follow Me!

Comments

  1. says

    This technology capability will become simple and performant enough that clients will deploy their own BDaaS on public clouds, private clouds, and hybrid clouds. The need for Qubole will shrink to the longtail until even the tail will deploy themselves. Look at past app platforms like Engine Yard and the like.

    • says

      Jim – would love to catchup. I almost wish I could agree with you. I wish I could go back to spending more time programming query engine internals and worrying less about our service. But what I have learnt in this business (and in prior life) is that running large distributed runtimes efficiently is hard. We are doing weekly pushes of new features and bug fixes without disrupting customers. That is hard. Building adaptive strategies into software that work reasonably well for the largest to the smallest customers *autmatically* – that’s hard. Being able to optimize workloads for customers *automatically* given their runtime history – that’s hard. Dealing with AWS and Azure and GCE and their quirks – and not losing sleep over it – that’s hard. Dealing with deployments all over the world at minimum operational cost – that’s hard.

      The comparison to running a web server farm (Engineyard) is naiive. Those are stateless. They aren’t coupled/distributed in the same manner. There’s just no comparison – and again in a way – I am glad if competitors think running Hadoop or other services is like running a web farm :-)


cloud-sponsorship

Add Comment Here