Hadoop is back. Sort of

5 reasons to use Hadoop-as-a-Service

Nilay Shrivastava
Female hands drawing picture image via Shutterstock

In this article, Nilay Shrivastava, Business Manager at IBM Cloud, explains why we should use Hadoop-as-a-Service. He claims there are at least five benefits that should spark our interest.

Without further ado, here is why you should consider using Hadoop-as-a-Service.

1. Cheap infrastructure : According to Forrester, organizations save 20-60% in Big Data projects as compared to on-premise infrastructure. Even if you have your own infrastructure, the biggest question is the following: Is it ready for the big data project? How much money, time and resources will you need to deploy to get the project started ?

2. Enable faster advanced analytics : The infrastructure of most organizations supports BI — which is just about descriptive analytics. The real challenge is to embrace the predictive analytics, machine learning, text analytics etc. A lot of data has to be churned and made ready so that actionable insights can be drawn from it. Organizations still don’t realize that they have actually tripled the data sources, a lot of data is flowing around which when fed in to advanced analytics system, can help them solve a lot of business problems.

SEE ALSO: Hadoop isn’t ready for the elephant graveyard

3. Scale to unlimited resources in minutes : MapReduce is the reason why Hadoop projects are usually done in cluster environments. How can you set up those clusters quickly and get started with the analytics ? A cluster requires a lot of upfront cost and if you are a startup or a business which doesn’t have (too many) IT resources, you will not be able to acquire those resources.

And the other side of the coin is that you might not need those clusters 24X7. Why would you want to waste your time and resources if your business doesn’t need them?

4. Self-service analytics platforms : Let’s say you have acquired the infrastructure, and have a relevant skill set, would you not prefer something which has already been tried and tested and which actually works? Why would you want to write the algorithms from scratch when you can customize them and quickly prepare them for precise needs?

5. Faster deployment:  There are services available which you can start using within hours. Here are a few examples :

  • Elastic MapReduce from AWS
  • BigInsights from IBM
  • HDInsights from Microsoft


In case you are planning to start your Hadoop project and you need my help, drop me a message or get in touch with me here.

This post was originally published on LinkedIn.


Nilay Shrivastava

Nilay Shrivastava is a marketing and sales professional with 5+ years of experience with expertise in Big Data and Analytics and Cloud Computing.

Inline Feedbacks
View all comments