Hadoop vs. Alternatives
June 05, 2019

Hadoop vs. Alternatives

Anonymous | TrustRadius Reviewer
Score 8 out of 10
Vetted Review
Verified User

Modules Used

  • Hadoop Distributed File System
  • Hadoop MapReduce

Overall Satisfaction with Hadoop

It is being used at our Fortune 500 clients. It is great for storage, but it is not well understood by the business. The challenge is that it requires very sophisticated data scientists to use properly and in parallel, but the data scientists turn the data on its head, causing IT execution issues. This has forced IT to restructure data in a denormalized form so the business users can actually be productive. This is a big trend in organizations.
  • Great for inexpensive storage, when originally introduced.
  • Distributed processing
  • Industry standard
  • Network fabric needs to be more sophisticated.
  • Need centralized storage.
  • The three copy of data should have been in the original design, not years later.
  • Consider deploying Spectrum Scale in these environments.
  • Hadoop was thought to be cheap, but it is actually a very expensive proposition.
  • Support is required for Hadoop, so it is not free from a support perspective.
  • The overall benefit of Hadoop is extensive scale out storage and processing, but it is difficult to tie it to ROI in a major corporation.
When comparing to the sophistication of IBM GPFS (Spectrum Scale) to Hadoop, it is clear that Spectrum Scale is a much better choice. That is maybe something you don't want to hear, but in all of our research, this has been the final decision of the client.
Massive processing in a distributed environment with data that can be distributed. Research environments. Lab environments would also be a good use for Hadoop. Hadoop can also be used in support of Spark environments and used by Frameworks if deployed properly. The best scenario is with a Data Scientist that understands how to program appropriately.