28
June
2011
|
05:27 AM
America/Los_Angeles

Pervasive Software: Betting On A Big Data Hadoop Future

I recently met with Mike Hoskins, EVP & CTO of Pervasive Software, based in Austin, Texas. It's an interesting company that combines a large legacy database management system with developing technologies for big data applications around the Hadoop database.

Here are some notes from our meeting:

- The Austin tech scene is doing well, there are a lot of startups. And there is a large Hadoop user group.

- We are 30 years old which is rare in the software business. We have had 41 quarters of profitability

- Our products help companies integrate their IT systems withy applications. Integration is a tough problem but it is also one that will be around for many years. If an application hasn't been customized it is easy but most aren't.

- Our integration platform is build it into many "cloud" platforms, such as ADP and Intuit's.

- We recently opened a marketplace for adapters, where people can resell connectors for various systems.

- We have an innovation lab where we focus on new projects. One of these on Hadoop and how to improve performance.

- Hadoop is really taking off because it lets companies manage very large amounts of data. A key feature is that it's scalable. But it's a brute force scalability, you need to add more processors to scale. There is no more "free lunch" from faster processors because of multicore architectures.

- Modern processors are now multicore. We are looking at speeding up Hadoop performance by making it more efficient at taking advantage of multicore architectures.

- Parallel processing will speed up Hadoop applications but it is difficult to do. We are developing a Hadoop accelerator for big data applications.

- The database market is very interesting, there is massive disruption happening.

- Data is immortal. Applications are just a skin that sits on a database.

Additional Info:

Pervasive Software Announces Pervasive TurboRush™ for Hive, Enabling Developers to Uncover Value in Big Data on Hadoop

Get started with Hadoop: From evaluation to your first production cluster - O'Reilly Radar