Faster Databases Without Special GPU Technology

DZone 's Guide to

Faster Databases Without Special GPU Technology

Demand for speed and scalability grow in conjunction with data volumes. Dive into this and more in this interview with Pradeep Bhanot, Director of Product Marketing at Actian Corporation.

· Database Zone ·
Free Resource

Thanks to Pradeep Bhanot, Director of Product Marketing at Actian Corporation, for sharing his thoughts on the current state of database performance acceleration.

How are you and your company involved in databases?

We develop and market multiple databases. The database types that Actian supports include ones for embedded use, relational, analytical, and NOSQL object technologies.

What are the keys to a successful database strategy?

We believe no one database is suited for all uses, so different databases, though Actian has coupled its OLTP database with a high-speed analytics engine to create a hybrid database. Every application needs security and data integrity as a first-order requirement. The ability to scale beyond a single system using a cluster of systems is vital for throughput. We take advantage of HDFS to scale on clusters.

As for speed, we have focused on looking beyond in-memory to take advantage of the high-performance computing support that Intel has included in all the 64-bit processors to accelerate analytics queries.

How can companies benefit from databases?

Without database management systems we would be accessing log files sequentially, without indexes. Reading every record, in sequence very fast was achieved 30 years ago using magnetic tape drives with lots of tracks and the media moving at several feet per second. Disk drives can read ahead if for a sequential scan, but usually have to jump around as updates get written where there are gaps. This process of seeking out the next block is relatively slow. Databases index blocks of data to speed access and optimize storage use.

Without modern databases computers could take days to get answers that databases can fetch in seconds.  

What are real-world problems you, or your clients, are solving with databases?

Expandium, a leading mobile operator has deployed a 12-node Actian Vector database on HDFS to enable troubleshooting for more than 20 million subscribers. They retain a rolling 10-day window of around 5 billion transactions per day which enables them to drill-down to more than 50 billion individual call records to improve the overall quality of service and control customer churn.

What are the most common issues you see companies having with databases?

Customers are demanding more speed and scalability as data volumes grow and cold, ad-hoc queries are required to perform well without traditional hand-tuning.

Increasingly, Customers want on-premise and cloud-based deployment flexibility without sacrificing query speed.

Where do you think the biggest opportunities are in the evolution of databases?

Organizations want to understand their customers browsing habits so they can proactively market to them. To do this they need to analyze masses of web-log data. Being able to do this on commodity servers and with minimal IT intervention is the segment that we see is an interesting growth opportunity.

IoT applications that need storage on-edge devices is another area we are seeing increasing demand.

What have I failed to ask that you think we need to cover?

In-memory databases have evolved to take advantage of commodity processors, without resorting to specialist GPU technology to gain speed. 

database ,gpu ,database performance

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}