Big Data Analytics

Kiss Goodbye to Expensive Upgrades – Smart New Technologies Will Help You With Data Analysis and Interpretation Instead

14 Apr 2016 | Share

In-Memory Database

Leveraging new technologies, such as in-memory, enable you to stay ahead of the competition instead of wasting money on costly hardware.

Data analysis and reporting is taking too long and soon the data you collect on one day won’t be ready for analysis until the next. But yesterday’s news won’t get you very far in a real-time world.

However, hardware upgrades are expensive, and sticking with the old isn’t a viable solution either. This article highlights four technologies you should be looking at, and explains how companies use them to supercharge their data analytics, BI and reporting.

1. In-Memory Computing – Why wait for tomorrow if you can have it today?

Data analytics is becoming increasingly popular, but many companies don’t know that fast analytics isn’t feasible without an in-memory approach. Although hard drives are the most common storage medium of data, the computing speed of CPUs outperforms the speed of spinning hard disks by a factor of thousands, making them a major bottleneck for fast analytics.

RAM’s capabilities on the other hand have held pace with CPU power. Put your data in-memory, and complex queries that would have taken days with traditional database systems now run in minutes or even seconds. Fortunately, RAM prices have fallen rapidly, so memory doesn’t cost the outrageous price you might think it does (a megabyte for a million dollars anyone?). What’s more, block management can substantially reduce the amount of RAM sufficient for performance to less than 10% of the data volume.

Econda, one of the leading providers of web analytics and personalization solutions in Europe accelerates analytics with in memory-computing.

2. Self-optimization – Your computer is better at this than you are

Searching for upgrades, everybody always looks at the price tag first. However, people often overlook the cost of managing a database.

An analytics solution needs to be able to account for many different data sources and use-cases but still offer optimal performance, without needing a custom optimization for every query. You want to concentrate on your business, not your database setup.

Self-optimizing software ensures each query runs at maximum speed without having to fiddle with code. Exasol monitors its own behavior continuously in order to provide the most accurate information to the query optimizer by using statistics about the system. If it’s possible for the optimizer to transform a query graph to an equivalent one that performs better, it does so itself.

By using software that optimizes itself you can avoid time consuming and risk-prone manual optimization, and instead turn your insights into value.

Webtrekk, a marketing analytics firm, profits from self-optimization.

3. Hadoop – Believe the hype, but use it right

Not only do popular websites and apps often have hundreds of millions of users, the data they produce comes in all shapes and sizes, and at great speed. How do you deal with such data? For many companies, Hadoop is the answer to that question.

But Hadoop’s original purpose was to get the job done in the first place, not for getting it done fast. The quick data analytics that many companies need to find actionable insights weren’t really what Hadoop’s creators had in mind.

For analysis that needs to be fast, Hadoop is the wrong tool. Instead, use an analytic database solution that is built for speed, and integrate it into your Hadoop architecture to complement your open source infrastructure while addressing its weaknesses. In other words, use Hadoop for storage, but use an analytic database for analytics.

King, a leading entertainment company, complements its Hadoop infrastructure with Exasol.

4. Advanced in-database analytics – The deepest kind of learning

Data scientists are transforming companies in all sorts of ways. By employing their knowledge in statistical programming, they perform advanced analytics. But bland reasons often foil grand plans. R and Python, the typical languages for these projects, are simply unfit to function in data sizes exceeding dozens of terabytes. The solution is “user defined functions” (UDFs) that allow you to write in the language of your choice but to run code in a fast analytic in-memory database, eliminating speed constraints.UDFs make it straightforward to integrate existing solutions right into the database’s SQL processing. UDFs move the analysis to the data, so you can utilize the complete power of Exasol’s parallel-distributed architecture.The performance and scalability of Exasol combined with capabilities in R such machine learning enables a new generation of advanced analytic applications.

CCV, a supplier of payment solutions, uses UDFs to run advanced analytics.

Get started today

The in-memory approach is arguably the most important of these technologies. Without the speed from in-memory computing, the “how” becomes an afterthought to the “when” (is it finally done?).

Exasol offers a number of ways you can get started quickly and easily in data analytics, all free-of-charge. Exasol is complementary to your existing setup and no “rip and replace” is necessary.

You can either download a single node copy of Exasol to test on your own hardware, or try out Exasol on a demo system in the cloud.

Our dedicated support team is available for any questions you have and would love to show you how Exasol can accelerate your data analytics within the framework of a proof-of concept.

For more on these 4 technologies, read our whitepaper on the topic.

10 trends impacting data analytics

Now that we’re well and truly in the age of data, what’s coming next?