Why CCV chose Exasol over IBM DB2 and Microsoft SQL for consolidation and analytics

For the EPoS company the price was right

CCV is a European provider of payment terminals, software and services to small and medium-sized retailers. In its native Holland the firm runs about 60,000 such terminals, which it manages centrally, including installing and updating customised software for bespoke services, and monitoring the performance of transactions. In order to do this efficiently the IT team builds analytical models, but doing this in an agile, iterative way proved difficult, largely because the databases and the front end tools were not optimised for the task. Moreover the data itself existed across many silos, and the firm needed to consolidate it as a first step.

"We needed a high performance low cost database where we could put all our relational data," said business intelligence specialist Cees van Leuveren.

"We had a desktop system based on QlikView that was fine for basic reporting but our business and information analysts needed more flexibility."

Van Leuveren's team looked at Microsoft SQL and DB2, databases that were already in use at CCV, but found them "very poor in terms of logical database structure".

"They needed lots of tuning and optimisation to get a good performance," said van Leuveren, adding that the firm eventually chose Exasol as a central data store for its raw data with a logical schema calculated on the fly. "It's self optimising and it's very rare we have to do any additional tuning," he said.

There was some scepticism before the initial proof of concept, van Leuveren said, because Exasol was not a well known brand, but the combination of performance and price persuaded CCV that it was the right tool for the job. "It took a couple of months of testing with heavy workloads for certain people to be convinced, but after that it was smooth sailing."

The proof of concept was run on an old generation 6 HP blade server with 30GB RAM but CVV will upgrade to an eight-core generation 8 machine this year with twice the memory and supported by a modern SAN. Deploying Exasol is the first step in the company's plans to create a new data warehouse for archiving, taking that duty away from the many source databases the company uses. Later the company may add on another stack for BI and reporting duties.

For now though, the main focus is on consolidation and analytical modelling. For example, when rolling out a new combination of hardware and software, or upgrading an existing system, CCV needs to know how successful the changes have been, and what optimisations or bug fixes might be necessary.

"Trial and error modelling is very good for this," van Leuveren said. "To get a model running is a very agile process, and with Exasol you have enough performance to do that. You can process all the historical data from all the terminals in 10 minutes rather than having to wait a couple of hours and you can run a test job in a minute."

A complete analytical model can now be built in a week, rather than the "month or two" it used to take using the old system, he added.

At the start Exasol came in at €30,000 including ETL tooling and remote support, but CCV will most likely scale up its investment as new systems come on line.

"We are testing what sort of load we can expect but whatever it is it will definitely come in at less than €100,000," said van Leuveren. "It's very hard to find another database that can do this for the same price."