We’ve all been there: you want to go out and meet friends in the city. You set a date, time and place and show up. But what happens? The friend you were expecting to meet is simply nowhere to be found. It happened to me one day when I was really looking forward to meeting up with that one particular friend after a number of months in Milan. I was hoping to join a typical Italian after-work thing: going out for an “aperitivo” (my personal recommendation when you visit Milan – you should definitely enjoy this once in your life!).
My friend who is literally always late, had quite an interesting excuse and I want to share what he told me that day. He works in IT or, more specifically, for an advisory company offering business consulting and IT engineering on a lot of levels. His job that day was to finalize a project on big data analytics, which was why he was unable to join our wonderful “aperitivo” on time.
He was programming a code which was to be used for a customer in the media publishing industry. He explained that this company wanted to identify their subscription customers’ behavior individually, so as to determine the best renewal price once a subscription ends. Since both their offerings, online and offline magazines, provide plenty of opportunities to trace readers’ behavior patterns, there was an immediate need for a structured database containing a lot of data input (such as payments, customer service interaction, websites visited etc.). Hence, the first challenge for him was to consolidate heterogeneous, unstructured data in one single database that could be analyzed. After the first test runs of the developed code, the company was quite surprised to discover that there was a correlation between contract renewal and customer service contact rates, for example. This was a new data-driven point for optimizing the entire business model that nobody had ever thought about before.
That night, he was deeply lost in thought. He was trying to figure out how the first sets of test data could be used to provide “training” in the code, so as to identify specific patterns within the data. In other words: to find correlations as initially intended.
He is a smart guy, so after a short while he managed to start the first test run with the newly enriched and improved code against a copied set of real data, testing and validating everything in a closer-to-real-life-setting. Oh, by the way: in this case, we’re talking about a volume of data measured in GB, if not even TB. However, it turned out that the test system was not as fresh as the code and it experienced problems delivering the performance required to enable my friend to be on time.
With the knowledge and options available today, something like this does not necessarily need to happen anymore: Fujitsu just released an entire set of new x86 machines providing enormous computing performance. One of them is the new PRIMERGY RX4770 M4, which finds its strength in the Intel® Xeon® Scalable family and a scalable, reliable, and variable system design. This can provide a huge amount of memory to cope with up to 12TB in-memory databases and offering headroom for additional devices to speed up an Ethernet connection. What’s more, the certification for SAP HANA® makes it great for business intelligence and data analytics tasks that require exactly these key specs.
Next time, I will be sure to let him know that powerful platforms and IT infrastructures don’t just exist for the sake of an IT engineer testing his code or for running live on a customer’s premises. In our case, if he had been able to test on a PRIMERGY RX4770 M4, this would have given us more time to enjoy a great “aperitivo” night and further strengthen our friendship.
P.S.: Regarding the case described above: the code got gradually better, ultimately delivering the information the company wanted. The solutions now have a database containing all information they already held, an improved set of instructions and, ultimately, an improved renewal rate, as well as YoY increases in their revenues in different segments.
Have you missed some of the blog series about Xeon Scalable Family? If so, please read the full series here: https://blog.global.fujitsu.com/index.php/tag/xeon-scalable-family/