Sybase: Data warehousing too slow for ‘big data’ analytics

Andrew de Rozairo is business development director, strategic partners EMEA (Europe, Middle East and Africa) at Sybase. Born in Brazil and based in the UK, his education includes a master’s degree in business administration from INSEAD and a bachelor’s in electrical engineering from the Massachusetts Institute of Technology.

De Rozairo has briefed SearchDataManagement.co.UK on two occasions this year about data warehousing and “big data” analytics. What follows is an edited version of those conversations.

Where does Sybase stand now?

Andrew de Rozairo: Sybase has its origins in relational databases, but about 10 years ago the CEO John Chen decided to build out the vision -- to unwire the enterprise and base it on how to turn data into information and to deliver at the point of action. Hence the mobility investment we have made in the last 10 years.

Unlike others we have believed from the start in different database architectures for different purposes: databases for mobile devices, for desktop, for transactional enterprise environments and one purely for analytics. And so Sybase IQ is a database specifically built for analytics.

Sybase IQ has a couple of thousand customers. It allows users to do complex queries on big amounts of data on commodity hardware. What it has not done so well in the past is run complex queries over a number of machines, maxing them out. That is what the Sybase IQ 15.3 with PlexQ technology release, this year, was all about. PlexQ can scale out both the data and the number of users, using shared everything MPP (massively parallel processing)  echnology.

How do you position IQ within the data warehousing landscape?

De Rozairo: Traditional data warehousing is antiquated. If you want to access anything quickly you won't find it in a warehouse. The whole concept of putting something into a warehouse doesn't fit with real time. There used to be this vision of an enterprise data warehouse where you would put all your company’s information. Do you know any company that does that? It will never happen.

Traditional data warehousing is antiquated

You’ve told me before that Sybase needs to get better at putting its technologies into context. What is the impediment to that?

De Rozairo: It’s not so much a barrier as [a function of] what we've chosen to focus on as an infrastructure provider. We don't try to develop our own applications for particular vertical markets, except for financial services, where we do address the capital markets [working with Reuters and others]. But the focus has been on building the platform for customers or partners to build on.

Sometimes when you introduce a disruptive technology, you need to give people ideas about what they could do that they couldn't do before. And typically they'll say, ‘Give me an example’, so here's an example: complex event processing, where you are analyzing streams of data. This would apply in telecoms where you would want to provide quality of service at a granular level at speed. So it is no use knowing that the network is providing good service on iPhones for YouTube but not for Gmail in a particular geographical area, one day late. You want the call centre operative to be able to know now. That gives a customer-service edge.

Most companies are, unfortunately, managed by HIPPOs -- ‘highest paid person’s opinions.’ They need to be managed more by data, as in Moneyball! And this means a move away from reporting on what happened yesterday and more towards predictive analytics. It is crucial to roll [this capability] out to everyone who's making operational decisions, not just the higher-ups. ‘How can I improve business processes this very minute?’ is the key question.

So, in your view, speed is of the essence?

De Rozairo: Speed that makes the analytics relevant. If you take two minutes to do a Google search you’d give up. Businesspeople now expect Google-like speeds. And they  -- especially the senior executives -- will expect to consume business intelligence on mobile devices. I believe the iPad has really changed things [in that regard].

Business intelligence now is like the Internet at the Altavista [search engine] and Mosaic [browser] stage. It is early days. What made the Internet take off was usability, through the browser, and relevance -- the ability to search for information that you wanted. The third factor was broadband. That's what we think Sybase IQ and SAP’s HANA delivers [SAP has owned Sybase since May 2010]. But all the big players in our market -- Oracle with Exadata, IBM with Netezza, Teradata -- have really started to invest much more in speed in recent years.

There used to be this vision of an enterprise data warehouse where you would put all your company’s information. Do you know any company that does that? It will never happen.

What’s your take on the big data concept?

De Rozairo: The expectation for speed plus the reality of big data is a perfect storm. And it is the big-user environment, too. Banco Macro in Argentina is a good example. They had six power users using business intelligence to create reports for other people to consume several years ago. Today, they have 1,700 bank managers with access to not just prebuilt reports but the ability to ask ad hoc questions of data. That represents a democratization of decision making.