Industry Leader Q&A with IBM Netezza's Phil Francisco
The market for data warehouse appliances - solutions consisting of integrated software and hardware - is heating up, with new twists emerging from both established and new appliance vendors. Netezza, an early proponent of the appliance approach, was acquired in November 2010 by IBM. Here, Phil Francisco, vice president, product management and product marketing for IBM Netezza, shares his views on what's changing and what's ahead.
DBTA: When we talked to you 2 years ago, you said appliances were moving from niche offering to mainstream. How would you characterize the market now?
Francisco: It is robust. We are seeing great penetration in the marketplace not only for our appliance but also the broader portfolio in general. If you look at the suite of products that IBM can bring to bear in terms of data warehousing and analytics, there is a nice portfolio of solutions that gives customers the option to go anywhere from fully customizable solutions to more refined solutions. This gives business flexibility to the appliance approach that we have through the Netezza system.
DBTA: Netezza was an early entrant in the appliance arena.
Francisco: When we first came to market, we were solving complex query problems that no one else could solve. Today, we have customers that use us across the whole range of data warehouse application solutions in anything from analytic sandboxes and the traditional approach of appliances all the way up to enterprise data warehouse solutions.
DBTA: The approach is well-understood?
Francisco: There certainly is a perception as to what an appliance delivers - flexibility, in terms of how easily it can be deployed into the data center, and how rapidly it can return value to a customer because it really does come online very quickly. But I think there have been a number of entrants in the marketplace that have claimed to have an appliance as well, but not everything that is claimed to be an appliance actually is.
DBTA: What are key appliance attributes?
Francisco: First of all, an appliance is optimized for the workload. It needs to be essentially a hands-off system, with the ability to bring it into a customer's data center, have it operational within 24 hours, and be able to start loading data and running queries within 2 or 3 days. And then, the ability for that system to be very easy to maintain so you don't have a lot of tuning, maintenance, and day-to-day functionality that is required of an administrator.
DBTA: TwinFin is Netezza's flagship mission-critical data warehouse appliance offering, but you have others as well.
Francisco: We have a smaller version called Skimmer, which has the same sort of overall performance characteristics and sizing characteristics, as TwinFin, except it is just on a much smaller scale. Typically, it is used for development and test environments, for QA environments, and for some customers that have smaller data requirements, it is used for their production systems.
And, we have announced a product coming to market called Cruiser which is a very high density product. It is meant to provide a queryable archive solution, with the ability to store as much as a half-petabyte of user data per rack, scaling up to about 10 petabytes.
DBTA: How are customer requirements changing?
Francisco: A lot of our customers over time have asked to extend the functionality into the domain of advanced and predictive analytics because, more and more, they have to compete on that basis to really be a factor in their individual markets. One of the other things that we have done with TwinFin is introduced a suite of analytics capabilities called i-Class that allow people to migrate advanced analytics functions and capabilities right into the database instead of having it run outside. You get the same kind of performance lift from these advanced analytics, and you reduce the latency of having to move big chunks of information from one system to another.
DBTA: What else do you see ahead for appliances?
Francisco: Still today, although some of our competitors would have you believe otherwise, it requires a focus on the kind of applications that you are trying to serve with your appliance. The notion that you can have a single appliance in your data center and it will do everything for you including OLTP as well as data warehousing all at the same time, and be best-at-class at both, is still a fallacy.
The other thing that you will see as we go forward, and maybe we can have a talk again in 2 years and you can tell me if I was right, will probably be very specific, vertically-oriented solutions that are built on appliances, whether it be for telco, retailers, or financial services. These appliances will take into account the kinds of data models that are required but also the kind of functionality that is required for those industries and relieve even more of the day-to-day administration requirements.