In-memory is not a step-brother

MUMBAI, INDIA: Among names like Netezza, Greenplum, Oracle, MS SQL Server and other erstwhile players, there is a new slew of concepts and trends to confront as smart data warehouse majors have sniffed out already. Some of these are not vapours to be waited out but  rather a trail of smoke that can be a precursor of strong flames somewhere. With mixed workloads, changing dynamics between transactional and unstructured data or between Online Transcational Processing (OLTP) and Online Analytical Processing (OLAP), and of course the surge of analytics at a new level; the market itself is levitating at a new angle.

Interesting to note here then, how Gartner had underlined 2012 itself with some hard-to-miss trends like simplicity-triggered pre-configured Data Warehouse Appliances for predictable performance; interest in Data Marts; optimization of storage via compression and usage-based data placement; Data Warehouse (DW) as a Service and Cloud; and the big trend called In-memory DMBS (IMDBMS) technologies. All this apart from matter-of-course capabilities like advanced functionality for hardware management of input/output (I/O), disk storage and CPU/memory balancing. Gartner analysts had said the data warehouse (DW) was set to remain a key component of the IT infrastructure while the demand for business intelligence (BI) and the wider category of business analytics increased, or optimization, flexible designs and alternative strategies became more important.

So just how much of this smoke is billowing changes at Teradata's drawing boards? We get to know that and more in this chat with Dinesh Jain, Country Manager India, Teradata.

What do you read into all the big and small changes flanking your industry right now? Is it a worry or good news?

Anything good with technology is always good news. Of course understanding the usage is very important. In-memory is certainly a big trend and with the way memory prices are coming down (at a clip of 50 per cent every two years approximately); the space is set to grow significantly. In-memory takes some of the data in our large data warehouses very well. But it can handle only a small portion of data like a CEO dashboard. For now, I guess scales like hundreds of terabytes cannot be pragmatically put inside memory. It can help with faster access with smaller portions of data for sure.

So you or your industry peers should not be worried that in-memory appliances may elbow out some of your DW market footholds?

No way! It can not affect our market. Traditional DW market is driven by large volumes and like I said, in-memory is great for limited amount of data. In future, there might be bigger loads like petabytes but you cannot put enough in-memory while the application space of this piece is also limited. It cannot be economically plausible.

How about flash-based appliances? Are e they going to be worth a note?

Yes, there is some change happening there. We have SSD (Solid State Disk) in our systems. Traditional drives are electromechanical. But the thumb drives at consumer level is different and not powered by electricity. It brings forth the power of flash. Now, can that be simulated at enterprise level? In SSD, data access is fast but expensive and with a lower shelf life. So now, our latest appliances try to make SSD and HDD (Hard disk drive) co-exist. The power in new solutions transfers data based on usage of data to SSD or HDD. Teradata was incidentally, the first one to introduce SSD in a new way. With DW becoming bigger and bigger it helps to distinguish data which is frequently used and the one which is not used very frequently. And SSD helps in strategizing it.

Is there a new market for a data mart or an application-specific analytic repository for a specific, relatively smaller group of users than those for DW? Analysts like those from Gartner had also hinted at this new concept. Are enterprises inclined towards Data Marts instead of data warehouses?

There were always there in some form actually. There could be two specific reasons for this. Core DW can become very big so departmental DW or Data Mart can take over. Or pre-aggregated data for faster access can be a reason. Also, it could have been there for departments which have some kind of specific data that it doesn't want to put into a core visibility zone for some reasons. But from my view, customers were not too much interested. Usage is coming down.

Do you have any peta-byte level customers in India?

We have more than 100 tera byte level of customers. No peta byte customer here. Tera byte level users are large companies and the way they use information is very different. Like eBay where holistic view of analytics is a key part. In India, the peta byte level still has to evolve. We are not far off though, I hope.

Any interesting examples from your key segments with new era-applications?

Teradata operates in many segments. One financial services customer has plans of putting in some 7 lines of business data into the Teradata platform to support and discover a lot of cross business synergies like cross sell/up sell and let management have greater visibility of the business. One insurance player uses Teradata to monitor new business and repeat businesses and has earned multiples of their investment in Teradata in the form of premiums within three to four years of launch. On the other side, n the Automotive segment, Teradata is used for predictive analytics in the areas of warranty, recall management, servicing trends. It is also used to track all the details of the vehicle using a single identification number ( VIN#). There are plans to include telematics and sensor based analytics to even further sharpen the company's ability to pre-empt any vehicle disruption and increase customer satisfaction.

Analytics in Telco segment is higher?

Talking of Telecom, perhaps telecom in India has emerged as one of the heaviest users of Teradata for analytics. This ranges from just being a very quick solution for aiding police investigation and caller location triangulation to advanced network optimization, QoS etc. Using such varied and predictive analytics lets the telcos offer the very best of calling packages, roaming packages, games, caller tunes, etc. to their customer just in time.

What keeps you excited about the company's traction and portfolio in the current scenario?

Most of the organizations are looking to piggyback on the capabilities of Teradata to manage their data repositories and slowly looking at a data mart consolidation strategy i.e. killing off department data bases build for specific end goals and linking the applications directly to Teradata thereby reducing maintenance, monitoring, reconciliation and skillset costs and needs. They are also planning to venture out beyond the structured data world and looking at creating a Unified data architecture which lets then use Big Data elements also in their analytics to gain competitive advantage. This is still quite nascent but a lot of interest has been shown. Also, a lot of interest has been shown in Teradata's marketing automation tool Aprimo which helps marketing departments to be automated and also handled campaign management and marketing ROI tracking.