Sybase China technical director Lu Dongming: Database challenges in the big data age

Recently, in the Sybase IQ 15.4 Media activities, CSDN United several technical media jointly interviewed Sybase China technical director Lu Dongming. Lu Dongming shared views on the impact of large data on traditional database vendors, the comparison of column-and row-style databases, and other hot topics.

Sybase China technical director Lu Dongming

Lu Dongming first briefly introduced SAP's 5 largest database products:

Journalist: Why are traditional row-type database companies developing column-type databases or hybrid delivery schemes?

Sybase China technical director Lu Dongming: At present, the veteran row-type database companies are in some way claiming to have the technology of the column. It is worth mentioning, however, that column-type storage, column-type storage indexing, column compression, and column-type databases are completely different concepts. Why the concept of the column is introduced, because the column database is now very strong momentum, has eaten up a large market. So they are confusing words.

Oracle, for example, is similar to Microsoft's (column-storage) solution, putting some of these columns in the user's choice to save rows or columns. Saving columns does bring a lot of benefits, such as high compression, a quick single-column analysis or query on a column. In a sense, this is very much like some of the features of a column database. But fundamentally, they are different things. Pure column database is no decompression and reverse operation process, Sybase to the data after the code, and then stored in these encoded form, but also do additional compression, but the efficiency of the additional compression is not the most important, more importantly, coding.

Based on the coding operation, the traditional line solution has to be read, repeatedly decompressed and disassembled before the operation begins. So many row-style databases are confusing and want to join the column database camp. But why not claim to be a column-type database because he can't. There are several reasons for not doing so, one reason is that Sybase still has a lot of patents, and another reason is that it is really hard for traditional data makers to abandon their baggage.

Journalist: How do you choose a row-or-column database?

Lu Dongming: Any database solution is not omnipotent, special tools to be dedicated. It's like no Swiss Army knife can cut trees and dig wood, and it can make wood into a special shape. The database is a toolbox with a variety of tools, each with a dessert of every tool, irreplaceable. Many database users or developers, still in the old database environment into the new environment, whether active or passive, they have not accepted the database market is a market segment. Always hope not to use so many databases, using a database to solve it. But like current telecom users, it is very clear that in such a data scale, it is impossible to use a so-called all-round players to solve this scale.

Reporter: Hana is there any technology coming from Sybase?

Lu Dongming: Not now, but these two technologies will certainly merge, one is Sybase's memory computing technology, there is an IQ inside a variety of column-type database technology, because there are all kinds of patent index, IQ inside 10 kinds of index, this is all other databases can not compare.

Reporter: The big data to the data warehouse manufacturer has brought what impact and influence?

Lu Dongming: Traditional Data Warehouse manufacturers in the past to solve the basic data type is relatively simple, structured data. There is no complete answer to how to handle structured and unstructured data seamlessly. On the scale of data, this wave of data growth did make the database vendors a bit unprepared, data growth suddenly increased exponentially. There are several ways to develop the database, IQ is also doing a lot of new index research, there is no updated index to data compression, data analysis, query and effect to do better, this is an aspect. In addition, the Hana brings the revelation of memory calculation. Hana used in a technology called Flash card, flash card is directly inserted in the machine.

From traditional storage to SSD to flash card to memory is probably 4 steps, each ladder is about 10 times times the gap. Hana used pure memory, flash card, with a hard drive, where the edge of these several technologies have, IQ is now basically only a small amount of memory and a large amount of storage. Sybase is now working on a solution by adding flash cards.

In addition, Sybase IQ 15.4 is learned to use MPP database and database analysis techniques, and to integrate MapReduce with Hadoop to address the analysis challenges of the big data age and to open up the ability to gain insight into key business.

On the other hand, according to reporter observation, SAP as a large company, the application software is also used at the bottom of IBM DB2, SQL Sever, Oracle database, and the future use of other database products will be the opportunity to shrink, and the shrinking share will certainly be transferred to ASE, IQ, Hana on. This is also an important factor in SAP's ambition to become the second-biggest database market in three years. (Zhangjiang/Wen)

(Responsible editor: The good of the Legacy)