Memory Computing Age has come

A few times when our team walked through the street night market, see the ground on a variety of small commodities, there is the iphone's foil, mobile phone shell, of course, there are a variety of memory and 8G/16GB small flash drive, this time there will always be someone shouted: "Look!" Memory computing age has come! "While this does not accurately express the advent of the memory computing age, this is basically an indication of a phenomenon," he said. Should our traditional computing architecture change in the age when memory capacity and CPU speed are no longer a problem?

Perhaps some people say that memory technology, or the memory database is not a new topic. Indeed, we can see the same history, just as we are constantly updating our handheld mobile devices. Everyone has spent so many years of mobile phones, various functions have been in fact, speech recognition in more than 10 years ago has begun to do, but this does not prevent a new mobile phone once launched immediately let the global users become its fans. The secret is that it redefined the concept of mobile phones.

SAP Hana, like Apple's iphone, is a technology-innovation product. The so-called innovation is based on the previous accumulation of technology, but also to break through these technologies accumulated inertia thinking and create new things. All of the technologies contained in both SAP Hana and the iphone have their shadow or technical prototypes in many of the previous products. But only if they make the most of the past years of knowledge accumulation and technological innovation, and become the true leader of this innovation force, can other market participants follow the trend.

Common problems in current information system architecture

With the expansion of business and the increase of data volume, the problem of performance is gradually exposed. Of course, this is also in the current massive data or large data background of an increasingly obvious trend. It can be said that the core performance problems of many application systems are directly or indirectly reflected in the disk I/O bottleneck.

So why does this problem occur? Most of the current application systems are based on traditional relational databases (RDBMS), whether transactional (OLTP) or analytical (OLAP) business systems. But the traditional relational database is mainly designed for the traditional transaction business system (oltp--online transaction processing application). Since the cost and architecture of computer hardware in the early years have restricted the design of database, relational database is a classic and once effective design has been used so far. From DB2, Sybase, Informix, Oracle, Ms-sql Server, MySQL and other areas of the emergence of many products, we can see how broad and important the market. The versions of these database products are updated more quickly each year, and the efficiency and functionality are indeed getting stronger. But we have to face the fact that these optimizations and improvements based on traditional relational databases and traditional architectures are based on disk technology.

The explosive growth of information and the ability of real-time processing of information are always contradictory. In the past, it has never been so strong and urgent for the enterprise to have the ability to deal with the massive data in real time when we want to get the necessary information quickly from the massive data. At this time, the inevitable problems of the information system will become an urgent need to solve the major challenges of enterprises. More and more data, more and more complex operations, response time requirements are shorter and slower performance but more and more slowly, the user experience is getting worse ...

In the past few years, the computer processor speed has been almost no breakthrough, but with the CPU processor from the single core to multi-core, from 2 cores, 4 cores, to the 8~10 core, and will soon have a 32~64 kernel of a single processor. By adding the processor's core, the CPU has a stronger, more concurrent computational power. In the new development trend of CPU, traditional software architecture can not effectively control and take full advantage of this new hardware architecture without revolutionary innovation. From the point of view of overall system architecture optimization, it is impossible to bring good investment return ratio to enterprises simply by increasing hardware investment. Here are some of the main reasons.

The potential performance of multiple CPU cores with large memory hardware architectures cannot be fully played.

Data flow is: Disk through memory to the CPU, the bottleneck of data throughput still exists.

Databases only "store" data, and time-consuming computations are still habitually done at the application level.

There is no memory database to support.

Large scale parallel processing computation is not supported.

The problem of combinatorial application of column storage, row storage, memory technology and parallel processing is not solved.

This article is selected from the "SAP memory Calculation--hana", Hu Jian Hidong and so on, published by Tsinghua University Press.

More Wonderful content: http://www.bianceng.cnhttp://www.bianceng.cn/database/storage/