The game between the x86 and the Big Machine

Over the last 50 years, mainframes have been the foundation of most of the world's largest IT projects, carrying most of the business data and running a large number of core business applications. A variety of technical trends have come and gone, but the mainframe, with its steady opening strategy and constant technological innovation, has been standing still, judging from IDC's data that by the two quarter of 2011 the IBM mainframe had achieved revenue growth for four consecutive quarters and expanded its share-leading position since 2001.

In recent years, IBM mainframe has been expanding new performance, and has introduced new products, so that IBM mainframe innovative performance and unique quality of the spread to more institutions, and will reach out to midsize enterprises and new markets. July 2010, IBM released the world's fastest, most scalable, most energy efficient large-scale enterprise-class mainframe zenterprise system;2011 year July, Zenterprise System added a series of enhancements to further expand the mix of patterns, Includes the ability to run management power and Linux-specific IBM System x server workloads. In addition, IBM's new mainframe Zenterprise System 114 (z114) for midsize enterprises allows businesses to start with smaller configurations (as low as 26 per second million instruction (MIPS)), enabling some additional features built into the server as they grow in the future. Without increasing the complexity and cost of the data center's footprint or system management.

x86 VS Mainframe Giants cannot be shaken

Server market has always been a volatile, from the 50 and 60 's only mainframe server, to 70, 80 's Unix-led, especially with the rise of 2000 x86, based on x86 distributed computing methods are becoming popular. Still, the position of a highly centralized computing system, represented by IBM mainframe, cannot be shaken, the initial acquisition cost of distributed computing is low, often a single application or database running on a server, low resource utilization, high energy consumption and large footprint, while mainframe is highly integrated, the sustainable utilization rate can be close to 100%, Virtualization and load management technologies are more mature, safer, and less energy-consuming and space-occupying. Below we will compare the distributed systems based on x86 and mainframe systems from the three aspects of RAS (reliability, availability, stability), security, and extensibility.

There is no doubt that the high RAS feature is a big selling point for IBM mainframes, and that its products are reliable for a generation. The mainframe system integrates a fairly high degree of redundancy and error checking technology, with redundant design or backup of memory chips, memory buses, I/O channels, power supplies, storage, etc. When the system is upgraded, the component can be hot-swappable while the system is running, even when the microcode upgrade can be done simultaneously while the system is running, thus avoiding the embarrassment of downtime. IBM mainframes are designed for high-volume workloads with up to 99.999% availability, reducing the likelihood of various outages, with an average downtime of up to dozens of years (score time unreported FAILURE,MTBF). Corresponding to this, because the principle of distributed cluster is based on the premise that the single point of failure will definitely occur, even the excellent distributed cluster of architecture design means that the entire system can achieve a maximum RAS characteristics, after all, in theory, any single point of failure can lead to interruption of service. Therefore, for the financial/insurance, Pharmaceutical, aviation/defense business, the distributed cluster, which is based on the RAS features provided by fault tolerant capability, is not able to meet the needs of RAS.

In terms of security, mainframes generally provide the underlying module level of encryption security, and based on the x86 system does not have this natural security, the need for system settings and application configuration to achieve, which requires a very high demand for security operation and maintenance to match. Distributed systems are heard when they encounter security issues, but this is rarely the case for mainframe systems.

As with security, the logical partition (LPAR) virtualization mechanism provided by the mainframe itself makes system expansion very convenient, dividing a single server into multiple logical servers, running separate applications with each other, and ensuring security. It is worth emphasizing that, as early as the 60 's, IBM Applied virtualization technology to mainframe hardware partitions. Mainframe virtualization is more mature than x86 virtualization technology.

(Responsible editor: The good of the Legacy)