Go to small machine thinking (a)--"software and information Services" 2015.02

In the past few years to the IOE has become the most popular IT industry vocabulary, especially the recent introduction of the rate of localization of some guidance, but also let a lot of people boiling blood, we feel the arrival of the year of nationalization. Many articles have been explored to "O" the difficulties and to "E" high requirements, almost everyone agreed to "I" (that is, the IBM minicomputer replaced with a x86 server, this article called "Go to the small machine") is the most mature, can stand immediately project operation. Is it difficult to go to a small machine?? Let's take a careful look at the process of small machine, call it a small machine thinking.

De-computerization includes two common scenarios: System new and legacy system migrations. For the new system, there is no history of the burden, can fully require the computing resources are all built on the x86 platform, and completely and minicomputer insulation. This kind of go to small machine thinking completely put the difficulty of the problem to the system integrators and ISVs, the user itself is not a great risk, users just need to assess the critical business SLA, to ensure that the final manufacturer submitted the solution to meet their requirements. Therefore, for the new system scenario, only the user needs to change the thinking, from the minicomputer to the small machine thinking almost finished, we do not discuss.

However, the old system migration scene is far more complex than the new system, the need to consider the same factors can not be sloppy. In particular, there are five main aspects: 1. The choice of the timing of the migration and the advantages of consideration; 2. Hardware system migration assessment; 3. Basic software migration assessment; 4. Application software migration assessment; 5. Data migration assessment; 6. Cloud-based assessment options. Below, we analyze the risks and migration points to consider for the small machine.

1. Timing of migration and advantages: for different systems, the timing of the migration varies, but the timing and benefits of the migration can be determined based on the following basic rules.

minicomputer service expiry or EOS: the maintenance cost of a minicomputer is expensive, and if you buy a small machine service that expires immediately, it is certainly a good time for customers to start to think in small ways, and of course, the EOS for hardware services, such as IBM's P6 and previous power CPUs, have terminated the factory service for 2014 years. In general, the operation of the minicomputer business is a key business, no factory back-end support is very dangerous. In order to ensure the security of the system, you must continue to get IBM services, and if you continue to minicomputer thinking, the only option is to upgrade the CPU or hardware system. But this is definitely not a small expense, and the upgrade when the customer often did not negotiate the weight, can only bow to the knife. And this is precisely the customer to small machine thinking start of a good opportunity, because by going to small machine, not only to reduce costs, often can also improve the elasticity of business, double benefit, he le it?

Application System General Upgrade: All kinds of application systems have their own life cycle, with the application of some core components of the system upgrades, often customers will invest heavily. As we all know, many standard basic software suites will have UNIX and Linux x86 versions, and the Linux x86 version tends to be much cheaper than the UNIX minicomputer version price. and the choice of the basic software is not a one-time, but also involves the follow-up service, through the replacement of the backend hardware platform, save the capex not only includes hardware, but also includes considerable software capex. Of course, the savings of opex can be underestimated by saving power and simplifying operations based on the Linux x86 version. Once an insurance user publicly stated that one of his systems migrated to x86, the cost changed to 6%, that is to say, save 94% of the cost. Therefore, only from the cost considerations, the application system upgrade is to go to the small-machine thinking of the golden age;

The innovative development of application systems: Cloud computing, big data, social, intelligent and other trends are constantly subverting the traditional industry, in order to remain invincible, the customer's application system has followed the innovation. So the applications that were launched in previous years, even when they were still in operation, could be a shackle for business innovation, even without the need to upgrade the hardware, and to do so, to revolutionize traditional applications, from the platform to the architecture that might need to be updated. This is also a great opportunity for the x86 platform to enter, not only to replace expensive and bulky small machines, but also to build on an elastic platform, roll out lightweight applications, and faster iterations (DevOps models) to meet the demands of a new competitive environment. A good way to reduce this and increase the efficiency, you must not miss a good time to think about the small machine.

2. Migration evaluation of hardware system: the migration of hardware system involves the server, storage, network and even the wind and fire power in the room, although the components involved are many, but the hardware system itself is not difficult to migrate. The difficulty of hardware system migration is how to evaluate the workload on the existing hardware, how to choose the suitable replaceable hardware platform, including the migration of various hardware components. The main discussion here is the migration of the server, which is the core of the whole thinking of the small machine. But because minicomputer and x86 do not have a simple correspondence, exactly how to determine the appropriate x86 hardware platform needs to be considered comprehensively, not only to consider the hardware itself performance indicators (such as IOPS, TPS, etc.), but also to consider whether the application type is OLAP or OLTP, in the end is i/ o-intensive or compute-intensive, etc. The above-mentioned series of factors to consider, not only need to carry out proof-of-concept, but also based on a well-designed scheme, the old system and the new system in parallel trial operation to finalize the system's ability to ensure that the system security over is absolutely the first consideration of the entire migration plan.

(not to be continued)

Note: Please subscribe to the "China Yunmeng" public number if you would like to know the latest views on cloud computing at @ Cloud Walk. can also scan the following QR code, direct subscription. After subscribing to "China Yunmeng", directly reply "10043" to read the article. This article has been published in the February 2015 issue of the "Software and Information Services" magazine, you can also buy magazines to read this article .

650) this.width=650; "src=" Http://s3.51cto.com/wyfs02/M02/5B/60/wKioL1UH1XfQO7HiAAEOBNi5uT4770.jpg "title=" Qrc.png "alt=" Wkiol1uh1xfqo7hiaaeobni5ut4770.jpg "/>

This article is from the "China Yunmeng" blog, please be sure to keep this source http://virtualman.blog.51cto.com/200540/1621452

Go to small machine thinking (a)--"software and information Services" 2015.02