Chapter 30 Developing Blade Servers
After reading the newspaper and seeing that the people in the company had arrived, Ling Shizhe announced the meeting, and he first said: "I called everyone here today to discuss the RPM+1 processor. As you all know, the RPM+1 processor has not been sold since it was introduced two months ago, and despite its excellent quality, those computer companies are reluctant to use our processors. The reason is that companies have their own processors, and they don't want to develop new computers on our RPM+1 processors. In addition, you have all been to see Ling Shishi Ranch, the whole ranch is the world's first fully computerized ranch, the computers used in it are either IBM or DEC, these computers are expensive, inefficient, power consumption is astronomical, and it is not particularly reliable, should we develop a new computer, it is powerful, reliable, and consumes as little power as possible, it is best to do only one-third of the current one-third, it is the best, this is my idea, I want to hear your opinions. ”
Ling Shizhe as the boss is not an arbitrary person, some major things he will listen to everyone's opinions through meetings, in order to make better decisions, this method he learned from the company he served in his previous life, encounter some major problems to meet and discuss, collect everyone's opinions, can effectively make the most beneficial policy for the company, so as to avoid major losses, this is the method commonly used in the previous republic, this method is also available in the West, that is, the company's board meeting, But now that the company is owned by Ling Shizhe alone, the board meeting is not good, and the enterprise development conference is just right for a meeting with the characteristics of the republic.
"Boss, do you have a plan?"
Ling Shizhe nodded to Li Li, who stood up and distributed the information in her hand to everyone. What they showed them was that Ling Shizhe according to the later server technology scheme, now there is no server at all, all of them are computers, or very honest machines, not to mention the trouble to use, and it is quite inconvenient in maintenance, maintenance and management. In this era, large and medium-sized computers cannot be called servers, and there is no RASUM system standard for later generations: r:reliability; a: availability; s: scalability; u: usability ease of use; m: manageability, i.e., the Rasum metric of a mainframe computer or server.
R: Reliability (reliability) needless to say, if the big central computer is not reliable and often has problems, it is simply a disaster means that it cannot be used, look at the emergence of the world's first tube computer, every time it is turned on, each tube will light up like a big bulb, half of the city will be without power, the computer engineer has to find a damaged tube from the computer room every fifteen minutes to replace it, replace it with a new tube, you must know that the computer weighs 30 tons, It covers an area of 150 square meters and has 18,800 tubes in its belly. Not only was the reliability poor, but the energy consumption was surprisingly high, and computers at that time could not be used by the civilian sector except for laboratories and the military. Reliability is the number one criterion for a mainframe computer or server, you may not be the most advanced, but you must be the most reliable.
A: Availability means that the mainframe computer or server must meet the requirements of long-term stable operation and cannot have frequent problems. It is equivalent to the reliability proposed by the later SUN company. It's not for individual users, it's for medium to large enterprises, and often requires servers to be uninterrupted. In some special applications, some servers have to work uninterrupted, even when they are not being used by users, because they must continuously connect to users, whether they are on or off work, whether they are on weekdays, or on holidays. This is the fundamental reason why servers must be extremely stable. For these users, the server has to work 7x24 hours uninterrupted, maybe the number of times it really works and starts is only once, that is, it has just been bought back and put into official use after being fully installed and configured, and since then, it has been working uninterrupted until it is completely scrapped. If something goes wrong at every turn, it won't be possible to maintain normal operation for a long time. In order to ensure that the server has a high "availability", in addition to the quality of each component, necessary technical and configuration measures can be taken, such as hardware redundancy, online diagnosis, etc.
U: Usability, today's computers are large or medium-sized computers, even if the PDP-8 minicomputer launched by DEC company is many times more complex than the PC computers we used in later generations, its structure is more complex than the servers of later generations, and the functions are quite few, to achieve so many functions, it is not possible without comprehensive software support. However, if there are too many software systems, the performance of the server may be reduced, and the management personnel cannot effectively control it. The ease of use is mainly reflected in whether the server is easy to operate, whether the user navigation system is perfect, whether the chassis design is user-friendly, whether there are key recovery functions, whether there is operating system backup, and whether there is sufficient training support.
s: scalability, the server must have a certain degree of "scalability", because the enterprise network cannot be unchanged for a long time, especially in the future network information age. If the server does not have a certain scalability, when the number of users increases, it will not be competent, and a server worth tens of thousands, or even hundreds of thousands, will be eliminated in a short period of time, which is unbearable for any enterprise. Scalability is reflected in whether the hard disk can be expanded, whether the CPU can be upgraded or expanded, whether the disk array shelf, PCI and memory module slots, and whether the system supports a variety of mainstream operating systems on the market.
m: manageability, an important characteristic of a large or medium-sized computer or server. Although we say that servers need to work continuously without interruption, even the best products can fail, and as people often say: not that they don't know that it can be broken, but that they don't know when it will be broken. Although the server is guaranteed in terms of stability, it should also have the necessary measures to avoid errors, detect problems in a timely manner, and maintain them in a timely manner if they fail. This not only reduces the chance of server errors, but also greatly improves the efficiency of server maintenance. In the case of Shishi Ranch, if the mainframe fails, it is impossible to fix it for maintenance like the personal computers of later generations, because it is simply impossible, which will paralyze the entire ranch network system, and the losses caused by it cannot be measured in money. Although IBM and DEC have provided the most advanced technology for this problem, it is still very primitive in the eyes of Ling Shizhe, a traverser, and although there are many mature methods to deal with this problem in later generations, it is still a problem that cannot be ignored. The manageability of the computer is also reflected in whether there is an intelligent management system, whether there is an automatic alarm function, whether there is an independent and systematic management system, whether there is a monitor, etc. Only in this way can administrators manage easily and work efficiently.
Now compare IBM's mainframe and medium-sized minicomputers with the Rasum standard, it is simply a scum and a product of the Stone Age, in order to digitize the ranch Ling Shizhe and the team he led, I don't know how many brain cells were killed to barely reach it. This so-called digitalization is still a new concept for people now, and people don't think it's very high-tech because they don't compare it, but from the perspective of people in later generations, this so-called digital ranch is completely a hype gimmick, not a digital ranch in the real sense.
After more than half an hour, everyone finished reading the information in Ling Shizhe's hand, and for the time being, no one came out to speak and discussed in a whisper, oh, oh, Alison knocked on the table and said: "Okay, don't talk to each other, this is a meeting, everyone has something to say, let's discuss it together." ”
"Boss, the computers used in the ranch are all produced by IBM and DEC, and the results are not bad, and we have received 50 orders for ranch renovation, if we are making such a computer with similar functions, will those ranchers spend money to buy it?" An engineer asks. yes, that's a very real problem, and they're definitely not going to spend money on it.
"We can not do those orders at present, the system is not mature now, it still needs time to observe, there are many problems that need to be improved, we do not provide such services for the time being, anyway, these orders are still only intentional contracts, it will take a lot of time to turn it into a formal contract, this time we are enough to make our new computer, and it will not be a problem to sell them at that time." Olson of the company's legal department will answer everyone's doubts.
Ling Shizhe asked: "You have all read the information, I have a question to ask, that is, is it possible to implement my blade server solution with the current hardware environment?" I don't want anything like the EUC to happen. ”
"No, boss, with our current technology, it can be done, this kind of blade server can be completely realized." Everyone can't help but think of a mistake made when the company was just established, when everyone was confused by Ling Shizhe's sci-fi EUC (driving computer) technology, only to see his good, and ignore the hardware conditions at that time, as a result, everyone found that the current technical conditions are not realistic at all, although it was finally made, the volume is not large, but the function is too weak, after all, the CPU used on it is only 8 bits, too reluctant, it can't meet the needs of commercialization, but it solves the problem of whether there is or not. What about a 16-bit CPU? It's not impossible, but with the current chip integration, even if the 16-bit CPU is made, it only solves the problem of whether it has or not, and it cannot meet the needs of commercialization, and its use effect is not as good as the use of 8-bit CPU. At present, this ECU can only be used as the company's technical reserve, and it must completely meet the requirements of commercial use, unless the performance of the CPU reaches the level of 386 or above.
"The key is how we can integrate the RPM+1 processor and the memory and application components on a single board, without being too big or too thick, and delivering the performance to the Rasum standard." Nien added. Seeing that everyone was not opposed to the server project, the next topic was how to divide the labor, and finally divided into the hardware group, Nien was responsible for the development of hardware, the system group, and Ollie was responsible for the development of the operating system and the application group, specializing in the development of various application software. After the meeting, everyone dismissed, and the leaders of each group returned to the laboratory to gather their members for further discussions. Ling Shizhe returned to his office, and was putting the finishing touches on the servers he had designed and the new digital transformation plan for the ranch to ensure that there were no loopholes.
The MPM+1 processor is a new type of processor based on the ARM core, which follows the PO of IBM in the future and the CPU of the traditional architecture CISC used by DEC are two diametrically opposed design ideas. PO+1 processor (no way,Continuing to use the real name will touch the trouble of copyright,Only change),R stands for RISC reduced instruction set computer,P stands for the power processor of IBM in the future,M stands for MIPS,When Ling Shizhe designed RPS+1, he also referred to the design experience of MIPS,This new PO company in the traditional sense of PO+1 just includes the first letter of the name of the above several processors,1 represents the first generation, In the future, with the progress of integrated circuits, follow-up upgrade products such as RPM+2 and 3456 will be launched.
Because this RPM+1 chip is diametrically opposed to the traditional chip design idea, it is born to face this embarrassing situation that no one cares about, and once those traditional architecture computer companies use RPM+1 processors, it means that the previously designed computers and various software developed are all abandoned, and it is impossible to redesign and develop new application software for him. Although RPM+1 is excellent, it is several times faster than traditional chips with the same 10-micron process.
No one specially designed the RISC architecture computer for RPM+1Z, and the computer used by the digital ranch made him very dissatisfied, so taking advantage of this incident, Ling Shizhe simply designed the RISC architecture server by himself. The server he designed is the later generation of the knife frame server, this kind of server was born in 2001, is a complete server integrated on a board, on a plate it is highly integrated processor, memory, hard disk, network interface and control chip. A card is a complete computer. Compared to traditional servers, blade servers are more like a fashionable consumer product: servers are designed in the shape of blades that can be neatly inserted into specific racks like books on a shelf.
They are plugged side-by-side on the backplane of the blade server chassis and share the same power supply, fan, mouse, keyboard, and display equipment, which not only takes up less space than traditional servers, but also removes the power consumption components in the rack server, reducing the heat generated by the system power supply. Due to the limitations of the hardware technology of this era, Ling Shizhe also added a GPU (Graphic ProcessingUnit) graphics processor to it, what is the use of adding this to the computer board? There is only one purpose, to achieve the purpose of a visual graphics operating system, otherwise in this era, a CPU will not be able to achieve a visual operating system at all.
When people see computers on the ranch using a graphical operating system, it's exciting to think about.