The advantages and disadvantages of the fine blade server

What is the difference between a blade server and a rack server?

The blade server trend started about a decade ago when RLX introduced its blade system, which was built into the chassis and inserted into a standard 19-inch cabinet.

The purpose of this design is that you can install a blade server or any other type of device that can be loaded into a server chassis, such as a management blade, and then it will automatically connect to the network and other services.

The blade server consists of a large motherboard with a chunky connector on the back; in addition to the processor and memory, there are several onboard components. You may find one or two USB ports for connecting a keyboard and mouse, but like any other server, the blade server is more likely to use the KVM system to manage it directly over the network.

Ultra-thin chassis

The blade server is designed to be easily interchangeable and can be plugged into the backplane; the backplane is part of the chassis, usually a passive bus, responsible for the input of the server and the back of the chassis. /Connection between output subsystems.

The chassis provides all the common services, such as cooling, power and connections, which are usually provided by replaceable modules, which makes the chassis suitable for the type of tasks the server handles.

Sharing services in this way makes each server more energy efficient. They are also easy to manage, with a single management platform.

Having said that, each blade may include one or two hard drives for booting, but it's wise to boot from a generic image on a storage area network (SAN): the more proprietary something on a blade Much more, the less you get from the blade concept.

The blade server also generates extra heat and adds power and complexity. Even so, you may need some local storage system - depending on the type of application, you may need a DVD drive to install the software.

One of the great advantages of blade servers is that they are very dense: they have a small footprint and they can get a lot of computing power.

Because several servers can be loaded into one chassis (HP's c-Class BladeCenter chassis can hold up to 16 servers), the cable to be dropped is significantly reduced. Now, as long as a set of lines can handle the input/output functions that previously required 16 lines to process, it may not be as simple as it requires redundant components and additional inputs/outputs.

Another advantage of blade servers is that they are usually hot-swappable: if a blade fails, you can pull it out and replace it with another blade.

In a fully managed, automated data center, the system is able to configure new blades so that the new blades are logically identical to the old ones, with the same MAC and IP addresses, operating systems, and applications.

Almost every removable part inside the chassis can be replaced during system operation. Hot swapping means higher reliability: if one system fails, another system can be on top of it.

Data center administrators can replace failed parts when they are idle, rather than knowing that the service has just stopped and the end user has complained about the haste.

Because you deserve to have

originally proposed, the blade is considered suitable for use in dense web server clusters to provide web services, but their functionality has since expanded a lot, and now can perform almost any Server task.

Having said that, most blades may still be running a single application for their interchangeability, but the trend toward consolidation through virtualization seems to change this.

Blades offer the advantages of ease of management, ease of deployment, and space savings. You can get this convenience by spending more money; but for many people, it's worth the money.

Some of the advantages claimed by the blade are precisely its shortcomings. One of the biggest drawbacks is that once you buy a blade chassis, each blade must come from the same vendor.

In fact, companies can only buy a complete computing system from a certain manufacturer, and there is no choice at all. Your hands and feet are firmly tied by this manufacturer.

It is also difficult to get out. Without a standard blade server chassis, it seems unlikely that a standard blade chassis will emerge because vendors lack the incentive to produce standard blade chassis.

Chengdu Density, Density and Density

This means that one manufacturer’s products cannot be loaded into another manufacturer’s products, which is a standardization and commodity that has significantly reduced the cost of technology over the past 30 years. The trend is running in the opposite direction.

So, although a single blade server seems to be less expensive than a rack-mounted server, if you count the cost of the chassis and general-purpose services, then each server will cost more than the equivalent stand-alone server system; Moreover, users have no choice.

In particular, because the blade is only filled with the chassis, the blade is economical: spread the cost of the chassis and other components to as many blade heads as possible.

The blade is a highly intensive system that saves costly space, but in fact it is a major drawback and may be the biggest drawback.

It is usually limited how many blades can be inserted into the cabinet, not how many blades can be installed in one chassis, or how many chassis can be loaded into the cabinet, but the power consumption and heat dissipation of the blades.

It's too hot to handle

If the density of the loaded blade exceeds a certain value - usually expressed in kilowatts /cabinet, the cooling system of the data center may not be able to handle it, or consume The power may exceed the power supply of the local power company.

It is especially important to have many high-performance servers installed in one device so that the chassis cooling system can operate normally. If you need to install a cooling device, or ask the power supplier to pull additional power lines, it also means additional costs.

Some companies even go back and use separate rack-mounted servers because the cost of the new cooling units and the space they occupy offset the advantages of the blades.

Blades are not suitable for every application environment, especially those that require a lot of processors and memory. Such tasks are often transaction-intensive tasks or attributed to virtualization technology because it combines multiple servers into one device.

Pros and Cons There are disadvantages

Blades usually have no place to place the extra hardware needed for such applications. It turns out that transaction-intensive applications have not moved from high-performance systems to virtual machines, and many organizations' IT departments still believe that high-performance systems are the best choice for running transaction-intensive applications.

So the blade server brings not all the benefits. They have advantages and they also have a place in the enterprise's IT system. However, buyers should pay attention: the cost may be higher than the cost originally thought. They must make sure that all aspects have been taken into account before they can afford to buy a blade server.