-
Step into Capney
Data center renovation should start from six key areas.
Release time:
2017-10-16
Recently, some engineers who frequently work on-site at data centers have begun to harbor significant doubts about whether the technologies employed in data centers (DCs) will continue to improve. The “transformation” of data centers referred to in this article actually pertains to ensuring that more appropriate models are applied within data centers. Although engineers each have their own working styles, this “transformation” holds practical significance whether they occasionally visit data center sites or have a fixed schedule of on-site work. Of course, we’re now leveraging remote modes to handle many aspects of our work. However, in actual projects...
Recently, some engineers who frequently work on-site at data centers have begun to harbor significant doubts about whether the technologies employed in data centers (DCs) will continue to improve. The “change” referred to in this article actually pertains to ensuring that more appropriate models are applied to data centers. Although engineers each have their own working styles, this “change” is highly relevant—whether for those who occasionally visit data center sites or for those whose on-site work follows a fixed schedule.
Of course, we now rely on remote modes to handle many aspects of our work. However, in actual projects, there are still quite a few tasks that simply can’t be completed via remote desktop access. For example, we often need to replace hardware, visually inspect the wiring conditions, reinstall blade servers, plug in and adjust equipment components, and occasionally encounter seemingly baffling operating-system installation procedures. Although most engineers don’t mind traveling to the data center in person to carry out these tasks firsthand, sometimes we still can’t help but wonder: Just who exactly designed these so-called “standard” components? And when they were designed, did they ever consider that human hands, fingers, and eyes all have inherent limitations?
After careful consideration of these issues, this article provides corresponding solutions for readers’ reference. By “standard” components here, we refer to the physical architecture encompassing 19-inch racks, hot and cold aisles, server entry and exit points, cabling and cable management systems, power connectors, KVM systems, lighting, and nearly every other aspect of the hardware infrastructure. Based on the observations above, if we were to evaluate the design from a human-centered perspective, virtually all of these elements would need to be completely rethought and redesigned. The only component that might not require significant improvement is the rather unassuming baseplate. Although the overall performance of this architecture appears remarkably impressive, we cannot assume that a qualified baseplate automatically ensures that the entire system is equally functional and reliable.
For convenience, this article lists the specifics of all components one by one and informs readers of potential problem areas as well as their corresponding solutions.
Rack structure
Although the 19-inch standard is an inherent, unchangeable foundational component, has no one ever considered making adjustments to the internal structure of the rack itself—adjustments that would make the actual installation of standard-rack systems (such as 1U, 2U, 4U, and so forth) significantly less daunting? For instance, to achieve the goal of directly routing power, network, SAN, and other cables into the rack, we could mount connectors along the edges and corners. That way, when we open the cabinet door, we’d no longer have to endure the harsh ordeal of being immediately confronted by a bewildering tangle of cables. Since the constraints imposed by the “cable racks” make specific operations involving connection systems extremely difficult, a more appropriate improvement would be to adopt plug-in connectors instead.
In addition, to ensure that staff can clearly see the operations they are performing while working, the designers need to install several LED lights for illumination inside the cabinet. These lights could adopt a refrigerator-style lighting mode—turning on immediately when the door is opened and turning off promptly when the door is closed. Since ceiling lights would cast shadows onto the top-mounted systems, they are not a suitable choice. To guarantee that every part of the cabinet—from top to bottom and from inside to outside—is fully visible, the lighting fixtures should be installed vertically along the inner edges at the front and rear of the rack.
Cooling system
Although such a design can theoretically function effectively, for on-site personnel, it means they’re essentially limited to working either in the hot aisle or the cold aisle. Clearly, no matter which side they choose, their actual experience won’t be very pleasant. To some extent, the real effectiveness of the air-exchange mode will far exceed that of the air-cooling mode. In other words, using flowing air to carry away the heat generated by the racks is a much more efficient cooling approach. In the initial design stage, the racks shouldn’t have been equipped with inefficient heat sinks in the first place—such heat sinks only serve to trap the heat inside the system and in the surrounding environment. Moreover, compared to vertical fans, horizontal fans are capable of removing significantly more heat.
To enhance the cooling performance of the fan, we should also shorten the length of the air duct, thereby preventing the situation where the fan can only effectively cool one end while leaving the other end inadequately covered. To put it more simply, we should establish an “air river”—in this way, cool air can continuously flow in through the fan, while hot air can travel along the duct to the exhaust fan.
Cabinet renovation
As for rack systems, a particularly practical design incorporates features that facilitate the movement of the upper unit. After all, installing one system atop another server is an extremely cumbersome task. Given how difficult the entire moving process can be, it often takes considerable time to carefully and gradually make the necessary adjustments. To begin with, the first step is to quickly loosen the hand-tightened screws or pins and pull them out from inside the rack. Next, to make it easier to remove the top cover and perform essential maintenance, you’ll also need to slide the system out of the rack almost entirely. Moreover, after pulling the server out from the front of the cabinet, you’ll have to go around to the back just to disconnect the power supply. If the rack is equipped with quick-connect/disconnect modules located at the rear or side, moving the system within the rack becomes much more convenient—allowing you to complete all related tasks in a single, seamless move.
Moreover, wouldn't it be a feasible design to reposition all components that need to be replaced into more conveniently accessible locations within the hardware device itself? Just imagine—how wonderful it would be if we could simply remove the hard drive from the front end of the system! Right now, however, accomplishing this task requires painstakingly disconnecting the memory, solid-state drives, and even the central processing unit—all of which stand in the way.
Noise reduction measures
Silent fans have already emerged. Unfortunately, however, they haven't yet gained widespread adoption. Yet the noise emitted by data centers is so loud that staff members must shout loudly just to communicate with one another. In this regard, we have good reason to believe that both racks and servers should opt for silent fans.
Security
What other places could possibly be safer than data centers? The answer probably comes down to banks—or places where large amounts of cash transactions take place. Although cash transactions typically don't occur inside data centers, it’s still essential to implement security measures at a professional level.
When entering the data center or internal work areas, staff members should wear a work badge. The badge will contain the actual serial numbers of the equipment they need to access, as well as detailed descriptions of their specific job tasks. This setup ensures that we can only open the racks that are necessary for the task at hand. Once the system detects that the badge has entered the work area, the corresponding racks in the relevant systems will be automatically unlocked. This approach significantly enhances security, preventing unauthorized modifications, disruptions, or even damage to the systems caused by accidental incidents.
In the real world, overly similar system names, incorrect rack placements, and careless mistakes are all significant causes of unexpected outages in physical servers. Therefore, only by deploying appropriate and well-placed security measures can data centers achieve the goal of ensuring that visitors access only the specific systems they’re authorized to handle. If errors in the badge information lead to inaccuracies in location assignments, technicians will need to re-verify both the server names and rack locations.
For data center managers, this model can also provide real-time, detailed information on the number of personnel inside the facility, their exact locations, the tasks they’re currently performing, and the reasons behind those actions. Currently, however, managers have no way of knowing which systems are being accessed, why they’re being accessed, or whether the necessary authorization for such work has been granted. With this approach, once technicians enter a specific maintenance area, a corresponding indicator light will illuminate, immediately notifying security personnel and management that work is in progress. Thus, we believe this represents an important step forward for data centers as they strive toward the goal of zero unplanned downtime.
Hardware Audit Mechanism
From a personal perspective, to prevent emergencies that might require the use of standard computer hardware, all data centers should stock essential emergency supplies for their technical staff. Technical personnel should also maintain a comprehensive inventory of mobile phones, laptops, rescue vehicles, toolkits, and other equipment necessary for physical maintenance. All external computing devices should be prohibited from accessing the data center’s internal network. Any violation of this rule would pose a significant security risk to the data center. As is well known, external hardware often serves as a major entry point for viruses, and the security risks associated with videos and photos, as well as equipment malfunctions caused by tools, are both highly dangerous and extremely difficult to effectively control.
In fact, our goal isn't to make the actual work in data centers more difficult; rather, we aim to make the entire operational process smoother and easier. This is precisely the underlying rationale behind the solutions outlined above—aiming to maximize efficiency. For example, imagine you’ve just arrived at the data center to fix a system, only to find that the system running next to the rack has just experienced a sudden outage. The likely consequence would be that others might immediately blame you as the primary culprit for the problem. However, by properly planning your data center according to the strategies mentioned earlier, you can avoid such embarrassing situations altogether.
Keywords:
Previous page
Previous page:
Recommended News
2018-08-28
Provide consumers with more solutions.
Power by : 300.cn Beijing ｜ CertificateQuery ｜ File information inquiry ｜ Anti-counterfeiting query