New vs. Used Server: Expectations, Risks, and Reality

Not everyone has the budget to purchase new commercial equipment. This is a typical situation for startups, where financial considerations are still early on. Therefore, the question arises of purchasing existing servers.

Is this a viable option? Certainly, if you approach the situation correctly, you can identify potential risks for your organization and clearly understand the purpose for which you need a powerful machine. This is precisely what we propose to discuss in this article.

Defining Goals and Priorities

First, ask yourself: “Why do I need a server?” Ask yourself a simple question: “Why do we need a server?” If this is simply a matter of time, then abandon the idea, because that’s how business is done. If you can’t articulate the tasks a server performs, you probably don’t need one.

Servers should improve your company’s performance and functionality, solve important tasks by storing information, processing large volumes of data, and integrating multiple employees into a single infrastructure. At the same time, 24 hours a day, 7 days a week. If you share this opinion, then the idea is justified.

However, it’s important to understand that servers aren’t entering the distribution market. There are many reasons for this.

• Outdated hardware;
• Low performance for certain tasks;
• Common errors, packet and data loss;
• Minor wear and tear.

Tasks at the other end of the scale. Let’s say the previous owner decided to give the machine away because a pair of Intel Xeon Platinum 8280 processors weren’t enough to meet the business’s needs. However, one such processor will be sufficient for the next two to three years. This is why, when purchasing a server, the first consideration should always be what’s new, what’s currently in use, and the potential load on the equipment. If these conditions aren’t met, you’ll simply be throwing money away.

When choosing equipment, it’s helpful to create a general list of priorities, which looks like this:

• Sufficient performance to solve the task at hand at an affordable price in the future;
• Reliability and overall condition;
• Recovery time in case of force majeure and sudden failures;
• Hardware relevance (technology, supported protocols, throughput);
• Seller reliability and trust;
• Price;
• Ease of purchase and potential logistical issues.

As you can see, price is not the first parameter that you should pay attention to. Either way, you’ll save money, but is the cost difference justified?

Let’s talk about functionality

Take the following paradigm as a basis: the potential server fully meets the performance requirements, taking into account possible upgrades. For example, a motherboard supports two sockets, an additional RAM slot, a controller, and PCI peripherals. And keep up the good work.

Server reliability

Have you thought about how servers can enter the distribution market? Theoretically, when upgrading equipment in large data centers. At least that’s what almost all the initial advertising claims. However, let’s be realistic. This possibility is very small.

Most “recycled” iron is stolen, semi-working, and even dead iron, and there are direct routes to recycling. And since this process is expensive, the previous owner is trying to implement the server under the motto “not to lose, not beautiful, one owner.” You can make a formal request to the manufacturer, but chances are they’ll ask, “Where did you get that from?” This is the most common objection to the series.

In the second case, the hardware is stalled by time and load, which can overload test runs and some simple applications. The reseller wipes the dust off the server, changes a few parts, brings the product into proper shape and then sells it, adding a personal margin on top. This approach raises questions rather than expectations.

Even legitimate equipment sold as a result of bankruptcy, natural obsolescence, or simple computational errors should be thoroughly inspected and tested. It is also recommended to check the previous state of the server, which may include

• Load intensity;
• Compliance of the facility with the operational requirements of the server room;
• Quality of ventilation;
• Service frequency;
• Staff literacy;
• Frequency of warranty cleaning and inspections.

According to GOST, was everything in the server room or was an unused basement allocated for it? Was air quality and temperature monitored? Has attention been paid to dust protection? Frequency of maintenance of cooling systems, for example. A competent administrator can easily see where there are “tricks”, if any.

Go back to the “data center” where the server operated several years ago. Why did you decide to replace it? How many units were on sale? How successful was this batch in terms of service and performance? Only the seller can answer these questions for you, and they’re not interested in discussing any issues with the equipment. Therefore, if you buy equipment without independent testing, you’re buying a pig in a basket. If you haven’t taken something into account or verified something, find out whether the service (if provided under the contract) is being provided by the same seller who sold you the equipment. They promise one thing on paper, but in reality, they deliver something completely different. Filing a lawsuit can help you protect your rights. But here the question arises: is it worth the candle?

From all of the above, the final point logically follows: warranties. Leading manufacturers, such as HPE, offer an official three-year warranty on their products. This includes a specialist, restoration of functionality, and replacement of components in the event of a malfunction. With used servers, the “warranty” is the seller’s responsibility. If, when signing the contract, you’re told they’ll give you a month to check it, it’s clear there’s no warranty involved.

Recovery and startup time in the event of a failure

Let’s assume you’re outdoors, there are no hidden obstacles, and the equipment is powered. Then the second question arises: How easy is maintenance?

According to expert experience, the seller’s wording should be interpreted as follows:

• Specialized equipment—finding spare parts is practically impossible.
• So, just solder a bridge, and everything will be fine. The board is clearly overheating and will die under load.
• The servers are fully functional, and you’re just starting out. They’re selling you potential corpses.
• The model is great, but not for us. The servers are clearly old, and we can’t find new parts.
• We take two and assemble one of them—both models fail the same way (really).

So, carefully consider the pros and cons of a particular model. The more popular the device, the more information about it will be on the forum.

Hardware Relevance

We’re all realists, and frankly, no one wants to buy an old device. So, immediately pay attention to whether the interface, socket, or chipset line bandwidth is sufficient.

Minimum threshold values ​​even for complex tasks:.

• Intel Scalable 1Gen/ AMD EPYC 1 Gen;
• DDR4 ECC/DDR3 ECC (high frequency);
• PCI-E 3. 0;
• Current drive support.
• Ethernet 1 Gigabit.

Such a solution can function for about two to three years until funds become available to purchase new, modern equipment. Please note that server performance improves by about 40-50% every two years. This takes into account not only raw power, but also useful work per 1 kW, so these boxes consume energy accordingly.

Servers based on the third generation of Intel and AMD processors are currently entering the market in large numbers. And it is more likely that vehicles will be accessed from the same data centers.

Let’s sum it up

The following terms are intentionally placed in parentheses

• Level of trust in the seller/organization;
• Equipment costs;
• Convenience of transport and distance.

This is not included in the material topics as these items are of equal importance when purchasing new or used equipment.

Our task today is to help you understand the priorities and functions of commercial vehicles, which have long been used to the maximum to solve enterprise problems. Despite the high performance of the servers, they are heavily overloaded. In addition, they are often exposed to conditions that are far from recommended.

Is the investment in a new server worth the risk? Decide for yourself. However, if you find that the costs are inadequate, a NORD-SERVER specialist will check whether you have what you need used server available market. Marvel, the company will always support you by offering the necessary model from the current line. They will focus on your organization and select the appropriate option in terms of price/quality. Specify the server type (mail, terminal, database) to get the most out of it. By cooperating with us, you receive high-quality equipment with an official guarantee and service support.