Servers are a group of computers that are designed for heavy-duty so that several people can serve them at the same time. And physically, they are designed in such a way that we can put them in special compartments (Rack). To provide better conditions for them in terms of safety and appropriate physical conditions (temperature and humidity) and ease of access. This category of computers is also the same as: other workstations are completely different from ordinary computers. And moreover, the following are also the main differences between them and workstations:
The difference between the server and other workstations
The server hardware is designed to be redundant. That is, in the case of hardware that mostly has a higher failure rate than other hardwares such as: power and external memory (HDD, SSD), more than one number are embedded in them in such a way that they are in the circuit simultaneously. And with the failure of one of them, in addition to not losing information, the server does not even shut down and the defective part can be replaced while the server is on and working (Hot Plug).
Keeping the physical security of information in servers is very important, so that a separate piece of hardware called Raid Controller is embedded in them. which can apply and manage various arrangements to create redundancy and more speed and productivity in the case of peripheral memories.
(For example: by choosing 4 disks and establishing a standard number (10 Raid) arrangement between them, in addition to achieving 2 times the speed of information exchange, this arrangement can withstand the failure of 1 or 2 memories, so that information is not lost or shut down.) Of course, note that all workstations and some regular computers also have this feature. with the difference that due to the lack of special hardware, this is done mainly in the form of software in them. Or that piece of hardware doesn’t have the same capabilities as servers. Indeed, this piece of hardware with PCI format can be installed on the motherboard and use their capabilities in workstations but servers have this piece of hardware by default.
The appearance of the servers is designed in such a way that all of them can be installed in large numbers in standard racks. And the air conditioning of them is designed according to this compartment in such a way that it absorbs cool air through powerful and numerous fans from the front line. And this air flow on its way to the end of the server cools all the required parts in an optimal way by the special fins installed inside and exits from the end of the server.
Smart middleware is embedded in the servers, through which BIOS access can be obtained in the computers when the server is on and working through the network. With this access, all necessary information related to the health status of all hardware parts can be viewed remotely. And some hardware settings are done without the need to shut down or restart the server. And also, the image of the server is recognized in this way, so that there is no need to connect a monitor directly to the server and we can perceive this image through the browser of a computer that is on the network.
Expandability is also a characteristic that has been given special attention in servers so that various hardware parts have been produced for every type of needs, which can be installed on the server. For example, in a situation where we need a large number of hard drives, more than 20 hard drives can be installed on a regular server. Or when we need a large number of network ports, many Network interface card can be installed on a server.
Long-term support on servers is very important. That is, the server manufacturing companies offer software updates for a longer period of time due to the fact that these devices are expensive and designed to last longer than other types of computers.
rackmount server or a desktop server (tower)?
Servers are produced in one of the types of rackmount or tower and the main question in most cases is, which types of servers is more suitable for you?
If you have a standard rack (with a depth of 100 cm) in your company, it is logical and optimal for your server to be of rack type So that you can provide more ventilation and security while installing it in the rack space.
And you can make the most of its vicinity to other rack components. But if you don’t have a rack, you can use its tower shape. But not all the differences are in the appearance and location of the server
also the following items are among the main differences between these two categories.
the air circulation in tower servers is the same as desktop computers, while the air circulation of rackmount servers is different as mentioned above.
Therefore, you can use all kinds of desktop series graphic cards in tower servers, while in rackmount servers, you cannot use any type of graphic card due to the limited space of the graphic card.
There are fewer expansion riser cards available in tower servers and the number of these PCIs is more in rackmount servers.
Usually, in terms of hardware, there are fewer modules for installing RAM in tower servers compared to the same rackmount models.
Rackmount servers usually have more ports for hardware development than Tower.
Which brand of server is better for you to purchase?
In short, only HP brand servers are suitable in Iran Why? We will explain further
- The highest performance among the brands belongs to the DELL. And despite the fact that the market share (according to the 2021 statistics) for this brand is slightly higher than the HP in all over the world. (The first place belongs to Dell and the second place in the market share belongs to the HP.) HP owns the market in Iran. The reason may have been the low price of HP and the nature of Dell company. which is an American semi-public company. And Iran-US relations have always faced obstacle and are affected by this communication.
- The performance difference between the two mentioned brands is not as much as the price difference, which means that HP is more affordable.
- DELL licenses cost is much higher than HP, and in most cases, there is no need to purchase HP licenses.
- In Iran, HP equipment are abundant and on the contrary, they are very limited for Dell.
- Other brands such as: cisco/IBM/ASUS and… not only do not have a price advantage compared to HP and mainly with a stable specification, the lowest price belongs to HP, but also in terms of performance, HP works in a different way compared to other brands.
- In other cases, such as: build quality/longevity /expansion modules/equipments/spare parts, etc., the HP is the unrivaled vanguard in Iran.
What is the difference between server generations?
The server standards are chosen much more strictly. And they will be updated downstream. For example: At the time of writing this article DDR5 standard is used in PCs. But for servers, not only has it not entered the market, but also the timing of its entry is obscure. in a nutshell, a server generation becomes outdated after personal computers generation. And usually, each server’s generation is in the market for about 5 years, but this number is less than 3 years in regular computers.
Server application are different from computers. And servers are prepared with specific goals, and maybe for that purpose for which you intend to buy a server, the server generation does not have a noticeable application and they are not affordable for you.
Usually, there are only new up-to-date servers in the market, so if you only intend to buy a new server and you cannot buy a stock server, your choice is limited to the latest generation. It is rare to find a new server from a lower generation in the market and it is high priced, which is not economical compared to buying the last generation server.)
For example, if your goal in preparing a server is to use it as an accounting server, the 8th, 9th, and 10th generations have no significant difference in performance for this type of application. And you can invest in number of servers instead of generation. That is, with a certain budget, instead of buying a 10th generation server, you can buy two 9th generation servers. And put these two in the circuit at the same time to support each other.
But if your goal is to launch VDI service through server, the 10th generation server performance will definitely be very remarkable. Therefore, the most important parameter in determining the server generation is your desired application.
servers parts and equipments
Server processors have more cores and memory (Cache) compared to personal computer processors and we know its function.
The main memory in servers has a special hardware part (chip) for error correction called ECC. (Error Correction Code) which is the reason for the presence of this important and expensive piece to prevent errors in the main memory. To prevent system hanging. This memory can be used in servers in large numbers and volumes. And in addition, it has various technologies with special applications such as: Persistent.
In addition to SATA which is widely used in personal computers, External storage (SSD, HDD, etc.) in servers supports the SAS standard. which provides us with much more bandwidth and simultaneous connections on one cable And of course, the SAS standard has more features compared to the SATA.
As mentioned above, this hardware memory (Raid Controller) is directly responsible for managing the arrangement of peripheral memories in servers which is very important.
This battery is able to complete the calculations that are being done in the controller during a power outage and not to cause data loss.
At least two power supplies are installed in the servers, so that if one of the power supplies has faced a problem for any reason, the other one will be in the circuit without shutdown. And the power won’t turn off, the server’s power is much smarter than regular computers and we can see its health status through iLO.
Risers are devices similar to the PCI port that exists in different types for servers’ hardware development so that you can put the hardware you need such as network card, memory controller card, graphic card and other cards on them.
It is the body of the server on which all parts are installed.
Like the motherboard in a personal computer, it is the main board that holds all communications.
It is the free space in the servers that can be used to place peripheral memories in different formats (SFF, LFF). These spaces are located both in front of the server and behind it.
This part is the special memory magazine through which the memory is placed in the bay in the form of a magazine.
In rackmount servers, all the air circulation lies with the server’s own high-speed and smart fans. And other parts do not have fans (in regular computers, each part have its own fan for optimal ventilation, such as the graphic card, power, etc.), so the parts that need to be cooled have aluminum sinks instead of fans. This sink performs the cooling operation by being in the way of the air created by the main fans. processors in servers do not have fans and only have sinks.
According to the above description, there are plastic vanes in the servers to direct the air flow straight to the sinks and components that generate heat.
Rackmount servers have built-in rails in order to facilitate the installation and inspection of the server so that we can take it away from rack like a drawer.
When we intend to slide the server out of the rack, so that the cables are not cut and the server is not removed, a metal arm is installed behind it which opens when the server is removed. and guides the cables connected to the back of the server.
This piece is placed on the forehead of the server for physical protection and delicacy. In order to integrate the appearance of the server with the lock to prevent the removal of external memories.
Of course, these are not all the hardware parts of a server, but as mentioned, they are the most important ones and are among the items that must be consciously selected and installed when purchasing one. Now that you are familiar with their nature, you can get free consultation from Rahaco experts to choose the exact type and specifications of each one.
- Most importantly, Raha does not provide servers for sale in any of its stores and offers all its servers to its service customers. Therefore, it is opportune that Raha does not sacrifice the stability and efficiency of its services to its server’s performance, so Raha’s delicacy in providing second-hand servers is justified. They provide their own services and the collections that only sell the server do not pay attention to the following. Here it is clear that the server vendors only offer their A-class servers to their customers, and the groups that only sell the server do not pay attention to the following items.
- Raha buys all of his second-hand servers directly from Jebel Ali, UAE, and individually waterproofs and shockproofs them at the origin.
- After entering the warehouse, the servers are completely overhauled, that is, the operator unfold the server with insulated gloves and anti-static bracelet and opens all the parts (Air flow, Ram, Power, Riser, Storage, cage, etc.) from the motherboard, The serial number of each part is recorded and the parts are classified separately in the warehouse. For example, all RAMs are categorized based on capacity, bass, brand, rank and working voltage.
The above parts are classified based on technical specifications; they are classified into the following three categories based on their appearance
- Almost new appearance: the price is 15% higher than the base price with a one-year warranty/ can be extended up to three years (7% in 2 years and 15% in 3 years)
- Normal second-hand condition: normal scratches/ (base price with one year warranty)
- Defective appearance condition: including deformity/ fractures/broken/twisted/repaired/rusted/non-original parts (15% below the base price with a one-week test deadline)
Server assembly process in Rahaco company
The RAM required by the customer is divided into 6 parts in order to use the maximum channels of each processor and achieve the highest performance. RAMs should be selected from the same brand, bass, rank and with a similar working voltage and capacity.
Both powers are selected exactly the same.
All the parts are taken to the insulated assembly table and the server is assembled by specialist, observing all safety measures (gloves, earth bracelet, server body anti-static connection).
The sink and processor are completely cleaned of silicone, and again, only kafuter Japanese silicone batter is used, which does not dry for a long time and has the highest thermal conductivity.
The server is turned on and the induction voltage of its body is measured (it should be less than 5 volts).
The server ROM is flashed.
Server’s bias firmware is updated to the latest version.
The latest version of ESXI is installed on the micro-SD on the server.
The customer’s storage is not placed on the server, but after assembling all the BAYs of the server is filled with Raha’s own storage to check the operation of the Raid Controller.
It is taken from the iLO server to verify the good operation of all components.
There are various systems for overloading the CPU/RAM/GPU/IOpS on the Raha’s storages, which will put the mentioned parts under 100% load for 4 hours.
The raid controller battery must be charged within one hour and reach the standard voltage. After the iLO test, it has been about an hour since the server is connected to the power, then the battery is disconnected from the cache and its voltage is measured.
Client storages are placed on the server and arranged.
The server is delivered to Information Technology section to install and run the service.
All the above items are done regularly according to the instruction and checklist, and about 10 hours of specialized work is done on each server before installing the service (overhaul + assembly and test).
So, keep in mind that according to the items mentioned above, there is a 60% price difference for a second-hand server with the same hardware specifications.
detect the originality of server components
This is the hottest, most complicated, most difficult and most important part of choosing a server, and customers always face many challenges at this stage.
Be assure that it does not have fake processor and motherboard, its technology is so high that it has only 6 main manufacturers in the world, and Intel processors are used by more than 99% in Iran. and in the case of motherboard, if a company with the production potential, due to the expensive production technology, they cannot compete with the old vanguards in this field.
But in the case of memory, external, peripheral components, connections and other items, there are many fake parts in the market which are not easily recognizable, and only by specific tests and examining their efficiency can you be sure of their authenticity. And mostly, they look very similar to the original versions and even experts might misunderstand their visual recognition.
In this case, try to purchase from a reputable place and be sure to pay attention to the product’s guarantee, because the non-originality of a seemingly unimportant piece, in addition to the fact that it can threaten the stability of your business, it may lead to the loss of information and security.
From what brand should we get the hard drive and memory required for the server and with what features?
This is very important, so please pay special attention. the following are mainly important in choosing a peripheral memory:
- the speed of reading and writing
- input/output operations per second or IOpS
- life span
- Appearance or Form Factor
- Communication port
In general, the types of external memory are divided into the following two categories (magnetic tapes are included in the category of cold memory):
- Memory based on rotating disk or hard disk – Hard Disk Drive: HDD Hard Disk Drive
Hard drives are mainly available in two small 2.5-inch (SFF: Small Form Factor) or large 3.5-inch (LFF: Large Form Factor) formats in the market.
- Flash-based memories or solid memories (without mobility) – SSD Solid State Drive
SSDs have more variety of appearance in the market including: 2.M, m-SATA-SFF SSD, PCI NVMe SSD, each of which has 6 different features (mentioned above).
Practical communication ports available in the market include: 2Sata, 3Sata, 6, 12SAS, 2SAS, M, PCI m-Sata.
How to choose the necessary resources for the server that meets our today’s needs and if we plan to upgrade in the future, we can upgrade with the least cost.
Decision-making in this regard is specialized and experimental and many cases are influential in it which cannot be included in this article! Therefore, in this case, it is necessary to ask help from an expert and experienced team that is well versed in the market and uses, Rahaco is always by your side.
Considering all the above contents and adding items:
– The originality and non-originality of the components
– New and stock parts
– The components are refurbished
– Server import type (legal or illegal)
– Buying a server directly from a representative in the UAE or from an intermediary
All of them affect the price of a server with fixed hardware features, and sometimes it causes a server with certain specifications to have an odd price difference between vendors, which needs consideration.
– Dollar price at the time of purchasing the server by the importer
– Importer’s purchase volume
– Whether the importer bought his server during the special offer or not
– Warranty type and after-sales service terms and conditions
At the end of 2015, with the aim of focusing more and providing better services to the market, HP divided its products into two categories enterprise and general, and for its enterprise products, it used the brand (Enterprise) HPe from this date onwards, and the public products presented under the same HP brand. Therefore Therefore, there is no difference between the brand HP or HPe.
The servers are designed and produced for permanent and long-term operation, and each part has a specific useful life that is mentioned in its specifications, and all parts connected to the server are required to report their health status and lifespan to the server and we can see these reports at any time and if a part is proceed toward the end of its useful life, we will receive a warning in iLO server, Be careful that the following conditions is very effective during the life of the server and its parts:
– online and sinusoidal UPS (Servers require special UPSs which are for example, very different and more expensive than the UPS used for CCTV systems.
– Optimum ventilation and standard temperature and humidity conditions
– Stabilizer or power oscillator (passive and active protection device against power fluctuations)
– Vibration-free environment
– Dust-free environment
– Ground connection
– Main assembly (non-static gloves and bracelets)
If all the above conditions are normal and favorable, the useful life of the essential parts, assuming that they are original and from high-quality brands, is estimated as follows:
- RAM: This part rarely fails and usually has a lifetime warranty.
- storage: The lifespan of this part depends on how much it is used, but in the world’s largest data centers, they consider its lifespan between 3 and 5 years (note that the average lifespan of an SSD is about 7 years, but if it used full-time and with maximum capacity (just like the Chia cryptocurrency extraction), its lifespan is about 2 months)
- motherboard: The average lifespan of this part is about 15 years.
- power : the average lifespan is not considered for power because it is absolutely affected by the conditions and the amount of use, but since it has a low price, its lifespan is not very important and it is always supported in the circuit, and when one of them goes out of the circuit It will be replaced quickly and easily (Hot Swap).
This is a very important issue, that you have purchased a server and deployed your services on it, and even you may think that the circumstances are desirable, but this server + services are not in optimal condition, this is very common, in almost all the cases that Raha experts examine, they notice this issue, the reasons for this lack of productivity can be some or all of the following items:
- • The server hypervisor is not configured correctly and is not installed on the correct storage.
- • The hardware configuration of the server is not optimal and is by default.
- • Proper arrangement for storages is not considered.
- • Necessary optimizations have not been done in the operating system layer.
- • In the services, the necessary settings to increase productivity have not been done correctly.
- • Information security requirements have not been regarded in all four layers of hardware, operating system, hypervisor and service.
that the result of not complying with the above items can cause risk in addition to the effect on productivity, it should be noted that the review of each of the above items is very specialized and the expertise required to check each one is different from the other and usually one person does not need reviewing knowledge of all the above items, and a team of experts is needed to act in a absolutely standard and coordinated manner.
Therefore, it is recommended to request an audit from professional experts in this field.
Be careful that the servers are sold as CTO (configuration based on customer needs : Configure to Order) and in addition to a high variety of parts, they are also expensive, and practically the availability of all various parts requires a large warehouse that is not possible for an Iranian company to have all the parts with different specifications, brands and generations, so taking advice from a company that is a server vendor will lead you to your product inventory, which ultimately may not be the most suitable choice for you to purchase.
So, it is strongly recommended not to ask any help from the vendor.
This is completely a misconception in the case of servers and in many cases, you can have more productivity, speed, stability, and security with a more accurate choice while paying less. Therefore, it is recommended to be more careful in choosing the server, which is one of the critical infrastructures of the organization, and be sure to benefit from an expert team, at least as a consultant.
Practical communication ports available in the market include: 2Sata, Sata3, SAS6, SAS12, M2, PCI, m-Sata. But be careful that common misconceptions in the market do not affect your choice, these misbeliefs include:
- • regular SATA SSDs do not have the ability to be RAID in the server.
- • If our server is SAS, we cannot use SATA storage.
- • The failure rate of SSDs is higher than that of HDDs.
- • SAS hard drives have the highest speed and durability.
- • M2-NVMe-SSD cannot be used in servers.
All the above beliefs are wrong and unfortunately, many sellers do not know that, to choose the right storage, it is necessary to pay attention to the following points:
- 1. The speed and lifespan of SSDs are always higher than HDDs. (High quality brands)
- 2. Only use HDD in cases where you need a large volume (more than 4 TB), in this case you have to use large LFF HDDs, so be careful when choosing a server.
- 3. SFF memory can be placed on LFF server, but vice versa is not possible, so if you need to have high capacity on your server, choose LFF type server.
- 4. Do not blow your budget to buy a SAS SFF HDD, buy a SATA SSD instead, it’s cheaper, faster, and lasts longer.
- 5. IOpS is very important and this parameter is much stronger in SSDs than HDDs.
- 6. Your first choice should be Sata SSD, except when you need more than 4 TB capacity, that you need to get LFF HDD, and in cases where you need high IOps (more than 100K), your choice is limited to SSD-NVMe-M2.
- 7. With the fair layout (Riad) you can achieve the desired features.
- 8. Raid family configurations (parity) are LOpS killers and increase your Raid Controller traffic to a great extent, avoid choosing such configurations and only use family 0 and 1.
Your type of user determines your storage type, considering all of the above tips, you can make the most appropriate choice, get free expert consultation from RahaCo.
What are the warranty conditions of the servers? And why does it matter?
Servers are among the goods that have corporate and organizational uses, that is, they are not for home usage, so manufacturers also produce them with the assumption that you must comply with the proper infrastructure to use them, which includes:
- 1. Online UPS and full sinusoidal. (Servers require special UPSs which are, for example, very different and more expensive than the UPS used for CCTV systems.
- 2. Optimal ventilation and standard temperature and humidity conditions
- 3. Stabilizer or power oscillator (passive and active protection device against power fluctuations)
- 4. Vibration free environment
- 5. Dust free environment
- 6. Ground connection
- 7. Main assembly (earth gloves and bracelets)
Therefore, neglecting any of the above cases usually invalidates the warranty, but be careful that in the case of the server, it is not the loss of a server or a part, but it can cause a general interruption in the business process, which costs much more than the rial value of the server itself or its parts. So, the issue of originality of parts and standard conditions of maintenance and assembly and periodic service of the server is much more important than warranty.
But the important question that whether Raha is a server vendor or not?
Answer: First of all, Raha is not a server seller and only offers servers for its customers, the reason is that server review is a very complicated process and as you can see, Raha does not introduce itself as a server seller in any media But Raha is a server importer and all the servers it imports are Bare Bone.
It means that it only includes the framework and motherboard, and you receive the other hardware including RAM, processor, storage and other parts based on the customer’s needs from the first-hand importers with the agency’s tariff, so Raha’s advice is based only on your needs, which If you finally intend to buy, the parts offered to you are procured from the market and assembled on the chassis according to the highest standards and assembly according to all safety precautions, and finally perform complex performance tests on the server to make sure the server works with maximum efficiency and optimally because Rahaco does not want to sacrifice the quality and efficiency of its services to the hardware.
The point that you should consider is that the first-hand server importers usually do not sell to the end customer and sell in high volume through the sales network of their representatives and colleagues, This point is important as the final customer usually cannot buy from the importer directly, and Raha supplies the equipment with the agency tariff from the first-hand importers, and offers them to the end customer at a price (Raha’s income is determined based on services), mainly Raha prices are competitive.