Hard Disk Drives (simply HDD) are different when it comes to servers. This segment has higher standards and requirements compared to regular home usage thus you need to pay more attention when choosing the right Hard Disk Drive for a server.
In this article we will consider the basic types of hard drives for servers: how they differ and whether they should be necessarily from the server manufacturer’s brand. Server hard drive should be always ready to respond to many requests with minimal latency and to provide a high level of data integrity.
Server hard drive works almost 24/7/365. It constantly reads and writes data thus it should provide minimal latency, maximum reliability, and decent speed and performance. Here are the most important server Hard Disk Drive requirements:
- Reliability — unrecoverable data loss can result in multi-million dollar and reputational losses;
- Performance — servers are designed to handle multiple requests that need to be processed quickly;
- Response Time — users don’t have to wait for an HDD to “wake up” and process their requests.
How to Choose Right HDD for Your Servers?
First, let us talk about primary categories in which you can choose a specific model for your server. Modern servers nowadays use the following types of HDDs:
- SATA. Operating speed varies between 5400 RPM and 7200 RPM. These drives are almost the same as regular consumer-oriented HDDs. Solid options include Seagate Exos 7E2 and Western Digital 6TB Ultrastar DC HC310.
- SATA RAID Edition or SATA RE. The operating speed is 7200 RPM. These drives support special RAID-controller commands.
- SAS. A special type of HDD with extremely high speed (up to 15 000 RPM) for storing frequently used data. Consider something like Seagate Exos 7E8 or Western Digital RE SAS.
At first you need to decide what type of the connection interface you want — SATA or SAS.
SATA or SAS?
The SATA interface is a further modification of the IDE, which was later renamed to PATA. This interface is initially focused for the use on home computers as well as industrial systems with moderate requirements for performance and reliability. At the same time, SAS is the successor of the classic server SCSI interface.
Initially SAS interface has more bandwidth than SATA. But in the 3-rd generation, SATA III has a maximum bandwidth of 6 Gb/s, as well as the second generation of SAS. Nowadays servers with a third-generation SAS-controller is available, with a bandwidth up to 12 Gb/s.
For connection, SAS-disk server must be equipped with the appropriate controller. This provides backward compatibility interfaces to SAS-controller can be connected to the SATA-drives, and makes vice versa impossible.
SAS provides full-duplex communication: the hard drive handle two commands at the same time (one read and one write command), and SATA-drive — either to read or write. But this advantage is noticeable only in case you are using large number of disks, if you compare SAS NL and SATA RE.
If we’ll take a look from intermediate results point of view, SATA-drives are good for creating bulk storage, which is not require maximum performance. And in case you need a fast disk subsystem, your best choice is SAS.
Avoid Regular Consumer-based HDDs
Sure, those are much cheaper and affordable, but they are not designed for use in servers. There are few solid reasons why it is so. For example, an ordinary HDD has much lower vibration resistance. In other words, the shock that server-grade HDD can withstand without any damage may destroy an HDD for home use. In addition to that, regular HDDs have a high level of non-recoverable errors and do not support RAID-controller commands.
At the same time, do not fall into extremes. A typical HDD will work just fine in a home-made NAS, so don’t spend tons of money on a server-grade drive just to create small server storage for your children’s pictures or home videos.
Server hard drives have many advantages comparing to regular ones:
- The reinforced spindle shaft is more resistant to shock and vibrations.
- Additional vibration control.
- A special technology significantly increases positioning accuracy and the height of the flight above the heads.
- Rich self-diagnostics software provides timely notifications about imminent disk failure.
Non-recoverable Error Level
Another important difference between the server hard drives is the level of unrecoverable errors. Ordinary SATA is about 10-14. It means that when overwriting 2 TB drive 6 times, you will almost 100% get one unrecoverable error. This is not a problem for home-use, but if you need to overwrite an important database each month, then bad things start to happen. For SAS disks level of non-recoverable error even lower — it is 10-16, for SATA RE/SAS NL — 10-15.
SAS drive provides a lower error rate thanks to an additional 8 bytes of data used for verification. In addition to that, it has advanced reading and recovery algorithms.
For example, the following table contains the value of the probability of error on rebuild for the different types of drives configured in RAID Level 5 array:
|600 GB Drive||1 TB Drive|
|SATA RE / SAS NL||2,4%||4%|
Note. Potential for error is proportional to the number of drives into a RAID-array.
How Does SAS-disk Provide a Lower Error Rate?
- The size of the sector on the SATA-disks — 512 bytes, on SAS-drives — 520 bytes. Additional 8 bytes are used for parity verification;
- Other reading algorithms;
- Advanced data recovery algorithms without controller.
Drive Form Factor (Drive Size)
Drive form factor directly determines the capacity of your server and energy efficiency. Currently, only two disk sizes are available: 3.5″ and 2.5″, also known as LFF and SFF.
3.5″ is the most widely used size. It allows you to place the maximum amount of data. 4 TB 3.5″ hard drives (such as WD Gold 4TB Enterprise Class Hard Disk Drive or Seagate 4TB IronWolf) are usually recommended for servers with higher storage requirements. But the downside is that they consume more power than smaller 2.5″ drives. Also do note, that SSDs are not available in this form-factor but you could use a special 3.5″ adapter. For example, ORICO 2.5″ to 3.5″ Hard Drive Adapter. The main benefit of 3.5″ hard drive is high capacity at a more affordable price. Their price per GB ratio is the best.
2.5″ is a common size for laptop HDD and a regular SATA-based SSD, although these drives are only one inch smaller in size, they use much less power compared to their big brothers. Currently, the maximum volume these drives is around 2TB (for example, take a look at DELL 2TB 7.2K SAS 2.5″ or Hewlett Packard Enterprise 2TB SATA).
This is the most important spec of every drive, no matter whether it is HDD or SSD, home-use or enterprise-grade. 3.5” drives store up to 14 TB data (for example, Western Digital 14TB Ultrastar DC HC530), while 2.5” drives go up to around 2-4 TB (like SEAGATE 1TB EXOS 7E2000 HDD). Buying the largest capacity is not always the best practice. Never install one drive on a server. Always combine several (at least two) to ensure reliability and redundancy. For example, for a 4 TB server, it is better to buy 4 1 TB drives or 2 2TB drives rather than 1 4 TB drive. Using a few drives allows you to create RAID and increase read/write speed or make a backup drive to avoid data loss. This is one of the reasons why the capacity of server hard drives is usually lower than for desktops.
Classification of the Hard Disks
In order to summarize all these aspects and simplify the choice of customers, the manufacturer often defines several classes for hard drives:
Economic (ECO) — disks in this class have a low price per unit. The level of performance and reliability of these drives determines their purpose for entry-level systems. They should be used in non-critical areas with low I/O loads and moderate speed requirements. High loads can lead to a deterioration in their reliability. ECO drives run at 5400 or 7200 rpm and usually have a SATA interface.
Business Critical (BC) or Nearline — disks in this class offer high capacity with a minimum cost per 1 GB. They are designed to provide good performance and suitable reliability. Depending on the implementation of the server, “BC-disks” can be equipped with SAS or SATA interfaces and have a speed of 7200 rpm.
Enterprise (EP) — disks of this class provide maximum performance and reliability. They are designed to cope with the maximum workload. This class uses the SAS interface and has a rotational speed of 10,000 and 15,000 rpm.
SSD Enterprise Performance / Mainstream — these drives offer the best performance and durability in the SSD segment and are thus ideally suited for use in systems with high demands on I/O. Enterprise Performance SSDs (SLC or MLC technologies) offer better I/O performance with the SAS interface. In contrast, Enterprise Mainstream SDD (MLC technology) has a SATA interface and is more affordable.
For these tasks, it is preferable to use SSD on a Marvel chip, for example, some Crucial SSDs (Crucial MX300 1TB) or at least OCZ on RD400 (Toshiba OCZ RD400 1 Tb). Disks on the chip SandForce (SF) are better avoided.
Choosing the Brand
Finally, the most important question: do you need to buy a brand drives? There is no secret that HP, IBM and DELL do not produce hard drives. They buy them from third-party manufacturers, and then test them, reflash and put their logos on them.
One hand, these drives have several advantages:
- Firmware considers features of controllers of various server models;
- Additional quality control and stress tests reduce the probability of purchasing drives with hidden defects;
- Vendor warranty and full support is available for the brand drives;
But “native” hard drives are about two or three times more expensive than the same model, but with the logos of non-native manufacturers – you can purchase Seagate, Western Digital, Toshiba, HGST and we are sure, that such difference in price is not justified for all the promises of increased reliability.
As the practice shows, the “non-native” hard drives operate without difficulties on:
- HP servers — at least from Gen 6;
- IBM servers — at least with the M2 generation;
- DELL servers — from 10 generations.
According to the statistics, most reliable are Hitachi drives. In second place Western Digital, on the third – Seagate.
Сheck Your Warranty
If you decided not to buy “non-native” hard drive, at first you must ask the seller: who provides the warranty? It is the fact, that many stores do not provide a guarantee on the hard disks, referring to the manufacturer’s warranty. But there is a subtle point: for example, some models of Seagate warranty period starts from the date of production. So do not exclude the situation that you can purchase a brand new drive, which is over the manufacturer warranty.
Before purchasing, you can check the warranty of specific instances on the manufacturers’ websites:
- Seagate: http://support.seagate.com/customer/en-US/warranty_validation.jsp
- Western Digital: http://support.wdc.com/Warranty/warrantyStatus.aspx?lang=en
- Hitachi: https://www.hgst.com/portal/site/en/support/warranty
When choosing a hard drive, at first you need to pay attention to the tasks that are going to be performed by the server:
- In case you don’t need a high-speed access and reliable data storage, and the number of discs will not exceed 4, we strongly recommend you choose the SATA RAID Edition drives. This is a great choice for low-cost entry-level servers, that serving a small number of users;
- If the server holds the database or the number of disks in the array will be 5 or more, it is better to choose SAS NL. Often, these discs are placed in servers that are used in medium-sized companies: under the accounting system, CMS, corporate repositories;
- If you need the maximum performance and/or reliability of data storage (for example, in the processing of financial transactions), your best choice is SAS Enterprise drives. It’s good for high-servers with a large number of users as well as for systems, which are operating with the most important data;
- In some cases, when you need to make a compromise choice, you can simultaneously use a mix of SAS + SATA drives for one server (you can also use SSD SATA drives);
To increase the reliability of the disk subsystem, it’s desirable to use disks from different parties on a single server in a RAID array.