Server hard drives are a critical component that impacts the performance, reliability, and total cost of ownership of a server. The type of hard drive used can significantly affect factors like I/O speeds, storage capacity, data protection, power consumption, and TCO over the life of the server.
With the rise of data-intensive applications like AI, big data analytics, and blockchain, choosing the optimal hard drive technology has become even more important. The right storage media can enable companies to efficiently run data-driven workloads and ensure maximum uptime and data integrity.
This article provides an overview of the major types of hard drives used in today’s servers. It examines the pros and cons of HDDs, SSDs, SAS, SATA, and other drive interfaces and form factors. Understanding the differences can help IT decision makers select drives tailored to their specific workload requirements and budget.
Hard Disk Drives (HDDs)
Hard disk drives (HDDs) store data on spinning magnetic disks called platters. A read/write head floats over the platter to read and write data as the platter spins (from https://community.fs.com/article/server-hdd-vs-ssd-which-one-is-better.html). HDDs have been used for decades in computers and servers due to their high capacity and low cost per gigabyte.
The capacity of HDDs has steadily increased over the years while costs have declined. HDDs now offer massive storage capacities up to 16TB for a single drive, allowing servers to store huge datasets cost-effectively (from https://serverfault.com/questions/986772/ssd-or-hdd-for-server). The large capacity of HDDs makes them well-suited for use in servers and data centers that require enormous storage for things like databases, virtual machines, backups, archives, and more.
In addition to high capacity, HDDs are inexpensive per gigabyte compared to solid-state drives. While HDDs have slower performance, the much lower price point allows organizations to maximize storage at an affordable cost when performance is not the top priority (from https://www.quora.com/Why-do-many-web-servers-still-use-hard-disc-drive-technology). For these reasons, HDDs continue to be commonly deployed in servers today.
SSDs
SSDs, or solid-state drives, store data in integrated circuits rather than magnetic platters like traditional hard disk drives (HDDs). This gives SSDs advantages in speed, reliability, power efficiency, and physical durability over HDDs (Source).
SSDs have no moving parts, allowing data to be accessed instantly rather than waiting for read/write heads to move into position. This gives SSDs much faster read/write speeds compared to HDDs. For servers where fast data access is critical, SSDs can provide huge performance improvements (Source).
The lack of moving parts also makes SSDs more reliable and durable. They are less prone to mechanical failure from shock, vibration, or component wear. SSD failure rates are about 2.5% over 5 years compared to 7-10% for HDDs. For mission-critical servers, SSD reliability is highly advantageous (Source).
Furthermore, SSDs consume less power, generate less heat, and operate silently since there is no spindle motor or head actuator. All these traits make SSDs ideally suited for server environments.
SAS Drives
SAS (Serial Attached SCSI) drives are enterprise-level drives commonly used in servers and storage systems. SAS drives connect via the SAS interface, which allows for much higher speeds compared to SATA drives. According to ServerMonkey (https://www.servermonkey.com/parts-and-upgrades/drives/sas.html), SAS drives are capable of speeds up to 12 Gb/s, whereas SATA drives max out at 6 Gb/s.
SAS drives are best suited for mission-critical applications that require high performance, high availability, and maximum uptime. As HP explains (https://www.hp.com/us-en/shop/tech-takes/sas-vs-sata), SAS drives offer superior performance benefits thanks to their full-duplex architecture, which enables simultaneous reading and writing. This makes SAS ideal for workloads with heavy I/O requirements like databases, email servers, and virtualized environments.
The SAS interface also provides advanced error and failure detection capabilities to ensure data integrity. SAS drives are built with enterprise-level components and undergo rigorous testing, resulting in high reliability and low failure rates. For these reasons, SAS drives are recommended for business-critical server applications where performance and uptime are paramount.
SATA Drives
SATA (Serial Advanced Technology Attachment) drives are the most common type of hard drive found in computers and servers. The SATA interface was designed for internal storage drives and uses a point-to-point serial protocol to connect the drive to the system’s storage controller. SATA drives communicate over cables with a 7-pin data connector.
SATA drives are a good fit for general computing needs and less demanding server workloads. Compared to SAS drives, SATA drives tend to be cheaper while still providing decent performance and capacity.
Some key advantages of SATA drives:
- Cost-effective storage solution
- Sufficient performance for many workloads
- Wide range of large capacities available
- Compatible with most servers and computers
SATA drives work well for tasks like:
- File and print serving
- Hosting websites and web applications
- Running databases and business software
- Media storage and backup
For mission critical systems or high performance workloads like transactional databases, SAS drives tend to be preferred over SATA. But for general purpose storage, SATA drives offer a versatile and affordable solution.
NL-SAS Drives
NL-SAS (Nearline SAS) drives are enterprise SATA hard drives with a SAS interface to allow compatibility with SAS controllers. They provide a balance of performance, capacity, and cost that makes them well-suited for use in servers.
Compared to regular SAS drives, NL-SAS drives typically have:
- Lower performance – NL-SAS drives have slower rotational speeds than SAS, resulting in reduced IOPS.
- Higher capacity – NL-SAS drives commonly go up to 12TB, while SAS max out at around 2TB.
- Lower cost per GB – The simpler SATA design allows for a lower cost per GB compared to SAS.
However, NL-SAS provides better performance and reliability than regular SATA drives. The SAS interface allows advanced features like dual porting. Overall, NL-SAS offers a hybrid solution balancing SAS capabilities with SATA economics.
Sources:
[1] How SAS, Near Line (NL) SAS, and SATA disks compare
[2] Dell EMC PowerEdge Enterprise HDD Overview
Flash Storage
Flash storage is a type of solid-state storage that uses flash memory chips to store data (https://www.netapp.com/data-storage/what-is-flash-storage/). It offers much higher performance than traditional spinning hard disk drives (HDDs) due to the lack of moving parts and fast data access speeds. Flash storage is commonly used for server storage in a few key ways:
Caching: Flash can be implemented as a cache layer to accelerate I/O performance. Frequently accessed “hot” data is copied from slower HDDs to the flash cache. Subsequent reads/writes to this hot data will be faster since it is handled by the flash cache rather than the HDDs. This improves overall storage performance without requiring all data to reside on expensive flash.
Index and metadata storage: Database indexes and filesystem metadata that require frequent access can be stored on flash storage to improve performance. The bulk data can remain on HDDs which are more cost effective for infrequently accessed “cold” data.
All-flash arrays: For applications that demand consistently high performance like virtualization, databases, and real-time analytics, all data can be stored on flash arrays. Latency is minimized when all storage operations involve flash.
Compared to HDDs, flash offers reduced latency, higher throughput, and lower power consumption. However, flash remains more expensive per gigabyte than HDDs. The tradeoff between performance and cost makes it well-suited for tiered storage implementations.
RAID Configurations
RAID (Redundant Array of Independent Disks) allows you to combine multiple hard drives together for increased performance, capacity, and fault tolerance. There are several RAID levels that use different configurations designed for specific goals:
RAID 0 – Also known as striping, RAID 0 splits data evenly across two or more drives with no parity or mirroring. This provides improved performance but no redundancy. If one drive fails, all data will be lost.[1]
RAID 1 – Also known as mirroring, RAID 1 duplicates (mirrors) data across two or more drives. If one drive fails, data can be accessed from the other mirrored drive(s) with no interruption in service. RAID 1 provides redundancy but reduced storage capacity.
RAID 5 – Uses block-level striping with distributed parity. RAID 5 requires at least three drives and is commonly used in servers for its balance of performance, capacity, and redundancy. If one drive fails, data can be rebuilt from the parity information.[2]
RAID 6 – Similar to RAID 5 but with double distributed parity. RAID 6 can withstand the failure of two drives and provides increased fault tolerance compared to RAID 5, but at the cost of write performance and usable capacity.
RAID 10 – Combination of RAID 0 striping and RAID 1 mirroring. Provides high performance and fault tolerance but requires at least four drives.
Drive Interface
The interface between a server and its internal hard disk drives is a critical factor in performance and connectivity. Two of the most common drive interface types used in servers are SAS (Serial Attached SCSI) and SATA (Serial ATA).
SAS drives use the SAS interface, which allows for much higher throughput and speed than SATA drives. SAS supports full duplex communication at speeds up to 12 Gb/s, while SATA is limited to 6 Gb/s. This makes SAS the preferred choice for applications that demand high performance like database servers, enterprise storage, and high volume servers. SAS is also designed for 24/7 operation and supports advanced features like native command queuing.
SATA drives use the SATA interface, which is cheaper and more commonly found in desktop PCs. The SATA interface caps at 6 Gb/s, but SATA drives still offer good performance at a lower price point. This makes SATA drives suitable for less demanding workloads like file servers, web servers, and print servers. SATA drives are also hot-swappable to allow replacing failed drives without downtime.
In summary, SAS delivers premium performance and enterprise-class durability, while SATA provides a budget-friendly option for basic storage needs. The interface chosen depends on the server’s intended workload and performance requirements.[1]
Choosing the Right Drives
There are several factors to consider when selecting the right type of hard drive for your server:
- Performance needs – If your server needs fast access to data, solid state drives (SSDs) provide much faster read/write speeds compared to traditional hard disk drives (HDDs). However, SSDs have higher costs per gigabyte of storage (Source: https://www.serversplus.com/servers/server-hard-drive-guide).
- Capacity requirements – HDDs tend to offer much higher storage capacities compared to SSDs. If you need to store large amounts of data, HDDs provide more cost-effective high density storage.
- Workload – For read-intensive workloads like web servers, SSDs can provide faster response times. For write-intensive databases or data analytics, enterprise-class HDDs are a better choice.
- Availability needs – For mission-critical data, RAID configurations with hot spare drives can enable continuous operations if a drive fails.
- Budget – The overall budget will determine if top-tier enterprise SSDs and HDDs are feasible or if more budget-friendly SATA drives are required.
For general purpose file servers, SATA HDDs offer a good balance of capacity, performance, and cost. For databases and other transactional applications, SAS HDDs provide better random access performance. For web servers and other read-intensive workloads, SSDs can dramatically improve response times (Source: https://community.spiceworks.com/topic/2354862-server-hard-drive-selection). By carefully matching drive capabilities with application requirements, servers can be optimized for best performance and value.
