▷ What to buy and not die trying?

In this article we help you to solve the eternal question: what NAS to buy and what are the characteristics to take into account. And is that when we need that several computers in a local area network (LAN or WLAN) can access certain information, direct connection applications (DAS) soon become inefficient from the economic point of view.

For this reason , network, NAS and SAN architectures are used, which facilitate the consultation of information from different terminals. In the case of SANs (storage area networks), the systems are usually intended for the accumulation of data blocks in databases; that is, structured data.

If you work with audio, video, text, code or similar files (unstructured data), then NAS storage applications are a high capacity, easy scalability and easy access alternative.

The operation of NAS systems is not complex. Within the LAN or WLAN network we can find several well differentiated elements:

• NAS head , NAS box or NAS gateway . It is the hardware element that connects networked storage with the Ethernet switch, generally through a Fiber Channel (FC) connection using NFS or CIFS protocols, among others. Ethernet switch and router. Using one of them excludes the other. These are devices that allow more devices to be connected to the local area network and facilitate communication between them. Servers Remote data providers connected directly to the communication element (switch or router). Clients Terminals of users who make I / O requests to the NAS head .

Networked storage offers an ideal framework for multidisciplinary teams, whether they work in a local office or from different locations. Its use is also common in home automation and Internet of Things (IoT) integration in the workplace, or even as a multimedia hub at home.

Creating backups and implementing disaster recovery plans are two other tasks that can greatly benefit from the use of a NAS.

Index of contents

Main distinctions among the NAS on the market

What's the difference between a 100-euro NAS and one that costs ten times as much? First of all, the range to which they belong.

The needs of different types of users are so varied that there are three well-differentiated market niches. The products that are destined to each of them meet the expectations of the buyers of these segments. So you have:

• High-end or professional. High-capacity NAS, bandwidth, and external scalability, suitable for servicing large numbers of clients and making virtual machine images or redundant copies of huge data volumes. They also provide great robustness and reliability in storage (for example through redundant power supplies or file systems and higher volume managers such as ZFS), as well as compatibility with SAN solutions, multiple protocols and remote replication processes. The price range is between 5, 000 and 7, 500 euros. Medium range. They differ from the previous ones in capacity (always notably below the petabyte) and in the expansion possibilities ( clusters are not supported, but file silos can be created). They usually have processors fast enough to serve about ten users, dual Gigabit Ethernet ports that open up the possibility of using IEEE 802.3ad link aggregation, iSCI (Internet Small Computer System Interface) and more. The price of models in this category rarely exceeds 1000 euros. Low range or consumer level. Intended for private users and small businesses in need of locally shared storage with basic features (real-time file synchronization and moderate storage memory). They are options that demand little investment and knowledge. This category contains abundant models below 500 euros.

But the economic and functional point of view is not the only one that allows a first classification of the available NAS. Depending on the manufacturing philosophy of networked storage devices, we can catalog the devices in three other divisions:

• Computer based NAS. We can find them in the catalogs of brands like Asustor, Thecus, Synology or QNAP, to give some examples; but also included in this distinction are systems improvised by computer lovers from commercial or custom towers. For its creation a personal machine or server is taken and the pertinent software is installed (FTP, FTP / SSL, SMB, CIFS, AFP, iSCSI, WebDAV, etc.). These are the most versatile models, but they require in-depth knowledge to maximize their performance. They are inefficient in energy consumption but their scalability allows to limit electricity consumption with precision. On the other hand, it also makes it easier to control the speed of information transfer, which is linked to the CPU and installed RAM. NAS based on integrated systems. They are produced by houses like Oxford, Marvell or Storlink. ARM or MIPS-based processor architectures, embedded operating systems, or real-time operating systems (RTOS) are used to run the NAS server. The use of this type of solutions rules out the possibility of modification; in exchange, the user has more efficient electricity consumption, information transfer rates between 20 and 120 megabytes per second and functions suitable for demanding applications. ASIC-based NAS. They use an application-specific integrated circuit to implement the TCP / IP and file system protocols. It is the most energy efficient NAS, but it is also the least versatile, accepting only the SMB and FTP network protocols, and the data transfer rate is at most 40 megabytes per second. The market for ASIC-based NAS is very small.

For general applications, computer-based NAS are the most desirable due to their extensive configuration possibilities. Fortunately, the manufacturing companies seem to have turned to this type of hardware instead of betting on NAS based on integrated systems or NAS based on ASIC.

NAS specifications and their impact on applications

When choosing the right NAS for our specific needs, it is vitally important to carefully examine the specifications of the device. In the technical sheet we can find abundant information, while other characteristics are evident just by taking a look at the external aspect of the hardware . Now let's analyze the most decisive technical aspects:

Number of available disk or drive bays

NAS drives with one to 24 bays are currently on the market. The number of available bays has a direct impact on the maximum storage volume of the device and influences the type of scalability to be used in the future.

For starters, single bay systems should be discarded when data is critically important. Thus, the only applications in which this type of model is acceptable is as a multimedia server in domestic environments.

For any other use in which the data is vital (labor applications, for example) we will want to have at least two bays. The existence of two spaces in the NAS enclosure indicates the possibility of establishing a redundant array of independent disks. Models with fewer bays are usually limited to RAID 0 and RAID 1. Thus, they are ideal for applications where data security cannot be compromised, but their volumes are not very high (Keep in mind that these RAID configurations cut the available space on HDDs and SSDs in half.)

When storage requirements are higher, NAS with four or more bays are a must. These come equipped with more complete RAID cards that allow the use of RAID 5, RAID 6 and RAID 10, among others. If you opt for a high-end or professional model, the data management functions are more complete.

Above the four bays, the only criteria that matter are economic, storage, and scalability.

First of all, keep in mind that to achieve a certain memory capacity it may be cheaper to purchase a NAS model with numerous bays and supplement it with lower capacity hard drives or solid state drives. Getting the most out of a small NAS using the latest generation HDD and SSD can be very expensive.

On the other hand, if the volumes of information to be stored are very large, there is no other option but to resort to NAS of many bays and, in addition, equip the largest direct-connect storage units. Considering the new generation of 16 terabyte HDDs and 24-bay NAS, the maximum achievable memory is 384 terabytes.

If scalability is important due to our growth forecasts or due to the inherent tendency of the specific application to manage higher volumes of data over time, we should consider the two ways of scalability existing in NAS systems:

• Internal scalability or NAS. To increase the capacity of the system, a greater number of hard disk drives or solid state is used (for which there must be free bays), or models with more capacity of these same hardware elements are used (for which the devices are replaced direct connect storage). Clustering or external scalability. This methodology is applied when the internal scalability options have already been fully exploited and the NAS does not support larger volumes of data. This is the connection of two or more clustered NAS units, for which a distributed file system (DFS) is used that works concurrently on all devices to facilitate access to existing files on the system, regardless of what is the physical node where this data exists.

On the other hand, the bays can be free ( diskless ) or have pre-installed drives. These last models can come configured to use certain RAID. There are two legitimate reasons to opt for one or the other product. First of all NAS with pre-installed drives are easier for medium or low level users. This prevents errors when setting configurations or choosing HDD, SSD and SSHS compatible for NAS. Secondly, some manufacturers give priority to full NAS as this ensures the sale of their storage items. To attract the attention of the buyer, they offer a set with a more attractive price than what would be achieved by buying the NAS and the DAS storage units separately.

Data transfer rate

Unlike DAS expansion units in which the only elements that limit the speed of reading and writing data are storage units, NAS systems can see reduced performance for various reasons.

Although it is true that speeds as high as in DAS can never be achieved, this is no reason to give up on maximizing the data transfer speed. When choosing the NAS model that best suits our needs, we must consider the bottlenecks that can appear in the following elements of the system:

• Local area network. You must know the specific standard on which the LAN is based. IEEE 802.11g, IEEE 802.11n and IEEE 802.11ax are the most common in WLAN, where the effect of network overhead must also be assessed. Other standards in the 802 branch are used for Ethernet, Fast Eternet, Gigabit Ethernet (Gigae), and the like. The nominal transfer rate of these standards limits the read and write speed of the entire system, so there is no need to purchase NAS with higher speeds if no network enhancement is anticipated within the lifetime of the storage device. Network card (NIC). Another bottleneck can be found in the network adapter. For the most common speeds currently in home and professional applications, a 10/100/1000 Gigabit PCI Express or higher network card is recommended. In some cases, and in order to improve performance, NAS systems have two NICs, one dedicated to the LAN or WLAN network, while the other is dedicated to file access using NFS or CIFS. Storage units. In the event that low-end HDDs are used, it may be the case that these prevent the full speed operation of the NAS. When selecting a hard drive for the NAS, you have to check what its rotation speed or transfer speed is, and make sure that it is compatible with the optimal use of our networked storage system. In the case of SSDs and SSHDs this aspect may also become relevant if the NAS is mid-high range or professional. Processor. The most diverse CPUs are used in computer-based NAS: Intel Atom (to avoid the C2000 family with early degradation of the circuitry), late generations of Pentium and Celeron, Core i3 and i5; AMD Bulldozer, Llano, Trinity, Phenom and Athlon; Supermicro A2SDI… The CPU can be relevant in the transfer speed when using a low-end NAS, but in general you can use old models without any problem: state-of-the-art technology is not required since file management is a lightweight application. The only cases where this may be of interest is when it is required for heavy file (multimedia) transcoding, Plex servers, and multi-user optimization; but even so an Intel Core i3 is enough. Motherboard to consider. A specialized motherboard is required for server applications. It can be segregated or integrated in the CPU. The number and type of ports (PCI Express, SATA, M.2) must be sufficient for the specific use that you want to give to the NAS system, and the compatibility with the RAM to be used must be monitored. RAM memory. For most common uses, the amount of RAM required is small (about one gigabyte). However, if you want to use modern file systems like ZFS, have a virtual machine, transcode files, establish a Plex server or use heavy applications, the demands may be higher. Insufficient RAM will result in misuse of the NAS. In these cases the memory range must be between one and four megabytes. It may still be of interest to double or even quadruple the available RAM for specialized tasks such as creating renders or serving large clients. When selecting RAM, you must choose the appropriate DDR and DIMM (SO-DIMM, LONG-DIMM, DDR3, DDR3L and DDR4 are the most common). Finally, high operating frequencies are of interest, since they define the communication speed.

Properly dimensioning each of the elements that interact or are part of the NAS allows avoiding I / O contention when multiple clients execute requests to the NAS, overloading it; and at the same time it will make the most of the network facilities that we already enjoy.

Connectivity, compatibility and

To ensure that the connectivity of the NAS device is optimal, it will be necessary to take into account the number of ports for Ethernet, its type and what other inputs are available.

As far as Ethernet is concerned, dual Gigabit Ethernet ports are of interest to enable link aggregation configuration. Additional ports always lead to better performance. It is also convenient to have dual RJ-45 physical interfaces, PCI Express 10/100/1000 Base-T connections (independent or truncated over AIS 3000 or 6000 series), Gigabit fiber channel for SAN, USB (normal, 2.0 and higher)), integrated network adapters (if needed), etc.

In terms of compatibility, it is sought that the supported file systems include CIFS and SMB for Microsoft Networks, NFS over TCP and UDP for Linux or UNIX, AFP for Apple, HTTP 1.1, HTTPS for web use, FTP and others such as EXT3, XFS, FAT, FAT32…

The network protocols that are available are also of great importance, some of the most common are: TCP / IP, UDP / IP, iSCSI, AppleTalk, NFS v2, v3 and v4, NDMP v3, SNMP MiB II, SSH, DFS, SNTP, TFTP, FC, etc.

As for the software , each SKU is a world so a detailed analysis by the buyer is required, considering at all times what the real requirements are in the desired application. Some of the most common proprietary programs deal with security, RAID configuration, virtual machine management, multi-channel management, information retrieval, image generation, backup procedures, remote administration interfaces and much more.

In general, the greater the number of bays and the higher the range of the purchased device, the greater the connectivity, compatibility and software benefits.

Most common applications and priority specifications

Depending on the use of the NAS, some technical or other specifications will be necessary. The following summary list illustrates some of the tasks that are most commonly performed with networked storage systems, and what are the technological aspects that most dramatically affect hardware performance in that application:

• Creation, management and maintenance of backup copies. It is a very light application for NAS, which can be executed with full guarantees by low-end models with few benefits. The price of the system, as well as its derived expenses (electrical consumption and need for cooling peripherals) and storage capacity are the main points of interest when selecting a NAS solution on the market for this particular use. Server and multimedia player. In this case, the computing and working memory needs are higher than the usual for NAS. Models with CPU and RAM above the standard will be chosen, if possible with the ability to do H.264 transcodes. Storage virtualization. In this case the CPU must still be higher compared to the previous point. The processor should have multiple cores and more memory. Use in wireless networks. The network card takes on special importance. If you do not use Gigabit Ethernet or higher (10GbE), the transfer speed of the NAS will be adversely affected. Database systems for intensive writing. In this case, it is worth having an SSD cache accelerator to reduce response times on NAS storage units in demanding IOPS applications. This improvement can increase the performance of these operations up to 10 times, while dividing the latency by 3 storage volumes. Virtualization host . Again, the CPU and RAM take on special importance. NAS sold as standard are generally not suitable for this type of service, the exception may be some professional range SKUs.

We advise you to read the following articles:

With this we finish our article on which NAS to buy. Has this short tutorial helped you? We are waiting your comments!