Comparison Synology Diskstation DS1819+ RAM 4 GB vs D-Link DNS-320L

The number of slots for drives in the form factor 3.5", provided in the design of the server.

Initially, 3.5 "is the traditional, most popular form factor of drives for server systems. It is noticeably larger than 2.5", but it allows you to create capacious, inexpensive (in terms of gigabytes) and reliable media, in which it is also easier to implement various additional functions. That is why, specifically in NAS servers, this form factor is also the most popular; slots under 2.5" are much less common in such equipment, and in most cases they complement 3.5".

As for the number of slots, it can vary from 2 (or even 1) in the most basic desktop systems to 8 or more in professional rack-mount solutions. And not only their maximum capacity depends on the specific number of drives, but also some other features of work — first of all, the physical possibility of using one or another RAID level.

This item characterizes the maximum capabilities of the device for connecting drives. This way you can understand how much maximum memory can be added to the NAS server.

The ability to remove one of the NAS server's internal drives and replace it with another without shutting down the entire server. Thanks to this, time is not wasted on rebooting, and the information on other media remains constantly available. Note that even if this feature is available in the NAS server, it may not be available when using RAID — some versions of this technology (see "RAID support") do not allow hot- plugging drives.

The third version of the SATA standard used in computer technology to connect internal drives. It differs from the previous version of SATA 2 (see above) in an increased data transfer rate — about 5.9 Gbps (600 MB / s) in fact — as well as a number of optimizations in power consumption. Even earlier drives can be connected to SATA 3, however, the speed of operation with such a connection will be limited by the characteristics of the drive itself.

The number of M.2 connectors provided in the design of the NAS server.

The M.2 connector is used to connect various internal peripherals, mostly miniature form factor. Note that two electrical (logical) interfaces can be implemented through this connector — SATA 3.0 and PCI-Express, and each individual M.2 socket on the board can support both of these interfaces at once, or only one of them. These nuances should be clarified before buying, since the possibilities for using M.2 directly depend on them. So, with support for SATA 3.0, such a connector is intended exclusively for drives, and the speed of SATA is noticeably lower than that of PCI-E; so this M.2 variant is mostly used by inexpensive SSD modules. In turn, PCI-E is somewhat more expensive, but it is faster and more versatile. Support for this interface allows you to connect both high-end SSDs and various expansion cards (for example, sound cards or internal wireless adapters) to the NAS server.

The number of PCI-E slots provided in the design of the NAS server.

PCI-E is one of the most popular modern interfaces for connecting internal components to a computer motherboard. Specifically, in NAS servers, it can be used, in particular, for wireless adapters and SSD drives; in the latter case, PCI-E allows higher speeds than SATA and fully realizes the potential of solid-state memory. And the number of such connectors corresponds to the number of PCI-E components that can be simultaneously installed in the server.

Note that the PCI-E connection can use a different number of lines (1x, 4x, 16x), and for normal operation it is necessary that the slot on the "motherboard" has no fewer lines than the installed component. In fact, this means that a component with a 1x connector will easily fit into any slot, but with a larger connector, the connection should be specified separately. However, in the case of NAS servers, even PCI-E 4x capabilities are rarely required, not to mention 16x.

NAS server supports RAID technology. The term is an abbreviation for "redundant array of independent disks", that is, "redundant array of independent disks". Accordingly, only models with more than one drive slot can have this feature (see “Drive Slots”).

There are several options for combining disks into a redundant array, they differ in a number of characteristics: some focus on increasing speed, others focus on fault tolerance. However, all RAIDs have two key differences from non-arrayed systems. The first is that the RAID array is perceived by the system as one single drive. The second is “redundancy”: the total volume of disks included in the array must be greater than the volume of data that is planned to be stored on them. This is due to the fact that the array uses service information, which must be stored on the same disks (however, the exception is RAID 0, see below).

The most common RAID versions today are:

— RAID 0. An array of two or more disks, information on which is written by interleaving: first, the data is divided into blocks of the same length, and then each of these blocks is written to its “own” disk in turn. For example, if a RAID 0 array consists of 3 disks, and the file is divided into 7 parts, then parts 1, 4 and 7 will be on the first disk, 2 and 5 on the second, and 3 and 6 on the third. that it is not actually a RAID, as it devoid of "redundancy" — the volume of the array corresp...onds to the sum of the disk volumes. The main advantage of RAID 0 is a significant increase in performance; it is higher, the more disks are included in the array. On the other hand, the reliability of such systems is lower than that of individual drives: in the event of a failure of any of the drives, the entire array becomes inaccessible, and the more drives are used, the higher the likelihood of this. The minimum number of drives for RAID 0 is two.

— RAID 1. In arrays of this type, information is recorded according to the principle of mirroring: two disks, the information on which is completely identical. This provides a very solid system fault tolerance: the data contained in the array will be available in full, without additional tricks and serious drops in performance, even if one of the disks fails completely. In addition, some gain in read speed is achieved in this way, and "hot swapping" (see above) usually does not cause problems. The disadvantage is the high cost of building: you have to pay for two hard drives, getting the volume of one. However, in some cases this can be quite an acceptable price for increased reliability.

— RAID 5. In such arrays, unlike RAID 0 and 1 (see above), not only basic information is stored on disks, but also service information — in the form of data for error correction (so-called checksums). In this case, both types of information are distributed evenly across all disks. For example, in RAID 5, consisting of 4 disks, the first "portion" of data to be written will be divided equally between disks 1,2 and 3, and the checksum will be written to disk 4; the second portion is between disks 1,2 and 4, with a checksum written to disk 3, etc. This provides good fault tolerance: the array provides data access in the event of a complete failure of any of the drives. In addition, RAID 5 is characterized by a very low level of redundancy: the working volume of the array is equal to the volume of the smallest disk multiplied by (n-1), where n is the total number of disks. The main disadvantages of RAID 5 are relatively low performance, which drops even more in the event of a failure; this is due to the abundance of additional operations associated with the use of checksums. In addition, if one of the drives fails, the reliability of the remaining array is reduced to the RAID 0 level (see above), and the remaining drives experience very significant loads, which further increases the risk of additional failure; if two disks fail, data can be recovered only by special methods. The minimum required number of drives for RAID 5 is three.

— RAID 10. A combination of arrays of the RAID 0 and RAID 1 types (see above): the disks are combined in pairs into mirror RAID 1 arrays, and the whole system operates on the RAID 0 principle, with sequential information writing to each pair of disks. This scheme allows you to maintain the high performance characteristic of the classic RAID 0, while eliminating its main drawback — unreliability. Regardless of the number of drives, a RAID 10 array is completely insensitive to a single drive failure and can easily survive the loss of half the drives if they are all in different mirrored pairs. At the same time, the simultaneous failure of one pair leads to an irreversible loss of information. Another drawback is the high redundancy characteristic of RAID 1: the useful volume of the array is half the sum of the volumes of all disks. At least 4 drives are required to build RAID 10, and anyway, their number must be even.

— JBOD. Abbreviation for "Just a bunch of disks" — "just a bunch of disks." This name, although rough, but quite accurately describes the features of arrays of this type: JBOD does not provide "redundancy", does not use additional technologies such as checksums (see RAID 5), and the volume of the array is equal to the total volume of all disks included in it. The discs are connected in a kind of series. This means that when writing each next file, the remaining free space on the previous disk in the queue is first filled, and if there is not enough space, the rest of the data is written to the next one. For example, if you write two 70 GB files to an empty JBOD array of 100 GB disks, the first file will fit entirely on the first disk, and the second will take up the remaining 30 GB on the first and 40 GB on the second. Similarly, if the volume of the file exceeds the volume of the entire disk — in our example, a 120 GB file will occupy the entire first disk and 20 GB on the second. The advantages of JBOD are good performance with a small load on the processor and the ability to combine disks with different sizes and speeds. In addition, they are somewhat more fault-tolerant than similar RAID 0 in many respects (see above): the failure of one disk does not necessarily lead to the irreversible loss of data of the entire array. At the same time, the reliability of JBODs is still somewhat lower than that of single disks, and therefore they can only be considered as a tool for improving performance.

Note that the variety of RAID standards used in modern NAS servers is not limited to the above. Additional options may include but are not limited to:

— RAID 3 and RAID 4 — similar to RAID 5 described above, however, in these formats, checksums are written to one dedicated disk, and are not distributed evenly across all disks. This improves performance (for RAID 3 — only in some cases), but reduces the reliability of the control disk. For a number of reasons, they are rather uncommon.

— RAID 6 is another analogue of RAID 5, differs in that it uses not one, but two sets of checksums, also evenly distributed over all disks. This significantly increases reliability, but reduces performance and increases the level of redundancy — the volumes of not one, but two disks “fall out” of the total volume.

— RAID 0+1. It can mean 2 options. The most common is an array of two RAID 0 (striped) combined into a RAID 1 (mirror). Some manufacturers use RAID 0+1 as a designation for an advanced technology that allows you to “mirror” information on an odd number of disks: for example, in a three-disk array, the first piece of data will be mirrored on disks 1 and 2, the second — on 2 and 3, the third — on 3 and 1 etc.

— RAID 50 and RAID 60. RAID 5 and RAID 6 arrays, respectively, composed of groups of disks combined in RAID 0. Provide high reliability and performance, but are expensive and difficult to maintain.

There are also other options for "combined" RAID — for example, in RAID 51, two RAID 5 arrays are made into a "mirror" pair.

The number of LAN ports provided in the design of the NAS server.

LAN — a connector used for a wired connection to Ethernet local networks (the most common "local" format today, it is also used to access the Internet). For a relatively simple network (say, within a medium office), one LAN port will be enough. However, models are produced where there are more than one such ports, mainly 2 and 4 connectors. They are designed for large networks divided into subnets with separate access to the NAS server: the presence of several LAN connectors allows you to connect each of the subnets directly without using a router. This simplifies the network architecture and optimizes the load.

The number of USB 2.0 ports provided in the design of the NAS server.

USB connectors are used in computer technology to connect various external peripherals. In the case of NAS servers, we are most often talking about external drives — flash drives, hard drives, etc. In this way, you can transfer information from an internal drive to an external one (for example, for backup purposes) or vice versa, and even expand the total working volume of the server . In addition, on models with a VGA output (see below), a keyboard can also be connected to USB, and on models with a print server function (see "Software Features"), respectively, a printer. For added convenience, the USB connector can be placed on the front panel (see below).

As for USB 2.0 specifically, today this version is generally considered obsolete due to the relatively low speed (up to 480 Mbps) and the low power supplied through the connector. Peripherals of newer versions can be connected to such a port, however, the speed will be limited by the capabilities of version 2.0, and the power supply may not be sufficient. Therefore, in modern NAS servers, such connectors are quite rare — mainly as an addition to the newer and faster USB 3.2 gen1 (see below), designed for relatively unpretentious peripherals like keyboards.