In the IT world of colocation, data centers and rack space, those in the know are familiar with many concepts and specifications. DC1.AMSTERDAM explains the most important, frequently encountered terms to you.
Table of contents
Bandwidth, in the context of IT, internet and telecommunications, indicates how much data a connection can transfer within a given time. The greater the bandwidth, the more information the connection can transport. For an internet connection, the number of gigabits per second (Gbps) indicates the maximum space for data transfer within this mentioned time frame. Greater bandwidth will lead to faster and more stable internet, because it will allow more data to be sent and received simultaneously at any given time. Thus, extra bandwidth leads to fast(er) page load times, downloads and file transfers.
The Border Gateway Protocol (BGP) is the basis for the internet’s routing protocol. BGP also allows independent provider networks to communicate with each other if desired. In other words, BGP allows it to route network traffic between the autonomous networks of different providers. Without the existence of BGP, no reciprocal routes between providers would be possible, and their networks could only operate in isolation.
Keeping websites and other essential online business activities and services running. This requires business internet. Its essential feature is a greater guarantee of its operation, often combined with contractually agreed availability.
Colocation is the provision of rack space for servers and associated hardware in a shared, secure, professional data center, with power, cooling, fire safety, security, certification and remote hands. Colocation is also called colo for short.
More information? See our comprehensive explanation page on colocation.
In the context of a server room in a data center, guaranteeing (internet) connectivity is key. After all, the reason a server exists is that it gives a website, for example, its place on the worldwide net. Internet connectivity is crucial not only for the server’s online connection to the web, but also for the rest of the business network.
Cloud connect refers to providing a direct connection to the public cloud, such as for hosting, backups and storing files and data. This is also known as hybrid IT. Public cloud providers include Amazon Web Services, Microsoft Azure, Google Cloud Platform, IBM Cloud, Oracle Cloud, Salesforce Cloud, SAP Cloud Platform and Nutanix Enterprise Cloud. Are you renting colocation in a data center with flexible cloud connect? Then you can securely, directly link your IT infrastructure to your own cloud environment via a low latency connection. This allows you to choose which infrastructure to offer each service from. This provides maximum flexibility and performance.
A cloud provider provides solutions for storing and processing applications and data in the cloud. Cloud computing is simple, scalable and free of investment in proprietary hardware. Well-known providers include Amazon Web Services, Microsoft Azure, Google Cloud Platform, IBM Cloud, Oracle Cloud, Salesforce Cloud, SAP Cloud Platform and Nutanix Enterprise Cloud.
A cross connect is an internal connection within the same data center; between one rack and another rack (belonging to a third party). It is actually a colocation service for establishing a private connection directly to another party’s facilities. A cross connect is generally achieved via fiber optic (or sometimes still via a copper connection).
Cross connects are common. In fact, connections within the data center are cheaper than longer connections to premises that are further away. Do you, as a colocation customer, need direct connections to services from other organizations and companies? Then it is highly advantageous if these parties are in the same data center.
For a cross connect, the colocation service provider provides the necessary connections in the rack. The owner of the (server) equipment, to establish an in-house connection, can use this to patch and plug in the necessary cables themselves, in their own rack.
Applying for a cross connect also requires an LOA: a Letter of Authorization (see later on this page).
“Do you have dark fiber for us?” That frequently asked question is about a fully exclusive fiber optic connection for a long-distance pathway, from point A to point B. Those start and end points of dark fiber are not necessarily in a data center.
There is literally no light on a so-called dark fiber; hence the name. It is the longest direct fiber optic connection. It is an extension of a metro connect – a fiber within the same city or metropolis. That means dark fiber goes beyond the boundaries of a metropolis. Think of a networking pathway from one city to another.
The dark fiber is, you could say, the most exclusive network connection, followed by a WDM connection and then a VLAN. See also the explanation of those terms elsewhere on this page.
A data center is a central location for securely and efficiently housing servers and associated mission-critical IT hardware of external customers. It provides backup power supply and facilitates (internet) connectivity in a business network environment. A professional data center provides colocation space and is fully focus on climate control, fire safety, redundancy, monitoring and security. Everything is designed to maintain maximum availability of the servers and applications at all times.
More information? See our comprehensive explanation page on data centers.
Data traffic indicates the consumption in internet use, expressed in the number of bytes. Internet use can include visiting and thereby loading a website, updating your site, sending an email, using an app, establishing a server connection, and many other uses of digital information.
The concept of data traffic is therefore common in the field of web hosting and colocation. Providers of these services usually express the data traffic offered in a maximum number of gigabytes. Or the total digital data transfer capacity available from the server(s) to devices such as computers, laptops and smartphones.
DSL stands for Digital Subscriber Line. This type of connection provides digital communication, including internet, over a copper wire connection. The landline telephone cable also uses this technique. DSL for the internet uses a modem to receive and convert a high-frequency signal from the telephone network. It is one of the types of internet connections, alongside cable, fiber optics and (mobile) 4G. DSL has long existed as ADSL: Asymmetrical Digital Subscriber Line. Its successor is VDSL: Very high bit rate Digital Subscriber Line. This offers higher internet speeds.
Fiber Optic Connection
A fiber-optic-based network connection allows for data sharing at potentially the speed of light. Optic fibers are literally thin fibers made of special glass. The thickness of such a flexible fiber is comparable to that of a human hair. To protect them, the optic fibers are encased in cabling with plastic sheathing.
A fiber optic connection transports light and data ultra fast and efficiently, in part because of its construction. The technique causes the light in the fiber to reflect internally, so it can send a lot of data at once, over a wide bandwidth, without loss of quality.
The overall fiber optic connection consists of a large network of underground cables. At the ends are branches that reach near, on or even into the business premises or home. The latter is referred to as an FTTH (Fiber To The Home) connection.
FTTH stands for Fiber To The Home. This refers to an Internet connection where even the very last part up to the home consists of fiber optic cable (and not a telephone cable or coaxial cable). The advantage of fiber optics is the high bandwidth capability over a long cable distance. In other words, FTTH offers high internet speeds, up to 1 Gbps (gigabit per second).
A full rack in a data center is a complete, standardized server cabinet. In the most common 19-inch width variant, the full rack is 42 to 47 rack units high. One Rack Unit (RU or U) is the standardized unit of 1.75 inches high. Installing a server, router and/or related IT hardware requires one to several units, depending on the height. The depth of a rack such as a full rack is usually 100 centimeters or 120 centimeters. In addition to a full rack, colocation in a data center allows you to rent either a half rack or another size shared rack.
Gbps is the abbreviation for gigabit per second. Another shortened designation is Gbit/s. It is the unit for the speed of data transmission, particularly applied in relation to the internet and computer networks. The unit 1 Gbps represents one billion (1,000,000,000) bits per second. For example, a normal (business) internet speed is 1 Gbps, while the better colocation and data center services also offer 10 Gbps and 100 Gbps.
A half rack, 1/2 rack or semi rack space refers to renting 21 rack units in one of the 19-inch server cabinets in a data center. In addition to your server, you also place the router and related hardware in there to keep your online business operations running full-time. A full rack, when it comes to the commonly used 19-inch variant, consists of 42 to 47 units.
If you rent rack space, such as a half rack, in a professional data center via colocation, then you are also getting services surrounding security and monitoring, climate control, fire prevention, remote hands, certification, power and backups for that.
With hybrid IT, you can flexibly use both cloud solutions and your own, traditional IT environment (on-premise). A hybrid IT infrastructure allows you to find the balance between, on the one hand, storage and computing power for applications and data in the cloud, and on the other hand, storage and computing power for the more precarious data in your own, familiar server environment. With hybrid IT, this relationship can be flexibly adjusted at any time. Cloud computing is simple, scalable and free of investment in proprietary hardware. In contrast, on-premise storage, within your own IT environment, offers greater security and control over sensitive data and its processing.
IP Connect (IP Transit)
Internet Protocol connect refers to establishing a business internet connection with an internet provider at the data center level: in that case, we call it the transit provider. They offer ‘data center internet’ with more options, capacity and speed than standard business internet, such as at the office. So if you rent rack space, you can request an IP connect for your internet connection: an IP transit within your rack. With this, you have a more advanced internet connection that is capable of higher port speeds, and usually has a speed of at least 1Gbps or 10Gbps.
IPv4 stands for Internet Protocol version 4. IPv6 obviously stands for Internet Protocol version 6. The Internet Protocol uses a standard address format: the IP address. This gives each computer a unique attribute, consisting of a numeric string. This is necessary in part because based on that unique identity, all systems can communicate with each other.
The way these addresses are distributed was first determined in the 1970s. That was the IPv4 internet protocol. This is a 32-bit string of digits, consisting of four numbers between 0 and 255, each separated by a dot. Because the more than 4 billion IP addresses provided by IPv4 are now reaching its limit, there is a successor: the updated internet protocol standard IPv6. Among other things, this standard takes into account the internet and the millions of computers still yet to enter the market.
The IPv6 internet protocol has a 128-bit sequence for alphanumeric combinations (i.e. consisting of numbers as well as letters) separated by a colon each time. This hexadecimal calculation offers longer, but more importantly, far more unique IP addresses: 340 sextillion to be exact. A sextillion… is a number consisting of a whopping 39 digits.
In addition to this essential, future-proof extension, IPv6 allows for faster, more efficient connections between unique devices. In fact, unlike with IPv4, direct communication is possible between IPv6 addresses.
IT infrastructure is the set of hardware, software and resources needed to share, transport and manage digital data. The dissemination and application of this data supports businesses, organizations and consumers in their daily information provision and operations.
kWh is the abbreviation of kilowatt hour. This is a unit for the consumption of electrical energy, such as when renting colocation in a data center. A power consumption of 1 kWh is equivalent to 1000 watts of energy for one hour. kWh may indicate, among other things, the power consumption of a device or appliance. For example, if it is 500 watts, then it consumes one kWh in 120 minutes. The prices of colocation packages and subscriptions can be either inclusive of power costs or exclusive; based on consumption of x euro cents per kWh.
It is a basic concept when it comes to connectivity: a Local Area Network (LAN). A LAN is a local computer network. An everyday example is the home or office network, where multiple residents, employees or other users use the same LAN with their devices such as laptop, mobile, smart TV and printer. These devices can communicate with each other, and computer files and data can be quickly exchanged between them within the LAN.
A home or office network – LAN – with internet access also needs, at a minimum, an internet modem and a (Wi-Fi) router. These are usually combined in one device. Instead of using wiring, Wi-Fi can also connect the various devices within the LAN. This is usually referred to as a wireless LAN, or WLAN for short.
LOA is the abbreviation for the term Letter Of Authorization – or Letter Of Authority. With colocation, it is a practical permission for the implementing party, such as the colocation service provider, to establish a connection, such as a cross connect or metro connect. In that case, the colocation service provider receives an LOA, issued by the owner of the other rack, through the requester of the rack connection.
An LOA allows a party to easily outsource the installation of the required patch(es) to the colocation provider. The authorization document provides established frameworks for this purpose. An LOA typically includes at least the following information: which location, rack, patch panel, port, connector type and cable type.
Low latency refers to the smallest possible delay in the transmission of data over an internet/communication network. In any digital data transmission, latency is inevitable, but in high-quality networks, the emphasis is on achieving low latency. Indeed, this contributes greatly to the fastest possible network connections.
A metro connect literally goes further than a cross connect. A cross connect in a data center is an indoor, direct connection from your server rack to a third-party rack. Likewise, a metro connect is a direct connection from your (server) equipment, but going outside the data center. Namely to a different building, usually also a data center, located within the same metropolis.
Applying for a metro connect also requires a LOA: a Letter of Authorization (see earlier on this page).
A network (connection) is a communication network for exchanging data between computer systems and electronic devices. The network connection is intended for network communication between companies and/or for the connection to the internet.
Patching in the context of colocation is the custom configuration of network connections into (and out of) a server rack. A patch is a schematic-based modification for establishing a new network connection, using physical elements such as network cables, patch panels, switches, network ports, routers and/or modems.
The rack in a data center is the central point for establishing a variety of connectivity links to other racks, and to and from the network equipment contained therein. In addition to a redundant setup with the aforementioned patch components, cable management is also part of the orderly layout of the server rack.
PDU stands for Power Distribution Unit. This professional power strip distributes and regulates electrical power to the servers and other hardware in the racks of the data center.
These power distributors provide sockets for data center equipment. There are also PDUs that have the capability of real-time monitoring, with remote control. The most modern versions of the PDU thus also offer power regulation, surge protection and real-time visibility into power consumption.
With a private cage in a data center, a company or organization has multiple server racks enclosed by transparent fencing. Furthermore, a private cage has the same features as a private suite. The latter is a server room that is not only enclosed, but even completely walled off.
With a private cage, the company can arrange and set up the rented array of full racks themselves, fully customized, with their own servers and hardware under their control. Are you as a company renting multiple racks in a private cage? Then you are also getting other important colocation services. Think of essential data center services such as around security, monitoring, (emergency) power, fire prevention, cooling and certification.
A private corridor provides a company or organization with a shielded space in a data center. With this colocation option, a (smaller) space with the racks containing the company servers is only accessible to the company’s own employees. A private corridor otherwise has the same features as the larger private cage and the private suite.
A private suite in a data center provides companies with their own, literally fully walled off and therefore enclosed space with racks for their servers. That private colocation space consists of a smaller or larger array of full racks. The company can arrange and set up the private suite itself, fully customized, with the necessary servers and hardware.
In addition to the racks, the data center and/or the colocation service provider will provide all other redundantly implemented colocation facilities in and around the private suite. Such as in the areas of security, monitoring, (emergency) power, fire prevention, cooling and certification.
There are several advantages of renting a private suite, fully separate from other data center customers. These include secured growth opportunities, protected access by own employees only, and a high-end corporate image.
A rack in a data center is a cabinet for safely and efficiently placing servers and other hardware for IT environments, internet and telecommunications, with a standardized width of 19 inches.
A (full) rack is typically 42 to 47 Rack Units (RU or U) high. Either way, each colocation rack has a height that is a multiple of 1U. Each such individual unit has a width of 19 inches, or 48.26 centimeters. The height of an RU is 1.75 inches or 4.445 centimeters.
1U fits one flat server or other IT hardware with that same standard height (1U). Obviously, 2U, 3U and 4U servers and hardware also require as many units.
The depth of a rack is usually 100 centimeters or 120 centimeters. Furthermore, colocation racks have rails for mounting the devices. Other terms for these rails include rack strip or rack profile. A rack strip is 0.625 inches or 15.875 millimeters wide. This leaves a space of 17.75 inches or 45.085 centimeters for the device itself; the area behind the front panel.
If you rent a rack, then it does not have to be a full rack, it can also be a shared rack, such as a half rack.
Rack space is a familiar term in the context of colocation. Racks are standardized server cabinets set up in a data center. Rack space rental to place your server(s), router(s) and other IT equipment can be in the form of a full rack, a half rack or any other size shared rack. The fixed unit that makes up racks is the Rack Unit (U). One unit in a (most commonly used) 19-inch server cabinet is 1.75 inches (4.445 centimeters) high.
Rack space is accompanied by other data center and colocation services such as power supply, business internet and redundant network connections.
More information? See our comprehensive explanation page on rack space.
Rack Unit (U) or (RU)
A rack unit, abbreviated U or sometimes RU, is the standardized designation for the amount of rack space taken up in a typically 19-inch server cabinet. The width of a rack unit is the aforementioned 19 inches or 48.26 centimeters, and the height of one such unit is 1.75 inches or 4.445 centimeters. A full rack of these dimensions is between 42U and 47U high. A 47U server cabinet therefore contains 47 times one rack unit. 1U fits one flat server (1U) or one other unit of associated 1U network equipment. Obviously, you need two rack units for a 2U server, and so on. The depth of a rack unit is 100 centimeters or, available from the better colocation service providers these days, rather 120 centimeters.
A redundant (fiber) uplink provides higher availability, as in this case, the network connection in question consists of (at least) two physical connections with (at least) two different network paths.
An uplink takes care of the network connection between a local area network (LAN) to a larger and wider network (Wide Area Network, WAN). The server hardware includes uplink ports for this purpose; for connecting the network cables.
Colocation services in a data center should be redundant. The aim is to ensure so-called availability on a permanent basis. Redundancy should keep servers, hardware and associated IT infrastructure structurally working and online. Even in case of emergencies resulting in an unexpected failure of (part of) the network and basic systems.
Redundancy is the result of multiple implementations of all primary data center facilities. Secondary backup systems, network connections and equipment will instantaneously take over these facilities in case of emergency. Think of standby generators for emergency power and secondary internet connections. For example, the specification N+1 indicates that there is one more facility available than would actually be needed.
However, redundant colocation services are also linked one-to-one with smart physical design. Such as a literal physical separation and the use of different cross-paths in the wiring and connections of network and hardware. This ensures multiple, independent connections, in order to exclude any single-point-of-failure (a term also used in the Netherlands, often abbreviated to SPOF).
Remote Hands (Smart Hands)
Remote hands is a server room service for having actions performed remotely in your server rack, by expert data center staff. These are usually routine tasks relating to your own IT network equipment. Think of things like visual inspections or installing some simple wiring or connection. But remote troubleshooting of minor failures is also a typical remote hands service. Another term for remote hands, in the context of remote physical assistance in a data center, is smart hands.
Having relatively minor technical operations performed on demand for your IT equipment in the rack saves you a lot of (travel) time. Moreover, in time-sensitive situations, especially at unfavorable times, remote hands are an almost necessary service.
A router connects multiple computer networks together. This enables digital devices to literally find the right route from the local network they are in to the internet. So this hardware allows multiple devices to link to the same internet connection. A network router controls these internet connections using the Internet Protocol (IP) and associated IP addresses. The router can also connect local networks and the various devices connected within for the purpose of mutual sharing of data, information and files.
A server is a computer that serves other computers; provides storage and/or processing of files and data; and keeps websites, domains, email and other web traffic online. In addition to hardware, the designation ‘server’ can also refer to software. After all, there are different types of servers. First, in the context of colocation, there is the physical web server. With its internal operating system, combined with software, it gives websites a place on the internet so that they are online. Another type of server is the storage server, such as those of cloud services that also offer online storage space.
A server room provides rack space for servers and other equipment belonging to the IT infrastructure. This rack space is located in a data center. In addition to the server racks, there are optimal conditions there for the hardware to operate safely and continuously. Providing a server room involves professional services focused on power supply, cooling, fire safety, security, certification and remote hands. Customers of a server room are businesses and other organizations; they can run their equipment there centrally, cost effectively and reliably.
More information? See our comprehensive explanation page on server room.
In the IT world, the term shared colocation means a shared rental of rack space – server cabinet space – in a data center. With shared colocation, companies and organizations can store their (server) hardware efficiently, cost effectively, securely and redundantly. This is done in a data center, which has a full focus on providing physical rack space and an uninterrupted power supply, among other things.
With colocation, companies share their rack space with other customers. This way, they do not have to invest in the necessary premises, infrastructure and other facilities, and manpower that help keep the servers and applications structurally online.
A shared rack is shared rack space: an x number of Rack Units (RU or U) in a server rack set up in a professional data center. Companies rent this standardized rack space to securely place their own servers, routers and related hardware. In addition to a shared rack, colocation is also possible with the rental of a full rack. Renting rack space in a data center, such as a shared rack, involves much more than just the physical use of a portion of that server rack. After all, colocation services such as backup power supply, cooling, fire safety, monitoring, remote hands and certification are also an integral part of it.
An uplink in the context of colocation refers to associated network connections from a Local Area Network (LAN) to a Wide Area Network (WAN). Within server setups and other computer networks, an uplink almost literally ‘goes up’ from a small, local station to a higher network. The server hardware includes uplink ports for this purpose; for connecting the network cables.
UPS is the abbreviation for Uninterruptible Power Supply. In short: UPS is a provision for an uninterrupted supply of power. This is a downright necessary feature for servers in a data center. After all, these servers must keep websites and other internet applications, including many basic ones, running continuously.
In case of outages or hiccups in regular power supply, UPS will need to prevent sites and apps from going offline, as well as possible data loss and other hardware and software failures. The backup emergency power supplies within the UPS are ready to take over the mains supply instantaneously, if necessary. Uninterruptible Power Supply often uses generators and charged batteries as a temporary power buffer.
To create multiple LANs on a larger scale and thus cost effectively, there is VLAN: a Virtual Local Area Network. This offers the possibility to digitally separate each LAN, rather than having to separate each LAN through physical wiring.
A LAN is a shielded, local computer network within which connected devices can quickly exchange data and files. Through VLAN, a large number of these independent LANs can be created in one fell swoop. In this virtual partitioning, each individual Local Area Network is given its own digital attribute.
A VLAN also provides rapid expansion capabilities for LANs. Suppose a business wants to link its LAN set up in its data center rack to its office elsewhere in the country. The colocation provider can then assist with this without having to do all kinds of physical wiring work. Instead, a VLAN can be assigned that traverses this route quickly and efficiently, within the colocation provider’s pre-existing network wiring.
VRRP stands for Virtual Router Redundancy Protocol. This is a redundant network protocol for routers. If the primary router or one of its interfaces fails, then VRRP automatically takes care of a so-called failover. In case of failure of the primary router’s functionalities, VRRP instantaneously engages a secondary router for this purpose.
With VRRP, the spare router is configured exactly according to, and directly linked to, the primary one. This way, these two physical, redundantly separated routers virtually form one and the same router. Thanks in part to an identical virtual IP address and identical default gateway for the firewall, the backup router can immediately take over from the primary router if necessary. This protocol increases the availability and reliability of the network connection.
In Wavelength Division Multiplexing (WDM), a single fiber optic cable consists of different wavelengths of light, each with a color. The input color is the same as the output color. On the input side, light from each individual wavelength/color is given a unique source of information each time. This way, a WDM fiber, over a single cable can simultaneously provide multiple, independent (internet) connections.
Using multiple connections through a single fiber is an efficient and therefore economical solution for those who need multiple, or even many, separate connections. After all, apart from an investment in the connection equipment, via WDM you only purchase one physical fiber optic pathway – instead of a separate pathway for each individual connection. However, you do depend on that one WDM fiber for all those (internet) connections.
With WDM Connect, there is a distinction between CWDM (Conventional/Coarse Wavelength Division Multiplexing) and DWDM (Dense Wavelength Division Multiplexing). With CWDM, a maximum of 18 colors can pass through the fiber optic cable. With DWDM, up to 160 colors are possible.
Want to know more about the IT world of colocation, data centers and rack space?
Contact us without obligation.