Glossary

Throughout our site, we try to use plain English but sometimes we have to use words and phrases that might not be generally understood. When we do, we'll expand on them here and often provide some further reading on them if you want it.

Air Gap (Networking)

A Network Air-Gap is the ultimate in Network seperation. It comprises a hardware device or specific network design that provides a seperation between two networks or systems. A Hard gap is hardware based and will intermittently connect two or more network segments to a common "output" based on a timer, either for specific times of each day or for a set amount of time on each "input". It provides a very secure method to allow access to restricted networks and when configured properly will defeat simultaneous connections from one network to the other or to the common output. LogiTEL manufactures a small, two-input device. More information can be found here.

NOTE. Networks seperated by a DMZ or firewall are not considered Air-Gapped as the correct compromise of the security between them could provide a network through-path.

Cloud Computing

A physical server is bound by the capacity of its components - when it has filled its disk, it can't store any more information and you have to physically alter it to improve matters. Same goes for memory, when you run out you have to add more physical RAM modules. If there are no free slots for these, you need to change the motherboard or worse still change the entire server for one with a bigger chassis. Another aspect of physical machines is that unused capacity is wasted. If it is only 10% busy, you can't re-direct the other 90% to servers that could benefit from it. Physical servers tend to have poor efficiency in this respect and are a major headache to expand - there is also a recycling problem - what to do with the old kit?

Cloud computing is a sea-change in the thinking about how we do computing. Cloud uses VMs - lots of them - all interconnected on a smaller number of big physical servers. VMs are not bound by the things that affect physical machines. As their needs grow and shrink, the capacity can be flexed up or down without rebuilding the server. If it needs more memory or CPU, you simply adjust with a slider and restart it (in these days or "hot-swap" you don't even need a restart!). You never have to rebuild your servers because you have out-grown them, now they grow and change with your business needs.

Cloud brings a utility based model to computing because each VM only uses as much resource as it needs - much like a lightbulb - it only uses power when it is working (i.e. making light). With the physical model, the bulb costs you even when it wasn't doing anything. Also, if you wanted more light, you have to change it a for a brighter bulb. With the Cloud analogy, you would just add more brightness to the existing bulb - Utility Computing is like a dimmer switch for servers :o)

Cloud stops you thinking of compute in limited terms of individual servers, in favour of a big lump of compute that gets used as necessary. It also has some other tricks to further improve efficiency, including:

  • "Thin-provisioned" disks which avoid having unused space on individual disks - the disks get what they need when they need it and no more.
  • "Over-commit" on memory allows the host to provide more memory to VMs than it actually has.
  • Prioritisation allows the host to dynamically suspend some VMs and give their resources to more urgent tasks.
  • V-Motion alows VMs to be moved between Hosts almost seamlessly without disruption to service.
  • Snap-shotting allows VMs to be frozen at a point in time, then if anything goes wrong you can instantly restore to an earlier, "known-good" version of the server
  • Easy backup of the entire machine - it's just files on a Host - easily copied to a safe place.
Clever mathematics and software takes care of any conflicts.

Cookie

Cookies are small packets of information that servers can place on your PC and then read back at a later stage. This is quite nice for making a rich web experience e.g. for remembering things about you personally and tweaking the web page accordingly. Sometimes, other servers can read these and so work out where you have been - it is a violation of your privacy to do this without letting you know, let alone gaining your permission. Since 2012 across the EU, it has become illegal for any sites to do this without letting you know and giving you the option to opt-out. UK has perpetuated this legislation after its exit from the EU political bloc.

Patsy

"Patsy" is a traditional term for a fall-guy or middleman that is expendible in the large scheme of things. In IT terms it usually means a program, data area or a server stood out on it's own to act as a go-between or staging area between two (or more) networks or systems.

Photo-Voltaic Array

"Solar cell" is a phrase that can mean many things, basically that energy is obtained from the Sun in useful amounts. This might be a case of warming air or water for domestic use in a series of radiators enclosed in sealed panels. The sunlight acts on the dark panels and shielded from the outside (like a green house) the heat builds up and a supply of hot water or air is generated. This is great if you want a shower or to heat your living room, not so good for running a data center. Others (called Solar Furnaces) are large scale setups that focus sunlight to an intense point and use this heat to boil water and then drive turbines - the concentrated sunlight replaces the coal, oil or gas. These types of power station are quite rare, controversial and confined to perpetually sunny parts of the globe.

The Solar Cells mostly seen on roofs and attached to street equipment are Photo-Voltaic, made with special materials that convert the Sun's light directly into electricity. This electricity is then converted into the correct type to power common equipment, for example - an inverter which will take a low voltage source (say, 12v DC) and output mains power (220v AC) into which most low-power mains equipment can be plugged.

Server

A Server is a type of computer which deals with one or two specific tasks at speed and scale rather than a general system like your desktop PC. Most often they do not have a keyboard, mouse and monitor like a PC as they are intended to work for long periods (maybe years) without anybody attending them. They are optimised for network and disk activity and usually have quite large memories. Some perform special functions and might possess a lot of very powerful processing cores. Web pages and e-mail use servers to provide their services. A specific type of server called a "Filer" specialises in storing large amounts of data. Database servers store their information in large specially structured files and often share these files with other database servers so they can take over if there is a problem ["clustering"] or share the load ["load balancing"].

Virtual Machine (VM)

A VM is a server that does not exist as a collection of hardware items; CPU, memory, disks etc... rather, these elements of the server are emulated by software on a physical server called the "Host". This host then can run several VMs at the same time. These VMs generally have no idea they are not "real" computers or that they are sharing things like CD drives and hard disks. VMs are a very efficient alternative to providing a server over building a physical box. Individual servers (physical or virtual) can spend a lot of their time doing very little, it follows then that their spare resources could be given to other tasks. The host approaches this same issue from a slightly different angle - it gives the VM only what it needs rather than re-assigning its excess. This then leaves the remainder to be used on other servers. VMs run very efficiently with rarely any difference in the performance of a virtual server from its physical counterpart. All the VMs on a host will co-exist and share the physical parts of the host with each other.

VMs are the building blocks of cloud environments. As the requirements of a server grow it cannot reach a binding limit on available resources like a physical server can. The machine can be flexed up or down to consume more or less of the host's resources without rebuilding.