This page contains reference information broadly related to computers, networking and the Internet
| SCSI hard drives systems cost four times more than IDE systems of the same
capacity. It's so tempting to say "IDE will be OK" but how do you make the right choice? Some Definitions:- |
IDE |
Integrated Drive Electronics | Before IDE, disk drives were simpler, mostly mechanical devices that were controlled by logic cards that plugged into motherboard slots |
EIDE |
Extended Integrated Drive Electronics | Starting in 1994, this extended the speed and capacity of the original IDE standard. The term EIDE covers the standards ATA-2 to ATA-7 |
ATA |
AT Attachment(AT comes from the 1984 IBM PC/AT and stands for Advanced Technology) |
Also called ATA-1 this is the official name of the standard known popularly known as IDE |
ATAPI |
ATA Packet Interface | This is the name of the CD-ROM and Tape Drive interface that shares the same cable as an IDE disk drive |
SATA |
Serial ATA | The new IDE standard which uses a thin cable and can transfer data at 150mbps |
SCSI |
Small Computer System Interface | Defined in 1986, this is the other type of hard disk interface |
I use the term IDE to mean all variations of the ATA standard.
The Advantages of SCSI
| Modern IDE drives are fast, but SCSI drives are faster. Here are some advantages of SCSI over IDE:- | |
| All device on the SCSI bus can be active
simultaneously. With multiple SCSI hard drives on the same bus, although only one
device at a time can be transferring data, the others can be moving
their heads into position and preparing data for transfer. Only one device at a time can be active on an IDE cable bus. |
|
| Mainstream SCSI drives spin at 15,000 rpm while IDE drives spin at 7,500 rpm. A high spindle speed means faster access. | |
| There can be 6 devices on a SCSI bus while and
IDE bus can only have 2. A computer often contains 2 x SCSI buses. It's unusual for a computer to have more than 4 x IDE buses and, because of the requirement to have a CD-ROM drive and not wanting to put 2 drives on a single cable, computers rarely have more than 4 x IDE drives. |
|
| In my experience, hot-swapping works better with SCSI than IDE. Windows certainly doesn't like drives appearing and disappearing but the hardware RAID controller should be able to hide this. | |
| A lot of high-end equipment such as RAID
controllers and backup devices only come with SCSI interfaces. Firewire and USB now offers more attractive alternatives for devices, such as scanners, that used to use SCSI. |
|
| A RAID 5 hot-swap 15,000rpm SCSI disk array is the best way to get a fast, fault-tolerant hard disk system for and this is probably the biggest difference between an entry-level small business server and a high-end server. | |
A Guide to Making Your Choice:-
Type Of Server |
IDE |
SCSI |
| Domain Controller with 20 or less users |
|
|
| Domain Controller with 50 or
less users with no disk-intensive server application such as a database |
|
|
| Domain Controller with more than 20 users and a disk-intensive server application |
|
|
| Domain Controller with more than 50 users |
|
|
| Web Server with up to 500 users |
|
|
| Mail Server with up to 200 users (less if using Exchange) |
|
|
| Terminal Services Server |
|
|
| SQL Server |
|
| You can run Windows 2000's Performance Monitor and view the disk performance
counters to determine how well your disk drives are coping but this is usually
only possible after you've already made your decision and bought the drives. Hopefully
it will confirm
you've done the right thing.
Mix and Match Maybe your company needs a total of 200 gigabytes of hard disk storage but only 30 gigabytes needs to be the fast-access type for your database. In this case it could be cheaper to have 2 x 200gbyte IDE drives in a 200gbytes RAID 1 array and 3 x 18gb SCSI drives in a 36 gigabyte RAID 5 array. With all hard disk specifications you should consider how fast your company's data storage requirements are growing and how easy it's going to be to increase your disk drive capacity when the time comes.
What's iSCSI? Internet Small Computer Systems Interface is an interface standard that converts normal SCSI bus traffic to a serial data stream which is then encapsulated into TCP/IP frames for sending over the Internet. iSCSI is supported by Windows Server 2003 but what possible use could it be? Storage Area Networks are used by large organisations as a central repository of all their data which is usually only accessed by servers and is an alternative to storing it on drives directly attached to servers. Geographically separate offices can be on the same SAN and be accessing data over large distances. Super-fast Fibre Channel connections are normally used to connect servers to SAN storage devices but the maximum distance that this operates over is 120kms, iSCSI can extend the range of SANs to wherever the Internet reaches. It offers a cheaper alternative to Fibre Channel but can't operate at the same speed. It obviously makes sense to some people to have the data you want to access stored on a hard drive 1000s of miles away. |
| In 1987 professors Gibson, Katz and
Patterson at the University of California published the first paper
describing a
Redundant Array of Independent Disks (RAID). Their research was sponsored by
IBM who hold many patents on RAID and who were the first to produce a
commercial RAID product in 1990. RAID is a great example of lateral thinking: instead of solving a difficult technical problem you find a way to make it far less important. This problem was that computer hard drives are delicate mechanical devices which, when in use, operate at a high stress level subjecting them to constant wear and tear and yet these same drives are used to store precious data as well as an operating system setup which is time consuming to restore. The ever-present prospect of a hard disk failure limited the degree of trust that could be placed in computers. RAID has developed many variations and options but the essential characteristics of RAID are:- |
The "I" in RAID originally stood for "Inexpensive" until the RAID Advisory Board decided that "Independent" was more appropriate. |
||
| 1 - | 2 or more hard disks are linked together in a RAID Array which the operating system sees as a single drive. | ||
| 2 - | If one of the individual hard disks fails the RAID array continues to operate normally as far as the host computer is concerned. | ||
| 3 - | When the failed hard disk is replaced with a new one, this disk is automatically rebuilt so that fault tolerance is re-established. | ||
| The method in which disks are
combined in a RAID array is referred to as the RAID Level which is usually a
number between 0 and 10. Some RAID levels aren't used anymore and were just
a step along RAID's evolutionary road. For example the improvements of RAID
5 over RAID 4 means RAID 4 is obsolete. The following table summarises the types of RAID levels in common use today:- |
|||
RAID Level |
Description |
Advantages |
Drawbacks |
| 0 | Striping | Fastest. Consecutive data is spread across different drives and there's no Parity to worry about | No Redundancy so not strictly RAID. All data is lost if any disk fails so less reliable than a single drive |
| 1
|
Mirroring | Safe and Simple. Each drive contains an exact copy of the data on the
other. Read performance is improved. |
The storage capacity of the array equals the size of the smallest disk in the array. |
| 5 | Data Striping with Distributed Parity | High performance while still able to withstand a disk
failure. The majority of the total disk space is available for data storage. |
Requires a minimum of 3 drives. Read and write performance slows when a drive has failed. |
| 6 | Data Striping with 2 lots of distributed parity | Any 2 of the drives can fail simultaneously without causing data loss! | It requires a complex and expensive controller. |
| 10 | RAID 1 + 0 2 or more RAID 1 Mirrors are combined into a RAID 0 stripe set |
Gives great performance while maintaining high fault-tolerance. | Requires a minimum of 4 drives. |
| Other RAID Levels you should know about | |||
| 0 + 1 | 2 RAID 0 stripe sets are combined into
a RAID 1 mirror |
Same performance level as RAID 10 | Less fault-tolerant and longer rebuild
times than RAID 10.
|
| JBOD | Just a Bunch Of Disks (The RAID Advisory Board hasn't got around to changing this unflattering acronym.) |
I can't think of anything good to say about it, buy a bigger disk instead | No redundancy or striping. Smaller disks are just combined, or spanned, to give one large capacity disk |
| What Happens when a Drive Fails? This is when your RAID array is earning its money so the answer should be "Nothing, as far as the users are concerned". In reality the data transfer rate of the array drops - especially with RAID 5 where missing data on the failed drive now has to be regenerated, on the fly, from the parity information. From a network management perspective swift action is now required as another drive failure will cause total data loss (unless you have RAID 6). The risk is now greater than if you weren't using RAID because:- |
|
| 1 - | If 2 or more disks are still functioning in the array there's now a greater overall chance that one will fail. |
| 2 - | The same type of hard drives, purchased at the same time and subject to the same operating conditions have an uncanny habit of failing within a few months of each other. |
RAID Features that Help Restore Redundancy after a Disk Failure
AlertsIf RAID works the way it should, a failed drive could well go unnoticed. A RAID system, therefore, usually has a monitoring system that detects a disk failure and tries to attract attention by:- |
|
| Making the computer beep. | |
| Sending out network alerts. | |
| Writing to the Event Log | |
| Sending emails to administrators. | |
Hot Swap DrivesIt can be a real problem finding an opportunity to shutdown a busy server in order to replace a hard drive. RAID systems often allow an individual drive to be removed and replaced while the power is on - a process called hot swapping. Hot swapping uses a high quality connector system where the earth pins make contact first and break contact last as well as a RAID controller that is prepared for drives that appear and disappear without warning. Hot SparesA Hot Spare is a drive plugged into a RAID array that's not doing anything. It's on the "substitute's bench" ready to take over automatically if one of the active drives fails. A Cold Spare, on the other hand, is a spare drive locked away in a cupboard somewhere. Replacement Drive PolicyThis is something that a Networking Department has and not a RAID feature. Whether you've got hot or cold spares for your RAID array you should have worked out in advance what action you'll take when a drive fails so as to keep up the full protection offered by your RAID system. This isn't as simple as "when one drive fails we'll just buy another". Hard drives have been undergoing rapid development in the last few years leading to increased performance and capacity and so particular models often only stay current for a few months. Trying to find the same model drive as one bought 2 years ago is often impossible and RAID does function best when all the drives are the same model. A sensible policy might be to buy 2 extra drives as cold spares when the RAID array is bought new and when the last of these has been put into service plan to replace all the drives, including cold spares, with a current hard drive model sometime soon. |
|
What about Software RAID?
| Software RAID means the operating system
handles the RAID controller functions in software - you still have to buy
all the physical drives. I'd give Windows 2000 Server's implementation of software RAID 10 out of 10 for effort but I would never recommend anyone to use it because:- |
|
| It was designed to save you the expense of hardware RAID controller cards. These are now cheaper and often built into the motherboard. | |
| It can be difficult to boot from a software RAID array, especially if a drive has failed. | |
| Every engineering bone in my body tells me the best approach is to use dedicated hardware for the RAID controller and leave the operating system out of the picture. | |
SCSI RAID or IDE RAID?
| From 1990 to 1997 only SCSI RAID existed.
RAID was elitist, high-end and provided a good profit margin for its
suppliers. There's no technical reason why SCSI drives are more suited to
RAID than IDE drives and so Promise took RAID "down-market" when it was one
of the first to develop an IDE RAID controller. Eventually Adaptec, a name
with strong links to SCSI, released IDE RAID cards and today they are
commonplace and a feature often incorporated into motherboards. "SCSI RAID or IDE RAID?" is the wrong question. "I need RAID, of course, but do I need SCSI or IDE drives?" is what you should be asking. There's a discussion on this here - find out and get back to me. |
What's the best way for a Small-to-Medium sized business to take advantage of RAID?
| 1 - | Decide that you will implement RAID the next time you have a server upgrade. It may seem scary and a hassle but it's nothing compared to the trouble a failed server hard drive will cause you. | |
| 2 - | Choose SATA RAID Level 1 - Disk Mirroring. | NOT SCSI RAID, Serial ATA is a new enhancement of the IDE standard |
| 3 - | Decide how much data storage space your company needs now and predict what this will grow to in 5 years. If the figure you get is greater than 300Gbytes you'll have to separate out the unchanging data, such as past correspondence and archives, which doesn't need to be on RAID to leave just new and changing data. If you still have a requirement of more than 300Gbytes then your needs are outside the scope of this article and you'll need a more high-end solution. | |
| 4 - | Buy 3 identical models of SATA hard drives with
a minimum storage capacity of 160Gbytes (300Gbytes is the current
maximum capacity of a single IDE drive). That 3rd drive - the cold spare - is important as we're going to do a nifty trick with it. |
|
| 5 - | Buy an Adaptec 2410SA Enclosure kit which is a combined SATA RAID controller card and hot-swap enclosure available from Misco for £333. It's got 4 hard drive slots but in this example we're only going to use 2 of them. You could use the 3rd and 4th slots for extra non-RAID data drives. It's a pity there aren't 5 slots as then you could have an additional RAID 5 data array. | ![]() |
| 6 - | Start by installing the Adaptec RAID card and
the Hot-Swap drive enclosure with no drives inserted but with all the power
and data cables connected and install the drivers
for the card and utility programs onto your original server hard drive. Then make a copy of your
existing hard drive onto one of the new SATA drives. Do this by connecting
a SATA drive as a secondary master or slave drive. If your
current server doesn't handle SATA drives use an IDE-to-SATA converter
cable. Use BootitNT, Ghost, Acronis or the server version of Partition Magic
to copy the server disk image to the SATA drive. It's likely that the new drive has a higher capacity that the original so use the same partition manipulation program to expand the partition so that it occupies the whole of the SATA disk. |
|
| 7 - | Fit the SATA drive into Slot 1 of the RAID enclosure, remove the original server drive - it's now officially retired - and boot from the SATA disk to check that everything works OK. Then fit a second SATA drive to a drive caddy and insert it into Slot 2. Use the Adaptec utility to join the 2 drives into a RAID 1 Mirror array and the process of building the second drive into a mirror of the first will start. It does this in the background as the server continues to operate as normal and takes around 2hrs for every 100 Gigabytes. When complete you'll have a fully functional RAID 1 array. |
| We all know we have to make
backups, but no one enjoys doing them. Backing up the whole server every day is usually too hard and so the task is split into 2:- |
|
| 1 - | Make daily backups of new data or data that's changed. |
| 2 - | Make periodic backups of the operating
systems and installed software. It might be appropriate to do this weekly, monthly or even 6-monthly, depending on how quickly such a backup goes out-of-date. |
| I can't help you with 1 but, with your Adaptec 2410SA now successfully installed, here's how to make a complete backup of the server hard drive:- | |
| Slide the drive out of Slot 2, stick a Post-It note on it and write the date and the words "Out Slot 2 in Slot1" and lock it away in a cupboard. | The Mirror Crack'dAn Agatha Christie novel |
||
| Slide the Cold Spare into Slot 2. | |||
| Oh! Sorry, that was it, we've finished. Did I go too quickly? Let's do it again. | |||
| 6 months later:- | |||
| Get the Cold Spare from the cupboard and read the Post-It note and see that it says "...in Slot 1". | |||
| Slide the drive out of Slot 1, stick a Post-It note on it and write the date and the words "Out Slot 1 in Slot 2" and lock it away in a cupboard. | |||
| Slide the Cold Spare into Slot 1. | |||
| This disk-rotation method ensures each drive has equal wear and if you're worried about RAID protection not being available during the, perhaps, 4 hours it take to rebuild the Cold Spare, remember that you have a very recent full backup locked away in the cupboard. | |||
| Someone who's performed their
company's backup procedure hundreds of times may never have had to do a full
system restore and, because of the inconvenience it would cause and the lack
of spare equipment, they've never even practised a restore: another thing to
worry about on the day your business is depending on it. So you've been successfully taking advantage of the hot-swap and auto-rebuild abilities of your RAID system to make the easiest ever complete server backups and one day disaster strikes: a rogue patch planted by a hacker on the Windows Update site is downloaded and installed on your server which then trashes the system folder making the server unable to boot and the boss is looking to you to fix it. Here's the drill:- |
|
| Turn off the server and slide out the drives in Slot 1 and Slot 2. | |
| Get the Cold Spare from the cupboard, slide it into Slot 1 and turn the server back on. | |
| Restore yesterday's data backup then tell the
boss we're up and running again. (Oops, better disable Windows Update first.) If that took more than 15 minutes you must have been milking it for dramatic effect. There are still some clean-up jobs to be done. |
|
| Slide one of the disks you pulled out into Slot
2 and wait for it to be rebuilt. During the hours this rebuild takes you
have no backup in the case a second disaster. A good time, perhaps, to
reflect on a 4-disk rotation scheme where a 2nd Cold Spare is stored
off-site. You could now also attach the 2nd scrambled drive to another PC and make a data backup in case there's been any important new data since yesterday. |
|
| Slide the drive out of Slot 2, stick a Post-It note on it and write the date and the words "Out Slot 2 in Slot1" and lock it away in a cupboard. | |
| Slide the remaining drive into Slot 2. | |
| This split-mirror backup trick only works with RAID 1 and, because of the speed and ease with which a system can be restored, it beats the standard method of using a tape drive. | |
CAT5, CAT5e, CAT5e Extended, CAT6, CAT7, 350mHz
cable or optical fibre?
|
|
| 1 - | Run multimode optical fibres between the servers and the network switch alongside the copper cable. The fibre's for future use so, for the moment, leave the ends unterminated. (Optical fibre has a far higher theoretical data rate than copper cable.) |
| 2 - | Locate the servers close to the network switch so that any future cabling will be a simple matter. |
| Remember that it's only the cable
we're trying to future-proof, you don't have to go to Gigabit and beyond
with the rest of the network equipment just yet. This equipment can be
upgraded relatively easily when the time is right however, though not
essential, I would also recommend getting outlet sockets, patchfields and patchcords that
match the speed of your cable. So back to the question of which cable to get. Obsolete CablesThinnet and Thicknet - coaxial cables.CAT3, CAT4 and CAT5. Yes, even CAT5 is on the scrapheap as it's not guaranteed to operate at 1000mbps especially over 100m runs. CAT5 VariantsThe CAT6 standard was only ratified in June 2002 so many cable manufacturers tried to anticipate this by producing their own improved versions of CAT5 which were intended for use with Gigabit Ethernet.If you've installed CAT5e cable capable of data rates up to 350mHz then don't worry, you'll very likely be OK but, from now on, CAT6 cable is the one you should use. Category 6 Networking CableGigabit Ethernet requires the individual copper pairs to operate at 200mHz. Gigabit Ethernet uses all 4 pairs in a network cable - 2 for sending and 2 for receiving - and a 1000mbps signal can be multiplexed over 2 x 200mHz cables - it can, the mathematicians say so and it's no use arguing with them. So a typical CAT5e cable operating up to 350mHz is good for Gigabit and a little bit extra besides. What CAT6 cable does, in addition to raw bandwidth, is to reduce the interference, or crosstalk, between the four cable pairs by incorporating a piece of plastic running down the centre of the cable to hold the individual pairs apart. In tests this has been shown to reduce the amount of packet errors and retransmissions. The central plastic separator increases the diameter of CAT6 cable to 6mm from the 4mm of CAT5e - so remember to allow for this with the size of your conduits.Surely CAT7 or Optical Fibre Cabling is more Future-Proof?Indeed they would be. CAT7 is, so far, a theoretical standard for 600mHz per copper pair and I haven't seen any for sale yet. CAT7 will require a completely different connector system - forget RJ45 - and a meticulous installation methodology will be required to achieve the high data rates. I think it will too difficult to go much beyond 200mHz per pair and will prove easier to use fibre everywhere instead. So why not install a wholly fibre-optic network now?Because it's will cost between 3 and 5 times as much as a CAT6 network. However, there are sometimes good reasons to have fibre runs in your current network :- |
|
| A network outlet is needed which is more than 200m from a network switch - optical fibre links can be up to 20kms long. | |
| A cable run must pass through a hostile environment e.g. near heavy machinery generating electrical interference, or under water. | |
| You have a once-only opportunity to lay cable after which recabling will be prohibitively expensive. Perhaps in a new factory or a heritage building. | |
| Optical fibre is very hard to eavesdrop on and so is favoured in high security environments. | |
The Other Questions Shielded or Unshielded? Unshielded, the cable you want is called Unshielded
Twisted Pair (UTP). Solid Conductor or Stranded?Solid. Cable Box or Reel?Reel. A reel inside a box is now quite common and gets you the convenience of a box without the kinks. Be Conscious of Fire SafetyNetwork cable comes with 2 types of plastic outer sheathing: the type that
doesn't produce poisonous gas when burnt and the type that does. It's not
mandatory to use the poisonous gas free variety in all situations but I
think we should all show that we're safety conscious and pay the 30% extra
that this cable costs. In the UK this cable is called LSZH (Low Smoke Zero
Halogen) or another similar name and in the US is referred to as Plenum-Grade
(the plenum is the space above a false ceiling). In the UK you can get some here. External Grade CableIf any cable runs will be exposed to the elements then you'll also need a
smaller amount of the exterior grade version of this cable which resists the
ingress of moisture and deterioration due to exposure to ultra-violet light. There have been some recent improvements in wireless networking which have made it more attractive:- |
|
| The 802.11g standard has been ratified and this specifies a data rate of 56mbps in the 2.4gHz band. | |
| The easy-to-crack WEP security standard has been superseded by the, hopefully, impossible-to-crack WPA security standard. | |
| The 56mbps that 802.11g offers has
to be shared between all devices using a particular access point while a
device on a gigabit LAN has 1000mbps all to itself, which is a huge
difference. Although wireless networking is fantastic for roaming access, temporary or problem installations you can't beat the simplicity, performance and reliability of copper cable. |
|
AD Andorra AE United Arab Emirates AF Afghanistan AG Antigua and Barbuda AI Anguilla AL Albania AM Armenia AN Netherlands Antilles AO Angola AQ Antarctica AR Argentina AS American Samoa AT Austria AU Australia AW Aruba AZ Azerbaijan BA Bosnia and Herzegovina BB Barbados BD Bangladesh BE Belgium BF Burkina Faso BG Bulgaria BH Bahrain BI Burundi BJ Benin BM Bermuda BN Brunei Darussalam BO Bolivia BR Brazil BS Bahamas BT Bhutan BV Bouvet Island BW Botswana BY Belarus BZ Belize CA Canada CC Cocos (Keeling) Islands CF Central African Republic CG Congo CH Switzerland CI Cote D'Ivoire (Ivory Coast) CK Cook Islands CL Chile CM Cameroon CN China CO Colombia CR Costa Rica CS Czechoslovakia (former) CU Cuba CV Cape Verde CX Christmas Island CY Cyprus CZ Czech Republic DE Germany DJ Djibouti DK Denmark DM Dominica DO Dominican Republic DZ Algeria EC Ecuador EE Estonia EG Egypt EH Western Sahara ER Eritrea ES Spain ET Ethiopia FI Finland FJ Fiji FK Falkland Islands (Malvinas) FM Micronesia FO Faroe Islands FR France FX France, Metropolitan GA Gabon GB Great Britain (UK) GD Grenada GE Georgia GF French Guiana GH Ghana GI Gibraltar GL Greenland GM Gambia GN Guinea GP Guadeloupe GQ Equatorial Guinea GR Greece GS S. Georgia and S. Sandwich Isls. GT Guatemala GU Guam GW Guinea-Bissau GY Guyana HK Hong Kong HM Heard and McDonald Islands HN Honduras HR Croatia (Hrvatska) HT Haiti HU Hungary ID Indonesia IE Ireland IL Israel IN India IO British Indian Ocean Territory IQ Iraq IR Iran IS Iceland IT Italy JM Jamaica JO Jordan JP Japan KE Kenya KG Kyrgyzstan KH Cambodia KI Kiribati KM Comoros KN Saint Kitts and Nevis KP Korea (North) KR Korea (South) KW Kuwait KY Cayman Islands KZ Kazakhstan LA Laos LB Lebanon LC Saint Lucia LI Liechtenstein LK Sri Lanka LR Liberia LS Lesotho LT Lithuania LU Luxembourg LV Latvia LY Libya MA Morocco MC Monaco MD Moldova MG Madagascar MH Marshall Islands MK Macedonia ML Mali MM Myanmar MN Mongolia MO Macau MP Northern Mariana Islands MQ Martinique MR Mauritania MS Montserrat MT Malta MU Mauritius MV Maldives MW Malawi MX Mexico MY Malaysia MZ Mozambique NA Namibia NC New Caledonia NE Niger NF Norfolk Island NG Nigeria NI Nicaragua NL Netherlands NO Norway NP Nepal NR Nauru NT Neutral Zone NU Niue NZ New Zealand (Aotearoa) OM Oman PA Panama PE Peru PF French Polynesia PG Papua New Guinea PH Philippines PK Pakistan PL Poland PM St. Pierre and Miquelon PN Pitcairn PR Puerto Rico PT Portugal PW Palau PY Paraguay QA Qatar RE Reunion RO Romania RU Russian Federation RW Rwanda SA Saudi Arabia Sb Solomon Islands SC Seychelles SD Sudan SE Sweden SG Singapore SH St. Helena SI Slovenia SJ Svalbard and Jan Mayen Islands SK Slovak Republic SL Sierra Leone SM San Marino SN Senegal SO Somalia SR Suriname ST Sao Tome and Principe SU USSR (former) SV El Salvador SY Syria SZ Swaziland TC Turks and Caicos Islands TD Chad TF French Southern Territories TG Togo TH Thailand TJ Tajikistan TK Tokelau TM Turkmenistan TN Tunisia TO Tonga TP East Timor TR Turkey TT Trinidad and Tobago TV Tuvalu TW Taiwan TZ Tanzania UA Ukraine UG Uganda UK United Kingdom UM US Minor Outlying Islands US United States UY Uruguay UZ Uzbekistan VA Vatican City State (Holy See) VC Saint Vincent and the Grenadines VE Venezuela VG Virgin Islands (British) VI Virgin Islands (U.S.) VN Viet Nam VU Vanuatu WF Wallis and Futuna Islands WS Samoa YE Yemen YT Mayotte YU Yugoslavia ZA South Africa ZM Zambia ZR Zaire (now called Democratic Republic of Congo) ZW Zimbabwe COM US Commercial EDU US Educational GOV US Government INT International MIL US Military NET Network ORG Non-Profit Organization ARPA Old style Arpanet NATO Nato field