Talk:Logical unit number

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search
WikiProject Computing (Rated Start-class, Low-importance)
WikiProject iconThis article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
 Low  This article has been rated as Low-importance on the project's importance scale.

Slang usage[edit]

I am removing the bit about the "Indo-Pak sub-continent" slang usage added on Apr 15 2006 by It is a) lacking relevance to the rest of the article and b) not appropriate to the English language wikipedia. RobLinwood 21:27, 26 April 2006 (UTC)

Whats he on? of course slang is relivent to wikipedia, its ment to be a encyclapedia for everybody and peaple use SLANG! please reply... (Unsigned comment by User:Derg999, 15:00, 21 May 2007)
Quite right, nothing wrong with slang as such on Wikipedia. Nevertheless the material was correctly removed, because this particular slang usage is an alternative meaning for the word – the proper place for such material is a disambiguation page, not an article. However, as the word is not used in English anyway it does not belong in the English wiki at all. --Richard New Forest (talk) 14:10, 16 June 2008 (UTC)

LUNs vs. SCSI targets vs. Volues[edit]

VERY CONFUSING: "To provide a practical example, a typical disk array has multiple physical SCSI ports, each with one SCSI target address assigned. Then the disk array is formatted as a RAID and then this RAID is partitioned into several separate storage volumes. To represent each volume, a SCSI target is configured to provide a LUN. Each SCSI target may provide multiple LUNs and thus represent multiple volumes, which does not mean that those volumes are concatenated."

Can somebody explain this so a mere mortal can understand? What is the distinction between a SCSI target and a LUN? How are they used differently? How do you partition without concatenating? Do you leave spaces between the partitions? This is totally problematic. —Preceding unsigned comment added by (talk) 00:15, 5 August 2008 (UTC)

True, the explanation is rather confusing. I guess this article needs some cleanup. Let's see.. A LUN is a logical unit. Think of it as a bucket of bits. When you send commands over a SCSI bus (or some other bus that use SCSI commands, like FC), you send commands like "Read 50K from LUN0 starting at address 2000" or "Write these 50K to LUN0 address 5000". So a LUN is the target address, if you will. Let's imagine that you have a regular home PC with a simple SCSI controller and a single SCSI HD attached to the controller. In this case, the SCSI HD is a target and it is mapped to a single LUN. So when you read/write to the LUN, you are reading/writing to this HD. Let's take one step up, and imagine that the SCSI controller is a RAID controller, and there are two SCSI HDs connected to the controller. Now, on the SCSI RAID controller you configure so that each disk is spilt in two partitions. You take one partition from each disk and make a RAID0 volume, and then with the two others you make a RAID1 volume. Now, each of these volumes will be mapped to a LUN. So when you now write to LUN0, you are writing to a RAID0 volume that is split across the two SCSI HDs. Likewise, when writing to LUN1 you write to the RAID1 volume. Take this a step or two further up and imagine an enterprise SAN, with 50+ HDs and some advanced RAID/storage controller in front. You use the SAN management tools to organise those HDs into different volumes - some perhaps RAID5, some RAID10, some perhaps just consisting of a single disk. Then you perhaps sub-partition some of the volumes. You then map/export these volumes (or sub-volumes) so that the servers can access them. From the point of view of the servers, these available storage volumes will be LUNs accessible over the SCSI (or FC/iSCSI/etc) bus that they are connected to the SAN with.
So a LUN is a logical "bucket of bits" that you can read/write to over a SCSI/FC/iSCSI/InfiniBand/SAS bus. Depending on what is on the other side of that bus, the LUN might be a single SCSI HD in your home PC or it might be a slice of a RAID10 volume spread over 50 disks in a SAN. (talk) —Preceding undated comment was added at 17:19, 11 November 2008 (UTC).

proposed new Logical Unit Number page[edit]

I rewrote the LogicalUnitNumber page but will look for comments before posting. I tried to pay attention to language, consistent, mere mortal language all with good Wiki references. I also tried to give simple to complex real world examples. Kleach (talk) 18:49, 19 November 2008 (UTC)

A couple of questions: Are the device names correct for Ubuntu or are the RAID devices sda1 and sda2? Sould there be Wiki pages for "rank" and "concatenation"?

How does this sound?

In computer storage, a logical unit number or LUN is classically a slice or subset of a physical disk which has been formatted into partitions. Each LUN is presented to the host computer as a logical or virtual disk. With the advent of RAID technology, LUNs now have another virtual layer and slices, subsets or supersets, of RAID arrays are presented as large logical disks to the host.

A single RAID array is broken into ranks, visualize a two dimensional array of disk drives in horizontal ranks and vertical columns. Each rank acts as one logical disk storing data across all disk drives including some level of redundancy. These logical disks are then sliced into partitions just as a single physical disk can be.

Regardless of the RAID level used in the rank or number of slices per rank, these slices can be formed into LUNS in many different ways. Each slice can be presented as a LUN to the host, or slices can be combined using additional striping, mirroring, or concatenation with the resulting LUN being presented as a large logical disk to the host. Even slices from multiple ranks can be combined as a single LUN, sometimes called a 'Meta LUN'.

The key concept here is that a LUN is a virtual or logical host view into the disk subsystem. On a host system, each LUN will be mapped to a device. That device in turn may be formatted into a "cooked" filesytem or perhaps used as a raw device, "uncooked".

Example 1:[edit]

A single physical disk in a laptop computer is sliced into two(2) LUNs during the install of Ubuntu, a Linux operating system. The first LUN is used as the root filesystem mounted on "/". The second LUN is used raw as virtual memory or swap space. The Ubuntu device names may be /dev/hda1 and /dev/hda2.

Example 2:[edit]

A simple RAID array has two(2) physical drives. It is decided that the RAID level RAID-1, mirroring, will be used to create a highly available, fault tolerant disk subsystem for a Ubuntu install. After the RAID configuration tools are used to create a single rank, this logical disk is then sliced into two(2) LUNs during the install of Ubuntu. Again, the first LUN is used as the root filesystem mounted on "/". The second LUN is used raw as virtual memory or swap space. The Ubuntu device names may be /dev/hda1 and /dev/hda2.

Example 3:[edit]

A large RAID array has eighty(80) physical disks, ten(10) ranks of eight(8) physical disks each. It is decided that each rank will have RAID level 5 applied giving ten(10) logical disks, we'll call them r1 through r10. Each logical disk is then sliced into ten(10) partitions, well call these r1_p1 through r1_p10, r2_p1 through r2_p10, .... and r10_p1 through r10_p10. Finally, it is decided that we want to present only ten(10) LUNs to the host and these LUNs will each be striped across all physical disks for performance. The first LUN will be a striped combination of r1_p1, r2_p1, r3_p1, through r10_p1. The second LUN will be a striped combination of r1_p2, r2_p2, r3_p2, through r10_p2. And so on combining the same partition number of each rank into a striped LUN. The host then accepts the LUNs as devices, say hdisk1 though hdisk10 using AIX notation, formats a filesystem on each and mounts the filesystem as /raid5_p1 through /raid5_p10. This scenario would provide very good I/O performance for something I/O intensive like a large DSS database.

Disambiguation Page[edit]

The stuff about Fortran LUNs is important but doesn't belong on the same page as SCSI LUNs. Perhaps there should be a LogicalUnitNumber disambiguation page. Mr. Shoeless (talk) 16:35, 4 May 2009 (UTC)

Confusing mess[edit]

This article is factually inaccurate and confusing in layout. The article is confusing the SCSI ID, which is related to the bus itself, with logical units, which (as more or less mentioned in the article) are addressed through a unique field in the CDB. All SCSI devices are addressed by ID during the selection phase and the LUN does not get into the picture. Only after the device has responded to selection does the CDB get sent, at which time the logical unit is selected.

Also, RAID keeps getting thrown into the equation, yet no SCSI standard specifically addresses RAID. RAID is an abstraction built upon SCSI, not the other way around.

Bigdumbdinosaur (talk) 04:49, 15 September 2010 (UTC)