DAS/NAS/SAN - Do I understand this correctly?

[MD5150]

Hello All,

I'm normally a database programmer, but I now have an opportunity to get more into the networking/data storage side of things, so I've been reading up on articles, products, and configurations, but I'm still not sure if I understand it correctly. Does this sound right?

1) NAS is normally used as file servers, right?
2) DAS is normally use for backups, right?
3) SAN...um, well I'm not too sure about this one. How is a Storage Area Network different from regular NAS? Aren't all components accessible as any server would be? Or is this a totally different concept?

Oh, and some models can be used as both NAS and SAN's, right?

I don't want to trouble anyone with a long essay, but I could definitely use a simple explanation of how each storage type is used.

I'm currently gathering up storage requirements and etc hopefully to eventually propose a good practical setup for our small business, which will soon grow to a critcal 24/7 process of semi-medium size. Money is an issue, so starting low, pay-as-we-go, and flexibility is probably the master plan. But I first need to better understand the model uses and options.

Any advice is greatly appreciated, thanks.

Mike

 

[Chacalinc]

1) NAS is normally used as file servers, right?

Yes, NAS stands for "Network Attached Storage" and it is a dedicated file server (with a custom OS in order to do that)

2) DAS is normally use for backups, right?

Actually, DAS stands for "Direct Attached Storage" and it is said when you connect -either SCSI or FC- directly to your host (with no switches between them).

3) SAN...um, well I'm not too sure about this one. How is a Storage Area Network different from regular NAS? Aren't all components accessible as any server would be? Or is this a totally different concept?

As you know SAN stands for "Storage Area Network" and it is refered as a network -using the FC protocol- between servers and storage (both disks arrays and tape libraries). SAN is a connectivity solution, where FC channel can make a FC network using FC switches. In SAN you speak about "zoning" and "LUN Masking"; zoning in order to setup the connectivity between servers and storage (what server going to see what storage array/port) and LUN Masking in order to present some lun to some server when you are sharing a FC port.

Hope this helps you (and my english could not be quite good, sorry for that).

Cheers.

Chacal, Inc.

 

[MD5150]

Your english is just fine. Thanks for the detail. Appreciate it.

Mike

 

[newbobbys9]

Some additional info: a SAN resides in a vlan unto itself vs a NAS which is attached to the LAN subnet. A SAN provides the ability to use an attached "appliance" for storage (independent of an OS to drive it) and allows for block transport of data over the network. A NAS utilizes file transport (NFS or CIFS) and is slower. iSCSI looks to be the way to go, and there are many products with reasonable prices. Hope this is of help.

 

[bran2235]

Good Stuff-
I am a Network person as well- as (me and HP) my company has just implemented a SAN / NAS combo- and I'm not sure I understand this...

We have an HP EVA 4000 series SAN (controllers) with two racks of disks. There are also two Fiber switches...

The NAS portion of this is 2 HP DL380s which are clustered and have access to storage that is 'carved' out on the disk arrays... (this would be the SAN, right?)

Here's a questions... the controllers- how are they attached to the network? Do they connect directly to the fiber switches? or where???

I think I just hav too much to try and understand at one time... can someone help me? I'm more interested in how all of the following are physically connected to one another:
the SAN;
the NAS;
the controllers;
the DISK Shelves; and
the two fiber switches...


THANK YOU IN ADVANCE!!
Brandon

 

[Chacalinc]

Actually. the NAS portion is just a simple server (2xDL380) may be with a sort of windows storage server or simply a windows/linux with shared folders (samba in case of linux). So, both servers are simply connected to both switches (to get high availability in FC channels) and the switches to the storage controllers, in this case the HP EVA4000. Then controllers have direct attached their own disk drives.

[tt]
[DL380] [DL380]
| \ / |
[FC-SW] [FC-SW]
| \ / |
[Ctrlr] [Ctrlr]
| |
[Disks drives]
[Disks drives]
[Disks drives]
[/tt]

FC-SW: Switches FC
Ctrlr: Storage Controller (EVA)

Note that an FC-SW has connectivity to both controllers and Both Servers have connectivity to both switches FC.

Cheers.

Chacal, Inc.

 

[xmsre]

Each host likely has 2 host bus adapters. One adapter connects to each switch. Likewise, there are (at least) two ports on each storage controller and they connect to each switch. The switch is zoned so that each host has four paths to the disk, HBA A - PORT A, HBA B - PORT A, HBA A - PORT B, HBA B - PORT B. HP's Multipath IO piece, securepath, manages the multiple paths.

The difference between SAN and NAS is the type of access to storage. SAN provides block level access, and NAS provides file level access. For SAN, you would use a block level protocol to connect; SCSI over FCP, or iSCSI (SCSI over TCP/IP). For NAS you would use a file level protocol like CIFS or NFS.

Some vendors offer a NAS head, a dedicated server that connects to storage via a block level protocol and serves up the storage via a file level protocol. Others build this functionality into the controller. The EMC Celerra would be an example of a dedicated NAS head, as well as the slew of Windows Storage Server based products by EMC, Dell, HP, IBM and others. Many use Windows Storage server, others use Unix/Samba or a proprietary OS with a custom SMB stack.