Library Table of Contents Index Previous Page (Alt-p) Next Page (Alt-n) Comments

Understanding how NetApp implements an FC SAN network

What FCP is  Top

FCP is a licensed service on the filer that enables you to export LUNs and transfer block data to hosts using the SCSI protocol over a Fibre Channel fabric. For information about enabling the fcp license, see Managing the FCP service.

What nodes are  Top

In an FCP network, nodes include targets, initiators, and switches. Targets are filers and initiators are hosts. Filers have storage devices, which are referred to as LUNs. Nodes register with the Fabric Name Server directory when they are connected to a Fibre Channel (FC) switch.

What LUNs are  Top

From the filer, a LUN is a logical representation of a physical unit of storage. It is a collection of, or a part of, physical or virtual disks configured as a single disk. When you create a LUN, it is automatically striped across many physical disks.

Data ONTAP manages LUNs at the block level, so it cannot interpret the file system or the data in a LUN.

From the host, LUNs appear as local disks on the host that you can format and manage to store data, using either the FCP or iSCSI protocols.

What a LUN serial number is  Top

A LUN serial number is a unique 12-byte, filer-generated ASCII string. Many multipathing software packages use this serial number to identify redundant paths to the same LUN. For example, VERITAS® DMP expects the serial number of a drive to reside in bytes 36 through 47. You display the LUN serial number with the lun show -v command.

How nodes are connected  Top

Filers and hosts have Host Bus Adapters (HBAs) so they can be connected directly to each other or to FC switches with optical cable. In addition, they can be connected to each other or to TCP/IP switches with Ethernet cable for filer and FC switch administration.

When a node is connected to the FC SAN network, it registers each of its ports with the switch's Fabric Name Server directory service using a unique identifier.

How nodes are uniquely identified  Top

Every FCP node has a unique identifier. Every manufacturer of FCP nodes assigns their equipment with a World Wide Node Name (WWNN) or a World Wide Port Name (WWPN).

Host

The host WWPN is derived by prepending 10:00 to its IEEE address and has a pattern of 10:00:00:00:nn:nn:nn:nn.

To know which WWPN is associated with a specific host, see the SAN HBA attach kit documentation for your host. These documents describe NetApp-supplied or initiator-vendor supplied commands or methods that show the mapping between the host and its WWPN. For example, for Windows hosts, you use the lputilnt command, and for Solaris hosts, you use the sanlun command.

You can use the fcp show initiator command or FilerView (click LUNs > Initiator Groups > Manage) to see all of the WWPNs of the FCP initiators that have logged on to the filer. Data ONTAP displays the WWPN as Portname.

Filer

When the FCP service is first initialized, it assigns a WWNN to a filer based on the serial number of its NVRAM adapter. The WWNN is stored on disk. Each target port on the HBAs installed in the filer has a unique WWPN. Both the WWNN and the WWPN are a 64-bit address represented in the following format: nn:nn:nn:nn:nn:nn:nn:nn, where n represents a hexadecimal value.

You can use commands, such as fcp show adpater, fcp config, or sysconfig -v or FilerView (click LUNs > FCP > Report) to see the filer's WWNN as FC nodename or nodename or the filer's WWPN as FC portname or portname.

Switch

Fibre Channel switches have one WWNN for the device itself, and one WWPN for each of its ports. For example, the following diagram shows how the WWPNs are assigned to each of the ports on a 16-port Brocade switch. For details about how the ports are numbered for a particular switch, see the vendor-supplied documentation for that switch.


 
Port  0, WWPN 20:00:00:60:69:51:06:b4
Port  1, WWPN 20:01:00:60:69:51:06:b4
...
Port 14, WWPN 20:0e:00:60:69:51:06:b4
Port 15, WWPN 20:0f:00:60:69:51:06:b4

How WWNNs and WWPNs are used  Top

WWNNs and WWPNs are used for the following purposes.

Creating an initiator group

The host's WWPN is used to create an initiator group (igroup). An igroup is a collection of WWPNs of initiators in an FCP network that you specify when you create the igroup. When you map a LUN on a filer to the igroup, you grant all the initiators in that group access to that LUN. If a host's WWPN is not in an igroup that is mapped to a LUN, that host does not have access to the LUN. This effectively masks access to the LUN.

Uniquely identifying a filer's HBA target ports

The combination of the filer's WWNN and WWPN is used to uniquely identify each HBA target port on a filer. Some operating systems require persistent binding to ensure that the LUN appears at the same target ID on the host.

How Data ONTAP supports FCP with non-clustered filers  Top

If your filer is not in a cluster, Port A handles access to LUNs. Port B can provide a second path to the LUNs for configurations with one or more hosts using multipathing software.

How Data ONTAP supports FCP with clustered filers  Top

If your filers are in cluster, Data ONTAP provides multiple modes of operation required to support homogeneous and heterogeneous host operating systems.

FCP is implemented over the target's and initiator's HBA ports. Target HBAs can have one or two ports, and are labeled Port 1 and Port 2 (if there is a second port).

Data ONTAP uses the FCP cfmode setting to control how these target ports:

These modes can be viewed with the fcp show cfmode command.

Caution
Changing the FCP cfmode setting on your filer might prevent hosts from being able to access data on mapped LUNs. Please contact Network Appliance Customer Support to modify the FCP cfmode setting.

FCP cfmode settings for clustered configurations

Clustered filers in an FC network require that the FCP cfmode settings are the same. You can use the NetApp Cluster Configuration Checker to detect inconsistent FCP cfmode settings on clusters licensed for FCP.

To download and run the Cluster Configuration Checker, log in to the NOW Web site at http://now.netapp.com and go to Software Downloads > Tools & Utilities.

Enabled options for cluster configurations

Clustered filers in an FC network require that the following options are enabled to guarantee that takeover and giveback occur quickly enough to not interfere with host requests to the LUNs. These options are automatically enabled when the FCP service is turned on.

How FCP cfmode settings affect target ports  Top

The following settings for FCP cfmode determine how the FCP target ports provide access to LUNs:

How to display FCP target ports  Top

Data ONTAP displays information about the ports by using the slot number where the HBA is installed in the filer. The display also depends on the FCP cfmode setting. For example, if the FCP cfmode setting is standby or partner and the target HBA is installed in slot 5 of a filer, the FCP commands display Port 1 as 5a and Port 2 as 5b. In addition, the local port name and node name addresses of the WWNN and WWPN have a pattern of 50:a9:80:nn:nn:nn:nn. The standby port name and node name addresses of the WWNN and WWPN have a pattern of 20:01:00:nn:nn:nn:nn:nn.

If the FCP cfmode setting is mixed, and the target HBA is installed in slot 6, FCP commands display the HBA as 6a, the virtual local port as 6a_0, the virtual standby port as 6a_1, and the virtual partner port as 6a_2.

For FAS270c appliances only

For FAS270c appliances, the FCP cfmode setting is dual_fabric, and each port is configured as a virtual port. The ports are displayed as 0c_0, 0c_1, and 0c_2.

How to manage target port resources  Top

Each target port has a fixed number of resources, or command blocks, for incoming initiator requests. When all command blocks are used, an initiator receives a QFull message on subsequent requests. Data ONTAP enables you to monitor these requests and manage the number of command blocks available for specified initiators. You can limit the number of command blocks used by the initiator in an igroup. This is known as igroup throttling. For information about igroup throttling, see Managing initiator requests.

Library Table of Contents Index Previous Page (Alt-p) Next Page (Alt-n) Comments