Top Solution to Real-World Deployment Challenges in iSCSI
This best practice report was made to share my generation involvement in architecting, executing, and overseeing numerous ONTAP 7-Mode HA sets and more than 30 ONTAP SAN groups. This information depends on certifiable encounters of iSCSI arrangements that I’ve experienced throughout the years. This record was utilized as reference material amid the iSCSI Best Practices whiteboard session at NetApp Insight 2018 in Las Vegas. On the off chance that you are searching for some tried prescribed procedures to enable you to be fruitful with your iSCSI arrangements, at that point continue perusing.
Before getting into it, I needed to set up a couple of key definitions just as a snappy outline of NetApp Unified Storage.
NAS – Network Attached Storage: File-based capacity (NFS and SMB (CIFS), SMB3) in which ONTAP controls the record framework.
SAN – Storage Area Network: Block-based capacity (FC, FCoE, iSCSI) in which the host controls the document framework.
LUN – Logical Unit Number: A legitimate portrayal of a joined SCSI circle.
SCSI – Small Computer Systems Interface: A lot of measures that characterize directions, conventions, and interfaces that are utilized to transmit information. SCSI permits low-level access to information in units of 512-byte squares. This is very productive and has low overhead contrasted with NAS (record level) get to. SCSI has an abnormal state of versatility that makes it ideal for an endeavor level convention. SAN utilizes SCSI-3 convention.
FC SAN – Fiber Channel SAN: Uses FC convention to convey over FC ports. FC embodies SCSI directions in FC outlines.
IP SAN – iSCSI SAN: Uses iSCSI convention to impart over Ethernet ports. iSCSI exemplifies SCSI directions in IP bundles. Utilizations TCP port 3260.
In a FC SAN, an overall hub name (WWNN) depicts a machine, and an overall port name (WWPN) portrays a physical port that is appended to that machine. In an IP SAN, the hub name portrays a machine, and the entry depicts a physical port. Each iSCSI hub must have a hub name. There are two bolstered hub name positions: IQN (iSCSI Qualified Name) and EUI (Extended Unique Identifier).
Set MTU to 9,000 on information and group interconnect ports and debilitate flowcontrol on all ports (net port show – fields mtu,flowcontrol-administrator).
FC and IP SAN ports inside the capacity controller use application-explicit incorporated circuits (ASIC) chipsets. Verify that objective ports are part crosswise over various ASIC chips inside the controller or extra card to dispense with a solitary ASIC as a solitary purpose of disappointment.
(IP) e0a and e0b are on the equivalent ASIC
(FC) 0a and 0b are on the equivalent ASIC
The right utilization would be:
(IP) e0a and e0c, e0b and e0d
(FC) 0a and 0c, 0b and 0d
ALUA (Asymmetric Logical Unit Access) Path Optimization Selection
ALUA is required in both FC and iSCSI executions.
Dynamic/Optimized and Active/Non-Optimized ways are both arranged for access with at least 1 way for each hub. For better execution, it is exceptionally prescribed to have different ways per hub to permit at least 2 ways for multipathing.
An Active/Optimized way utilizes immediate or essential ways between the initiator and focus on the hub that possesses the LUN. An Active/Non-Optimized way utilizes a roundabout or optional way between the initiator and focus through the bunch interconnect, with expanded idleness.
Path Selection (Round Robin)
The host utilizes a programmed way choice calculation pivoting through every single dynamic way when associating with dynamic aloof clusters, or through every single accessible way when interfacing with dynamic exhibits. Round Robin (RR) is the default for various clusters and can be utilized with both dynamic and dynamic detached exhibits to actualize stack adjusting crosswise over ways for various LUNs. NetApp is arranged into an Active/Active arrangement as a matter of course. It very well may be arranged in Active/Passive relying upon client necessities.
Kind sized edges are bigger Ethernet parcels that diminish the proportion of bundle overhead to payload. The default Ethernet outline size or MTU is 1,500 bytes. With gigantic edges, MTU is ordinarily set to 9,000 on end hubs, for example, servers and capacity, and to a bigger esteem, for example, 9,198 or 9,216, on the physical switches. Kind sized casings must be empowered on every single physical gadget and coherent elements from start to finish so as to evade truncation or fracture of parcels with the greatest size.
On physical switches, the MTU must be set to the most extreme upheld esteem, either as a worldwide setting or strategy alternative or on a port-by-port premise (counting all ports utilized by ESXi and the hubs of the NetApp bunch), contingent upon the switch execution. The MTU should likewise be set, and a similar esteem must be utilized, on the ESXi vSwitch and VMkernel port and on the physical ports or interface gatherings of every hub.
At the point when issues happen, it is regularly on the grounds that either the VMkernel or the vSwitch was not set for gigantic edges. For VM visitors that require guide access to capacity through their own NFS or CIFS stack or iSCSI initiator, there is no MTU setting for the VM port gathering; be that as it may, the MTU must be designed in the visitor. The picture beneath shows gigantic casing MTU settings for the different systems administration parts.
In the wake of making ifgrps and VLANs, evacuate the ifgrps and information ports from the default communicate area and make an information communicate space. Move the ifgrps and information ports into the information communicate area and set the mtu setting to 9,000.