SAN management tweak 2: In case of SAN based backups, it should be noted that the HBA driver version remains consistent across all the SAN backup clients, as we may face recovery problems in case of directed recoveries and also in case of backups. In such cases, a backup client will be unable to read the data backed up on tape by the other client. Similarly, we may face errors that indicate a hardware error during backups and mislead us, as the problem lies with the inconsistency of drivers. I have seen these problems arising at many sites because of the inconsistency of drivers. So this issue should be taken care of while updating the HBA driver on each backup host. This SAN management best practice also applies to hosts accessing a common storage box.
SAN management tweak 3: It's best not to intermix FC switches and HBAs from different vendors in your SAN environment. Such mixing may lead to compatibility and performance issues.
SAN management tweak 4: It is recommended to always perform a single initiator zoning. Use only one initiator per zone.
SAN management tweak 5: Soft zoning (WWPN base zoning) should be preferred over hard zoning, since change in cabling of the fabric will not result in loss of storage access to hosts.
SAN management tweak 6: SAN topology also plays an important role with respect to scalability, resiliency, performance and security of a SAN. So the core edge SAN topology should be considered for building SANs. If there are two switches connected by an inter switch link (ISL) between them, then you should ensure that there is a minimum of two ISLs between them. The initiator and target ports should not be more than six per ISL.
SAN management tweak 7: In an iSCSI SAN, use a dedicated VLAN for storage traffic. Ensure that you use the latest ISCSI initiator version for respective operating systems, CAT 6 cables, and set iSCSI port settings to Ethernet full duplex. It has been observed that Ethernet port settings have a visible effect on the performance—full duplex settings have given best results on this front. When it comes to management of iSCI SANs, it's a good practice to use jumbo frames.
SAN management tweak 8: We have observed that storage administrators mostly deploy RAID 5 protection, which is suitable for most applications. But at times, RAID 5 may not suffice the requirements of applications which require high performance. IOPS requirements are quite high in such cases. So ensure that you use RAID 10 and if feasible FC Drives or EFD drives.
SAN management tweak 9: Most storage boxes contain two backend buses. So the disk array enclosures (DAE), load should be divided equally among the two backend buses. So if we have eight DAEs, then four DAEs should be connected through one bus and the remaining through the other bus.
SAN management tweak 10: Cache also performs a vital role in storage boxes' performance. It has been observed that arrays perform optimally when the read cache is set to 20% and write cache to 80% of the total cache.
SAN management tweak 11: In case of EMC storage arrays, we should not bound LUNs on vault drives that receive IOPS of more than 200, as it impacts the array's overall performance. In case of EFD vault drives, the IOPS threshold is raised to 1500. It's not recommended that vault drives (0.0.0 to 0.0.4) be left unbound. If drives are unbound, they are not being regularly verified by flare. This means there will be no early warning of drive faults, which can cause booting problems for the storage processor. Therefore, if no user data needs to be bound on the first five drives, then a small (for example, 1 GB) test LUN should be bound across any unbound vault drives.
About the author: Anuj Sharma is an EMC Certified and NetApp accredited professional. Sharma has experience in handling implementation projects related to SAN, NAS and BURA. One of his articles was published globally by EMC, and titled the Best of EMC Networker during last year's EMC World held at Orlando, US.
This was first published in March 2010