05-01-2016 09:30 AM
iSCSI SANs enable the use of shared storage while avoiding the cost, complexity, and compatibility issues associated with dedicated separate Fibre Channel switches within a SAN. These iSCSI SANs are becoming increasingly popular in small medium-sized businesses (SMBs) and even Enterprise accounts.
However, the challenge for iSCSI in the large enterprise market has been performance and reliability since it is TCP-based as opposed to the FC protocol, which has been designed to be intolerant to frame loss and uses flow control to enable a lossless environment. The adoption of the Data Center Bridging (DCB) protocol suite enables iSCSI networks to be configured so that the iSCSI traffic is not dropped due to congestion, thus allowing the network to be lossless and avoid TCP drop/re-transmit mechanisms. DCB has aspects that enhance the iSCSI SAN environment to make it more reliable and customizable.
Additionally, unlike native FC, iSCSI can be implemented over 1GbE, 10GbE, 40GbE, and 100GbE TCP/IP networks enhancing the performance benefits of iSCSI SANs even further.
An integrated solution using DCB-iSCSI compliant enterprise 3PAR StoreServ storage systems (StoreServ 20000 and 8000) along with enterprise-class switches, such as the HPE FlexFabric 59xx switch product portfolio, is perfectly positioned to form the backbone of a modern high-performance, highly available iSCSI storage network. These Hewlett Packard Enterprise solutions are built on a lossless Ethernet network which has been validated for minimizing packet discards and for preventing packet losses and unpredictable performance due to congestion and initiator TCP window recovery procedures.
HPE has some great docs which can help guide you through use cases, general recommendations, and best practices around building an iSCSI SAN.
Visit the Hewlett Packard Enterprise Information Library to find some of the docs. Below are a couple of examples specific to iSCSI SAN design and best practices: