Home > Storage > PowerMax and VMAX > Storage Admin > iSCSI Implementation Guide for Dell EMC Storage Arrays Running PowerMaxOS > How iSCSI works
As said earlier, the iSCSI protocol allows for the encapsulation of SCSI commands on a standard TCP/IP connection and transported over an Ethernet based network between a host and a storage array. These actions can be separated into two processes: the Login Process and the Data Transfer Process.
When an iSCSI host initiator wishes to communicate with an iSCSI storage array, it begins with a login request. The login request contains information about “who” is sending the request and “what” storage target the host wishes to communicate with. If CHAP is being used, the request will contain CHAP information. The iSCSI storage array will authenticate the host initiator using the CHAP information. If the authentication is successful, the login is able to complete, and a “session” is established between the host initiator and the storage array target. Once the session is established, the transfer of SCSI commands and data between the host initiator and storage array target can begin. It is not uncommon for iSCSI sessions to remain active for days or months. When either the host or the storage array decides to close the session, it will either issue a logout command. When the session closes, the ability transfer of SCSI commands and data between the host and storage will also end.
iSCSI transports block level data between an initiator on a host and a target on a storage array in the following manner:
This process is shown in the following figure:
When the target side receives iSCSI Ethernet frames, the target datalink layer will remove the frame encapsulation and pass the results up to the Network Protocol Layer. The Network layer will remove the IP datagram encapsulation, and the Transport layer will remove the TCP segment encapsulation, leaving a PDU to be passed up to the session layer (iSCSI protocol layer). The iSCSI Protocol Layer will remove the SCSI data from the PDU and pass it to the presentation layer for interpretation and processing.
The figure below shows the different components in an iSCSI Ethernet frame.
Essentially there is no difference between an iSCSI Ethernet frame with a standard Ethernet frame except what is the payload in the TCP segment - the iSCSI PDU. There is nothing in the TCP Segment Header to indicate that the TCP Data Segment contains data of a specific protocol. The TCP/IP definition does not prevent iSCSI PDUs and other network data from being transmitted on the same network. Similarly, there is nothing that requires that they be mixed, so a network administrator can determine whether an isolated subnet for iSCSI is necessary or not. The ability to carry multiple types of data in TCP Segment header is what allows modern Ethernet switches to the transport of iSCSI, IP, and Fibre Channel over Ethernet (FCoE) on the same infrastructure.