One reason mainframe and open systems staff without cross system experience don't communicate clearly is: language. Although both groups think they are speaking clearly, neither group actually understands the other. For reasons having to do with job security and industry FUD, vendors have used different terms to mean basically the same thing.
Mainframes and midrange systems can be divided into smaller logical system images. LPARs are logical partitions of the mainframe hardware. Each LPAR runs a separate copy of the operating system. Large RS/6000 systems can be physically subdivided into nodes. Each node runs a separate copy of the operating system.
While this provides a similar result, the logical versus physical distinction is very important, especially in terms of I/O. An LPAR can be defined to access any I/O device connected to the mainframe hardware. An RS/6000 node can only access I/O devices physically connected to that node.
Disk I/O connectivity for mainframes and RS/6000's is accomplished by connecting an internal adapter of some kind to an external physical device. (In making this comparison, I am ignoring the capability of both to have some internal devices.) Mainframes use an adapter called a channel. RS/6000 systems use a host bus adapter (HBA). Channels come in copper bus/tag, ESCON and FICON. HBAs can be copper SCSI or Fibre Channel (FC).
Mainframe FICON channels and RS/6000 fiber channel HBAs are both variations of the FC standard. At the physical level and the lower protocol levels, both are identical. It is easy to think of a channel and an HBA being the same.
Where channels and HBAs differ is the amount of connectivity that a mainframe LPAR or an RS/6000 node can provide. A mainframe LPAR can have 100 or more FICON channels associated with it and share those channels with other LPARs. An RS/6000 node can have no more than two fiber channel HBAs installed in it and can not share these HBAs with other nodes.
Using an IBM Shark disk subsystem as an example, the mainframe LPAR can have sixteen FICON channels from one LPAR. A single RS/6000 node can have two FC HBA connections. Mainframe operating system software will automatically support I/O load balancing across the sixteen paths and handle fail over. The RS/6000 requires additional software drivers be installed to handle load balancing and fail over. These differences must be understood by the SAN designer.
Another area of confusion is switches. The only switches in a mainframe disk I/O environment are McData or Inrange enterprise class directors with all the high end availability features. In the RS/6000 world, switches can be either low/mid range switches such as the Brocade models or enterprise class directors. Since the word switch is used to describe both, it is important to define the actual type and model of switch.
About John Weinhoeft:
For the past 30-plus years John Weinhoeft has had his hand in the computer industry. He recently retired from designing and managing the State of Illinois' centralized computer systems that served 100 agencies. John has authored and edited a number of analytical books published by Computer Technology Research Corporation. He is, or has been, a member of several computer organizations including the Computer Measurement Group and Central Illinois Personal Computer Users Group.
Send John your comments