The eServer i5 has a complex architecture. To effectively operate, administer, and program this machine, you must understand the main concepts about this architecture. This chapter explains them.
The i5 is a single-level storage machine. To the operating system, both memory and DASD (Direct Access Storage Device, or "disk") are treated the same. When a program runs, and it must retrieve a piece of information, all the program needs to know is a single address. The operating system uses this address to retrieve the information whether it is in memory or DASD.
In addition, the i5 doesn't need to store an object in a single, continuous space on DASD. If you have a large file, for example, the file may be split into many small pieces and spread across several physical DASD units.
The advantage is immediate: Neither the operating system nor the person who uses the system needs to worry about reserving enough space for objects, because the system automatically uses whatever space is available, even if the pieces are many and far apart.
Along with this advantage, there is a price to pay. If one of the DASD units breaks down, you have a long and complicated recovery procedure ahead of you, because many thousands of objects could be affected. All these objects could have small splinters of their information in the DASD unit that failed and have the rest of their information in other units. Consequently, none of the DASD units (except maybe one or two, if you are incredibly lucky) would have complete and usable objects. Thus, IBM has developed technologies like check-sum and RAID (redundant arrays of inexpensive disks).