| Previous | Table of Contents | Next |
Other storage devices also will benefit from SSA over time. For example, new disk-array controllers that attach to the SSA links should result in reduced costs and improved performance, and can provide new low-cost RAID solutions for all of our systems.
The second area I want to cover deals with compression technologies. Data compression allows more efficient use of storage media, such as disks. It also can improve the bandwidth between the disk devices and the I/O adapter in the system, thereby potentially giving higher performance. Compression technology has the advantage that it lowers the cost of disk storage on a system without the need to change applications or data-access methods.
The maximum benefit is achieved when the compression and decompression are performed without performance loss. This result requires a technology that can operate at the speeds of today’s I/O buses and that has the capability to pipeline data through the compressor without the significant store-and-forward penalties that can result with some compression algorithms. IBM has developed a fast, highly reliable data-compression chip and algorithm, specifically for storage systems, that meets these performance requirements.
The IBMLZ1 compression algorithm was designed not only for highly efficient compression, but also for extremely high reliability.2 This reliability is important because compression removes redundancy in the source, making the compressed data extremely vulnerable to data corruption. IBM has achieved the high reliability of this compression technology through the combined use of a reliable CMOS technology for the chip and a compression-decompression-coupled checking scheme, where the CRC of the original data is checked against the CRC of the decompressed data. And remember, all this is accomplished on the chip with no performance loss for reading or writing operations to the disk.
2If you would like to know more about data compression in general, or the IBMLZ1 algorithm in particular, I highly recommend the new book by Roy Hoffman, Data Compression in Digital Systems, Chapman & Hall, 1997.
Use of this IBM compression chip on disk adapters provides benefits in disk capacity for AS/400s (the IBMLZ1 compression algorithm typically achieves better than a 3:1 savings in disk capacity for high-end systems) and also potential system performance improvements due to the reduced use of the I/O buses. Future disk adapters will all have this disk-compression technology.
The year 1997 marked the beginning of a new generation of AS/400 systems. The industry shift to network computing is driving many of the enhancements we are making to the AS/400e series. The ultimate beneficiaries of these enhancements are AS/400 customers, no matter what form of computing models they use in their businesses. In this chapter, I have touched on some, but by no means all, of the enhancements you can expect to see in Version 4.
We also are living in the era of the “Web year,” where a Web year is defined as being three months long. Changes are coming so fast that we are now measuring our development cycles in Web years, which helps explain why there will be many releases in Version 4. It should come as no surprise to any of us if, in the next few years, other changes occur that we cannot even predict today. Such is the nature of our business. With this uncertainty in mind, how can we begin to predict what lies beyond Version 4? Sometimes predicting the distant future is easier, because we can take a broader view of what is possible. In the next chapter, I attempt to make some of those future predictions.
| Previous | Table of Contents | Next |