MAINFRAME > Storage > Disk

The Bedrock of High-End Disk

IBM System Storage and System z partnership paves the way for innovative information infrastructure solutions


 

The IBM System Storage DS8000 Series, IBM’s flagship disk system, has had a unique relationship with the IBM mainframe that dates back almost half of a century. Through the years, IBM researchers have continued to advance the mainframe’s I/O architecture to deliver new ways to accelerate data processing for the world’s most demanding clients in both government and industry. No other data-processing system has contributed so much to modern computing than the IBM mainframe and in fact, much of the technology we see in today’s popular server platforms have their origins in the IBM mainframe. The same can be said for today’s popular storage subsystems, which owe much of their success to the innovation stemming from the I/O architecture of the IBM mainframe, which debuted in 1964 with the IBM System/360.

Looking Back

The System/360 introduced the Count Key Data (CKD) architecture that enabled the host OS software to optimize data transfer between the host and its spinning disks. CKD gave way to Enhanced CKD (ECKD), which enabled host software to communicate “hints” about the host’s I/O operations to the storage control unit. These hints were meant to provide faster performance between the host and its disks.

In the early 1980s, IBM introduced the novel concept of disk virtualization with the IBM 3880 cache disk controller. This cache innovation enabled much faster access to data since the controller could now pre-stage data to its memory–instead of retrieving it from the spinning disks–by analyzing the operating system’s hints. The invention of a cache control unit enabled the disk subsystem to autonomously manage the physical aspects of the rotating disks, which helped optimize end-to-end disk performance.

It wasn’t long after that in 1987 IBM introduced two new storage virtualization concepts that advancedI/O performance even further by decoupling the disk subsystem from the mainframe host. The first was the introduction of non-volatile storage, or NVS, which offered write-cacheing capabilities that enabled the control unit to effectively manage how data was written to the disks. After introducing both read caching and write caching, the team decided to formally declare the control unit had gone under such a transformation that it deserved to be called a disk storage controller, thereby creating the first modern, external disk subsystem.

This, incidentally, sowed the seeds for IBM’s storage business and has led to the tremendous success of the IBM Enterprise Storage System (still affectionately known as “the Shark”) and its successor, the IBM System Storage DS8000 Turbo. The close collaboration that spawned this success still exists today, and clients continue to praise IBM’s longstanding tradition of technology synergy between IBM’s mainframe (System z) organization and its System Storage organization.

Partnership Yields Performance

In addition to the long list of performance synergies, this close collaboration continues to produce synergies that offer the highest levels of information availability, system reliability, scalability, and information security.

On the performance front, the two teams continue to collaborate to reduce I/O service times and make more efficient use of resources throughout the System z server’s I/O architecture, including the DS8000 disk subsystem. These include, but are not limited to, advancements in the System z channel architecture, data-access methods and caching algorithms. The latest advancements include High Performance FICON for System z (zHPF), Intelligent Write Caching and extensive collaboration on solid-state drive (SSD) technology. zHPF is the result of recent enhancements made to the System z architecture and the FICON interface architecture, which improves the performance of online transaction processing (OLTP) workloads running on z/OS. When exploited by the FICON channel, z/OS and the DS8000 storage system, this new zHPF capability can not only help improve performance by up to 200 percent but can also help reduce overhead and improve end-to-end system reliability, availability and serviceability (RAS). For its part, Intelligent Write Caching is a new algorithm from IBM Research that combines and balances existing algorithms to determine how the write cache should be managed for optimal data throughput.

 

Allen Marin is a worldwide marketing manager in IBM's System Storage group and is currently responsible for marketing IBM’s flagship DS8000 series storage platform.


comments powered by Disqus

Advertisement

Advertisement

2018 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

Improving I/O Performance with HyperPAV

The new HyperPAV disk technology delivers a one-two punch with improved I/O efficiency and performance and reduced I/O resources required.

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store
Mainframe News Sign Up Today! Past News Letters