Originally posted by: CUK

Recent news about the new IBM System z13 were the initiator to look back over the last years and summarize IBM synergies between IBM Storage and System z mainframe computers.

IBM System Storage synergy with System z has long tradition at IBM, resulting in many benefits like improved application efficiency, performance and highest possible system availability, due to resiliency functions like remote copy, which are managed by System z software and services packages for effective DR setups.

Already in the 1980s IBM provided S/390 and MVS with the capability to manage the cache of the disk subsystems of those days, the 3880 / 3380. MVS was capable to define how IO operations should be treated by the 3880 controller resulting in a better utilization of cache resources and improved application performance.

In the 1990s IBM expanded the cache management to the 3990 / 3390 / RAMAC Disk Subsystems and added write capabilities to caching. That was a significant improvement for write intensive applications and the base for implementing remote copy functions like PPRC ( Peer to Peer Remote Copy ), since it allowed a fast data transfer between the controllers, resulting in a low impact to application performance.

In the early 2000s, with the introduction of the Shark or ESS ( Enterprise Storage Server ) IBM went to an even tighter integration of Storage and System z, by providing functions like PAV ( Parallel Access Volume ) and MAV ( Multiple Allegiance Volume ). These functions allowed z/OS to run several IOs to disk devices in parallel, either from a single LPAR with PAV or from different LPARs with MAV. With PAV applications benefit from a dramatic reduction in response time, allowing OLTP ( Online Transaction Processing Systems ) to run much more transactions, i.e. leading to an improvement of e-business applications.

MAV on the other site allowed clients to run different types of processes simultaneously, like OLTP and batch to a device, which is a typical scenario in z/OS. MAV gave companies the possibility to keep their business online to their clients, while running their internal processes in parallel , something that was not possible before.

On top, IBM added the WLM ( Workload Manager ) the possibility to define quality of service to IOs and efficient management of PAV. That led to a better performance and administrator productivity, since it reduced the efforts of performance tuning by user defined policies.

Since the 1990s IBM storage and z/OS also went hand in hand concerning business continuity solutions. The disk systems got the capability of running remote copies of the data they were storing, either synchronously with PPRC, today called Metro Mirror, or asynchronously with XRC ( Extended Remote Copy, today called z/OS Global Copy ) or later Global Copy.

z/OS capabilities to manage these disk functions allowed developers to create solutions like GDPS ( Geographically Dispersed Parallel Sysplex ) to fully automate DR plans for clients, resulting in a better and much higher system and business availability.

Other functions like Concurrent Copy or today's FlashCopy, allow users to create immediate copies of their data while business and applications do not need to be stopped for backup purposes, resulting in more efficient backup processes, by reducing backup windows from several hours to just minutes or less.

Over the last decade IBM continued to improve the different functions. With todays IBM System Storage DS8870 and System z, clients are offered a great variety of synergy solutions to fulfill todays business goals. Some of them are:

• zHPF Enhancements including all z/OS DB2 I/O, BxAM/QSAM

• Extended Distance FICON

• Caching Algorithms – AMP, ARC, WOW, 4K Cache Blocking

• DFSMS Recognition of SSDs and Storage Tiering

• Easy Tier - Easy Tier Application and Heat Map Transfer

• DB2 HyperWrite

• z/OS GM Multiple Reader Support

• SSDs + DFSMS + zHPF + HyperPAV + DB2

• zWLM + DS8000 I/O Priority Manager

• HyperPAV

• HyperSwap Technology

• TPC for Replication, GDPS

• Multi Site DR and Target Options

• GDPS and GDOC Automation

• GDPS Active / Standby/Query/Active

• Space Efficient FlashCopy

• Multiple Target Inncremental FlashCopy

• zCDP for DB2, zCDP for IMS – Eliminating Backup windows

• Heat map transfer

• 1 TB EAVs

• Quick Init for CKD Volumes

• Dynamic Volume Expansion

• z/OS Distributed Data Backup

• System z Discovery and Automatic Configuration (zDAC)

• Alt Subchannel Exploitation

• Disk Encryption

• and more ...

It would take a long time to describe every item in the list above, but due to the recent market strategies like Big Data Analytics or Security, some of them are worth to cover in some more detail.

z/OS DB2 HyperWrite improves DB2 Log Write Performance with DS8870 Metro Mirror because DB2 performs parallel log writes, reducing DB2 log writing time up to 40%.

zHPF together with the DS8000 provides a lower elapsed time for batch IOs, resulting in service time improvements of up to 32%, but also other typical DB2 actions, like formats, scans and lately also DB2 List Prefetch will get improvements ranging up to 111%.

The IBM System Storage DS8870 offers the option to use integrated Flash Memory either in a hybrid way by mixing it with HDDo or Flash only. Using it in the hybrid way or with SSDs will enable Easy Tier in its latest version allowing z/OS applications to inform Easy Tier about the appropriate tier for best performance results.

These type of z/OS and DS8870 synergy solutions benefit institutions running Big Data Analytic applications i.e. to improve their findings, results and business. In one case a globally acting bank in the USA was able to improve their fraud detection scan nearly by 10 times, resulting in several millions of US$ cost savings already in the year it was introduced.

An insurance company also experienced mayor benefits in implementing DS8870 with Flash memory and z/OS integration solutions. Their applications got 90% reduced RT ( Response Time ) compared to a well tuned pure HDD system. More important was that they got a predictable and flat RT no matter what applications were running, resulting in a 50% reduction of individual query completion time and 30% less host processing rates. That on the other hand meant freed server resources for other business applications.

Another example of z/OS and DS8000 synergy is Extended Address Volumes ( EAV ).

Over the years the 4 digit addressing and the capacity of the emulated 3390 devices was not enough to address the needs of todays business, but also the bigger getting amount of data for DR purposes challenged the z/OS System z addressing capabilities. The solution for that has been EAV, a z/OS and DS8000 feature allowing the addressing of large 3390s up to 1 TB capacity. Thanks to PAV, HyperPAV and WLM an application will not suffer in performance because of that large size of the volume and a user can manage more capacity with less addresses, simplifying the device management and DR implementation.

The DR solutions that exist for z/OS and DS8000 have been improved over the years massively. As said, all started with PPRC and XRC and the possibility to manage these DR functions from MVS, OS/390 or today z/OS. That allowed developers to create DR automation software and services like GDPS. Initially with the capability to manage 2-site DR scenarios in a consistent and automated way. Today also 3-site DR and even 4-site DR setups are possible. RPO and RTO values are reduced from hours to minutes by intelligent replication technologies and automation solutions, such as GDPS or Tivoli Storage Productivity Center for Replication ( TPC-R ). HyperSwap capabilities allow a transparent switch between replicated copies of data within seconds, further increasing application availability significantly.

The latest DS8000 DR features allow a multi-target Metro Mirror, meaning that one primary device can have several secondaries defined. This allows more flexible Multi-Site DR setups and a higher business availability.

Other then that new functions have been incorporated recently like

1. Ficon Dynamic Routing, allowing higher performance at lower cost
2. 16Gb FICON adapters that improve transfer rates by 2x and benefit DB2 log writes with zHyperWrite
3. Forward Error Correction to ensure an even saver data transfer by adding more redundancy
4. zHPF Extended Disance II that improves data transfers for Metro Mirror by 50%
5. and Fabric Priority which adds resiliency capability and enhances FICON Dynamic Routing

As said, one could continue to describe a lot of more examples of z/OS and DS8000 synergies. The just made announcement of the IBM System z13 is again an example for that, but more important is that these type of capabilities started over 30 years ago and IBM will continue to provide more functions with the intention to improve clients businesses in terms of cost reduction, application availability and performance.