The Consolidation Revolution

August 18, 2016

In the history of the IBM i hardware and software, we’ve gone through trends of decentralization as well as centralization, or consolidation. This has resulted in different things as variations of IBM i infrastructure have been brought together into a single location. At first, single data centers had multiple AS/400 systems, with user environments controlled strictly through subsystem technology. Later, the trend changed to multiple partitions on a single iSeries system running OS/400, which was connected to physical I/O adapters and possibly virtual TCP/IP interfaces.

Today the latest Power Systems have multiple IBM i, AIX, or Linux partitions—potentially fully virtualized micro-partitions—hosted on a single Power systems server. These developments in IBM i and Power7 have put us in the middle of a consolidation revolution. Here are three technologies to help you join it.

1. Subsystems

Thanks to subsystem technology, the IBM i operating system and OS/400 (and i5/OS) could be designed to allow multiple environments to run on it concurrently. Subsystem definitions allow the administrator to segregate work (interactive, batch, server) into their own memory pools and their own separate and distinct job queues, all sharing a portion of processor time. This was the first form of consolidation technology and is still strong today. In fact, with the horsepower available on a single system today, it is quite possible to consolidate multiple systems and thousands of users onto a single Power Systems server with a single partition.

Have multiple companies or divisions? Break them apart by subsystem, with each user environment assigned to a specific job description that will direct them into the proper subsystem and the appropriate libraries and directories. For motivation, consider one of our customers who, thanks to subsystem technology, consolidated more than 800 dedicated store system AS/400s to six Power7 partitions running IBM i. Imagine how much easier the administrator’s job got! And the energy savings!

2. Independent Auxiliary Storage Pools

Independent auxiliary storage pool (IASP) technology is another form of consolidation. IASPs let administrators isolate data onto groups of disks that can switch between partitions. Multiple IASPs can be attached to a single system. For example, this could allow multiple divisions or customer workloads to be consolidated on a single partition. You could consolidate different business operations, multiple development or simulation environments, or even multiple versions of an application onto one partition—using the same library names housed in separate IASPs while keeping the data totally isolated.

To take advantage of IASP technology, software vendors sometimes have to make program changes because library and directory objects are referred to with the same name but with an IASP prefix. Changing the IASP group for the job with the SETASPGRP command, by using the Initial ASP group parameter on the SBMJOB command is another option, as is specifying the Initial ASP group on the appropriate job description (*JOBD). IASP technology is also the basis for PowerHA, FlashCopy, geographic mirroring, and more. The technology will also sound familiar when we discuss moving virtual partitions between systems in the following section.

3. Virtualization

Consolidation via virtualization is not new to OS/400, i5/OS, or IBM i. The IBM i single level storage technology is a form of virtualization and was first implemented on the System/38, allowing for virtual memory where the OS regards all of primary (RAM) and secondary storage (DASD) as one large pool. Single level storage allows processes and jobs to run without the programmer having to estimate how many programs or how much data to load into memory at one time. This greatly simplifies handling a large company’s “big data” needs and simplifies programming and database management.

Another technology is virtual media, which allows backups without physical tapes and, thanks to CD image catalogs, product installs to run without physical CDs. Virtual media allows program installs and updates to be shared across multiple partitions with ease and without having to physically load media. Virtual tape can be combined with IASP technology for backup to a storage area network (SAN). This IASP could then be attached to another partition for recovery or archiving.

Virtual partition technology is available with PowerVM which, in combination with IASP technology, adds tremendous capabilities to our IBM i world. But virtualized partitions weren’t invented on Power Systems—it has been around in the mainframe environment for years. Power just improved on the technology. And it is truly exciting technology, as it allows as many IBM i, AIX, or Linux partitions to be created and enabled as are needed. All you need is disk, memory, and CPU (and appropriate licensed programs from IBM).

Why might you create a virtual partition? Simple: To better utilize resources. For instance, a fully virtualized partition could be created to test a new cumulative PTF package, or new version of the operating system, then deleted when the testing is complete. A test partition could be created to validate a new release of application software. Virtual partitions can be copied. Virtual partitions can be suspended when unneeded, thereby freeing up resources for other partitions. Suspended partitions are saved to disk much like a file or a directory. This is very similar to virtual memory technology, where pages of memory are sent to disk when not needed and brought back as required.

Virtualized partitions cut power consumption and allow for CPU and I/O resource sharing. Another feature is active memory sharing, which allows the dynamic exchange of memory between active virtual partitions. And then there is the crown jewel of virtualization: Live Partition Mobility, which allows you to move a virtual partition—while it is active—between frames, with no system outage to the end user.

Live partition mobility (LPM) is similar to PowerHA in that it copies the entire partition’s data at the disk level out to a SAN on your network, which in turn is attached to another Power Systems server frame. The advantage is that you can activate this copied partition on the target frame in seconds; and any remaining updates happen after the partition comes up on the target. This is not a substitute for a high availability fail-over plan using PowerHA or another mirroring tool. LPM is used for planned system outages; workload balancing by moving a partition from busy to light systems; or for consolidation of multiple virtual partitions onto one hardware frame.

IBM i, UNIX, and Linux consolidation

One IBM Power Systems frame can host IBM i, Linux Red Hat or SUSE, or AIX UNIX partitions in a single “footprint.” This is a great opportunity for server consolidation. However, this does cross “political” boundaries that may exist in your IT structure, since those systems are all supported in one place, from one PowerVM or HMC console, and are sharing CPU, memory, power, and rack space. This might be a great opportunity to consolidate those duties or, at a minimum, cross-train those administrators. Bring factions together to discuss the advantages of this consolidation: reduced power consumption, having a single vendor, and a simplified, more flexible environment.

Ready to join the revolution? Read our guide to learn how automation helps with consolidation.