The IBM Power i with its concept of a single address space (Single Level Storage) and its built-in database management in the microcode remains a unique operating system of its kind. Its design any object imposes a particular management of its storage if desired leverage its power and its integration.
Historically, storage is an integraded part of the offer AS / 400 through internal controllers and disks, largely developed by Rochester. Over time, the performance characteristics of these materials have evolved with adding of different generations of servers, as well as CISC RISC.
Since the announcement of version 6.1 of the IBM i in February 2008, PowerVM virtualization resources through the VIOS have greatly enriched the choice of available storage subsystem - facilitating support for SAN arrays and use features that accompany them.
Long reserved only for high-end Power systems connected to the DS6000 and ESS8000 or the external storage, has been evolved with the introduction of the range IBM Storwize recently renewed with the V7000 and V5000 models, not to mention the second generation and supply SVC different models of FlashSystem.
Today, the range of Power7 and 8 servers can benefit from connecting to these external bays with or without VIOS and take advantage of features specific to the equipment such as hardware synchronous replication (Metro Mirror) or asynchronous (Global Mirror), the Flash Copy or compressing (RTC).
To facilitate operations linked to rocker procedures while incorporating the specifics of the IBM i, laboratory Rochester imagined the software "PowerHA SystemMirror for i" and SYSBAS / iASP architecture to provide High Availability hardware solution additional software solutions developed by their high availability partner publishers whose Quick-EDD / HA solution.
To date, few publishers have developed their software to make it "compatible iASP" and some customers have made the effort to transform their house applications to make them compatible with this architecture PowerHA IBM in the context of a true "High availability" of qualified material.
In synthesis, this solution consists of a portion of the storage space that remains attached to the partition carried by a Power server (SYSBAS) and a "detachable" (iASP) that can be connected logically to SYSBAS another partition in the same server or on another server installed or long distance from the first. The "detachable" part (iASP) is supported by a copy of the SAN array of functions to obtain a clone on a more or less identical remote bay and forming a form of logical cluster. This is the switching and activation of this iASP that will allow the remote partition backup to take the place of the ceding score, knowing that the copy functions will be reversed to synchronize data in the other direction.
It is in this detachable portion that the user will have to migrate and organize the application objects (libraries of data and programs, IFS and all necessary items to both applications and the IBM i), knowing that many of essentials must remain in the SYSBAS.
One use of IASPS is to isolate multiple instances of the same product independently of each other using the same object names. This use is made possible through the implementation of "containers" (IASPS) can independently manage these bodies against each other in preserving the uniqueness of names of objects seen by the inside iBM i d a same partition.
A feature included in the product "IBM PowerHA SystemMirror for i®" called "Cluster Admin Domain" is responsible for the timing of certain objects such as user profiles between the two SYSBAS cluster excluding all user objects as well as some important libraries or as QGPL QUSRSYS. These restrictions prohibit the replication of applications that can not be migrated in iASP when we want to implement this type of solution.
If this is the case, the solution is to use a software replication as Quick-EDD / PowerHA® to ensure synchronization of SYSBAS objects in addition to SAN arrays copy functions for IASPS or if you want to dispose of a true high availability solution.
If control of the RTO (Return Time Objective) is not a major factor in production, it returns to a classical architecture without iASP (except coexistence / segmentation of multiple instances of the same product using the same object names as mentioned above) architecture using a global copy via the SAN arrays and restart "cold" when shifting to the backup server.
The vitality of the IBM i and technological developments now allow to consider several architectures for the same application project. It will be extremely important to weigh the pros and cons of each solution with advantages and technical and financial drawbacks to find the best solution for the project.