AIX

AIX strengths:

* It is based on UNIX System V and has much in common with other versions of UNIX.

* System Management Interface Tools smit and smitty. These are very good system admin tools and cover 95% of your admin work. If you want to learn AIX command line interfaces, just hit F6 and smitty will tell you what commands it will use.

* The IBM compilers for AIX are very good but fussy about your code. Other compilers may let you get away with ambiguous code and make assumptions, but IBM compilers will issue warning messages where your code (for example) relies on side effects or ordering that is not defined in the C standards. Initially, this is a pain but it ultimately makes the code more robust. IBM compilers cost extra. GCC compilers are also available.

* Built in Logical Volume Manager (LVM) - AIX was developed from day one assuming large systems and lots of disks and so has an LVM to manage groups of disks. Some of the terms used may be different than those of other UNIX versions but the LVM is flexible, powerful, and easy to use. LVM is delivered with base AIX.

* AIX has a Journaling Filesystem (JFS2) which supports multi-TB filesystems. The older JFS filesystem is still available, too. Both are delivered with base AIX.

* Workload Manager (WLM) allows multiple applications to be segregated into classes and resources (CPU time, memory, and disk I/O) assigned to the classes, which allows coexistence, monitoring, and management of multiple workloads on a single OS image. (Think Solaris Containers.) WLM is delivered with base AIX.

* HACMP - High Availability Cluster Multiprocessing allows AIX to run a cluster of machines with automated take-over and/or fail-over of applications and data between machines. HACMP is recommended for all critical solutions where minimum downtime is a requirement. HACMP is a separate product which runs on AIX.

* Reliability, Availability and Serviceability (RAS) - marketing speak for a) things don't break often, b) if something breaks, the machine usually either keeps running or comes back up immediately after crashing with the failing component offline, and c) the machine is usually able to identify the failing component and the failing component can often be replaced while the machine continues to run. RAS is built in to AIX and the underlying hardware. For example, POWER4 and POWER5 CPUs can sometimes detect internal errors before they become critical. AIX can then stop scheduling work to the failing CPU and take it offline while applications continue to run.

* Logical Partitions (LPARs) - machines with POWER4 and POWER5 CPUs can be partitioned to run multiple copies of AIX (and/or Linux) at the same time. Workloads formerly run on multiple machines can be run on a single machine, allowing hardware resources to be dynamically (without reboot when running AIX V5.2, V5.3, or Linux) moved as workloads fluctuate. If workloads mesh well (encounter peak loads at different times of the day, week, month, or year), workloads consolidated onto a single machine can be run with fewer resources than required to run them on multiple separate machines, thereby reducing cost. AIX does its part to support the LPAR infrastructure and cooperate with the machine firmware, referred to as the Hypervisor.

* Micro-partitioning - LPARs can share CPUs, which the Hypervisor time-slices. (Requires AIX V5.3 or the Linux 2.6 kernel.)

* Virtualization - LPARs can share disks (and disk & network adapters) to reduce costs. (Requires AIX V5.3 or the Linux 2.6 kernel.)