Vous êtes sur la page 1sur 11

Solaris to AIX

Practical migration information


Christian Pruett
Senior Systems Administrator
Freelance

24 August 2011

Are you moving away from Oracle or Sun Microsystems equipment to IBM System p
servers? Are you more familiar with zones and LDOMs than HMCs and VIO servers? This
article provides practical tips for migrating from Solaris to IBM AIX servers.
Most flavors of UNIX and Linux on the market today are similar enough that even a novice
systems administrator can navigate the various operating systems with ease. But when you need
to switch from one operating system to anotherbe it because of a change in business direction,
technology enhancements, or cost savingsmaking the switch can be tough.

Frequently used acronyms

FTP: File Transfer Protocol


JFS: Journaling file system
LPAR: Logical partition
MTU: Maximum transmission unit
NFS: Network file system
RHEL: Red Hat Enterprise Linux
RSH: Remote shell
SAN: Storage area network
SSH: Secure shell
WPAR: Workload partition

This article looks at practical migration information for switching from the Oracle (formerly Sun
Microsystems) Solaris operating system and associated hardware to the IBM System p platform
with the IBM AIX operating system. It covers key information that administrators should know
about what differentiates the two systems and shows how you can start running quickly.

Systems management
The most basic thing a Solaris administrator should know about managing System p hardware
is how to access the servers and get things running. In the Solaris world, you do this through the
Advanced Lights Out Manager, Remote System Control, or other forms of console access; for
most AIX systems, you use the Hardware Management Console (HMC).
Copyright IBM Corporation 2011
Solaris to AIX

Trademarks
Page 1 of 11

developerWorks

ibm.com/developerWorks/

The HMC is a multifunction stand-alone system that can interact with multiple System p computers
simultaneously. It communicates with the servers' firmware through a special area called the
hypervisor, which manages the servers' resources. Using the HMC, you can allot CPU, memory,
disk, and other hardware resources to individual LPARs, which serve as virtual AIX servers within
the host System p hardware. In addition, resources can be shared or micro-partitioned among
multiple LPARs.
The HMC is accessible either directly in the console or by configuring remote connectivity, which
allows login functionality by directing a web browser to a secured page available on the HMC.
Begin setting up your system by creating a spreadsheet with the location of your Fibre Channel
and network cards on a per-drawer basis. Cables can get plugged into the wrong places easily and
logging in at the HMC won't show the drawer or rack elevations, so the spreadsheet is invaluable.
To set up your spreadsheet, complete these steps:
1. Log in to the HMC as hscroot.
The default password, if you've never logged in before, is abc1234.
2. In the left column, expand the Systems Management view and then the Servers menu. Click
the managed server you need to examine.
3. Click the Properties menu.
4. On the I/O tab, record a list of all of the cards by drawer and physically compare them to the
server, noting at what elevations the particular drawers are located.

Low-level management
One of the first tasks a Solaris systems administrator learns is to access the open boot PROM
also known as the OK prompt. This low-level area of the hardware allows you to reset a hung
system, boot from a specific device (like a CD or a network boot), or get into single-user mode.
Although there is no direct equivalent on System p hardware, there is a similar low-level interface
known as System Management Services (SMS).
SMS is a numerically driven menu system that allows you to set the boot list order (that is, CD,
hard disk, network), set the boot IP address in the Remote Initial Program Load subsection for
a Network Installation Manager build-out (similar to Solaris Jumpstart), or perform lower-level
diagnostic work.
You access this menu by pressing the F1 or 1 key when you power on a System p computer or
LPAR. Alternatively, an LPAR can boot directly into the SMS menu by changing the Boot Option
sub-menu through the HMC when activating the LPAR.
If you are in the SMS and going to boot from a disk but have multiple hard drives mapped from
a SAN across a number of adapters, the SMS can take significant time to probe all the hardware
addresses. To boot from a hard drive quickly, perform these steps:
1. If applicable for a SAN-boot architecture, have your SAN administrator map only the root disk
to the server. Or, pick a unique size for the rootvg disk so that it will stand out from the rest
easily.
Solaris to AIX

Page 2 of 11

ibm.com/developerWorks/

developerWorks

2. Boot the system into the SMS, and open a terminal window (if applicable).
3. Click Select Boot Options, then click Select Install/Boot Device.
4. Click Hard Drive and navigate to the type of disk desired, or click List all devices.

The Object Data Manager and System Management Interface Tool


In the Solaris world, everything is tracked through flat files. Networking is covered by files like /
etc/defaultrouter, /etc/netmasks, and /etc/hostname.bge0. Software packages are tracked down
via the /var/sadm/install directory path, and file systems are mapped to device files like /dev/dsk/
c0t0d0s0 with /etc/vfstab. There are a few menu-driven systems for administration, but tools like
admintool are limited in scope and functionality.
AIX uses a specialized database similar to the Windows Registry called the Object Data
Manager (ODM). The ODM retains information on those same aforementioned tasks typically
managed by flat files on Solaris. Although not directly editable by vi or text-based editors, ODM is
controlled by several higher-level commands that add routes, detect new hardware, and so on.
One of the simplest ways to interact with the ODM and administer the server is through the System
Management Interface Tool (SMIT). This robust menu system allows you to do everything from
shutting down the server to changing tuning settings. By typing smit or smitty in a command
shell, SMIT allows you to perform common tasks and, more importantly, learn the command-line
instructions (by pressing F6 to view the commands executed behind the scenes). Fastpaths to
specific commands, such as using smitty shutdown to shut down the server, are also available;
you can determine these fastpaths by pressing the F8 key.
You don't manage everything using the ODM or SMIT, however. There are still plenty of flat files
that the two operating systems have in common, such as /etc/passwd, /etc/group, and /etc/hosts,
and that you can modify directly. But subtle differences do exist, such as /etc/shadow versus /etc/
security/passwd or /etc/nsswitch.conf rather than /etc/netsvc.conf. Unfortunately, there is no rule of
thumb as to which files are identical between the two operating systems or what the ODM may not
cover, so be sure to take notes and check online resources to learn the differences between files.

Hands on with SMIT


Within the SMIT menus, it helps to know what the prompts mean and how to select items. An
asterisk (*) means that a field is required. A plus sign (+) means that you can press F4 to see a list
of options. If you want to search through those options, press the slash key (/) to look up items by
keyword, or press F7 to select multiple options. For example, to create a user with SMIT, perform
these steps:
1. As root, type smitty mkuser.
2. Give the user ID a name in the User NAME field.
3. In the Group SET field, press F4, then use the F7 key to select the staff and lp groups.
4. In the HOME Directory field, give the user ID a different home directory location.
5. Press F6 to view the command that will run. Press F3 to go back, then press Enter to run the
command.
Solaris to AIX

Page 3 of 11

developerWorks

ibm.com/developerWorks/

6. Press F10 to exit SMIT, and set the user's password by running passwd $USER.

Device control
One other virtue about AIX is its ability to detect and manage devices easily. With Solaris,
determining which device correlates to something like /ssm@0,0/pci@1c,700000/pci@1/
SUNW,isptwo@4 (scsi) is sometimes complicated. Or, if you have the wrong set of device drivers
installed, running devfsadm might not configure that Fibre Channel adapter you're trying to set up.
In AIX, the cfgmgr command detects any hardware devices present on the system and even tells
you which device drivers may be missing. A common set of commands, such as lsdev, lscfg,
and lsattr, provide information on the devices installed in plain English, their base configuration
information, and any attributes or tunables set for the devices. Devices can be presented in one of
two statesavailable or defineddenoting whether the device can be used. All of this information
is tracked in the ODM, where it persists across reboots and can be ported easily.

Hands on with device control


AIX often creates logical devices that point back to physical counterparts. For example, an
Ethernet adapter of ent0 has the logical device of en0 bridged off of it. This is done, because
although a setting like the MTU size controls the physical aspects of the system, an IP address
does not physically change the adapter. For example, to check a Fibre Channel device for parentchild devices and configuration settings, perform these steps:
1. Run lsdev C | grep fcs to view all Fibre Channel devices on the system.
2. Note the address (third) field, and pick one of the adapters.
3. Run lsdev C | grep $ADDR, substituting the address field to view the child devices.
4. Run lscfg vl $FCS to get the Worldwide Number of the adapter.
5. Run lsattr el $FSCSI to see the tunable options for the device.

Software and patching


Solaris breaks down its operating system software and patch management into two main levels:
versions and patch clusters. These levels are based on the system's kernel levels and software
package subsets. In contrast, AIX is broken down into four main operating system levels: version,
release, technology level (TL), and service pack (SP). Version and release are most commonly
referred to in the name of AIX, such as AIX 7.1. TLs are major releases of the operating system
that have been updated, and SPs contain minor updates. Running a command like oslevel s
identifies a code like 6100-04-05-1015, which shows that this server is at AIX 6.1, TL 04, SP 05.
Each installed software component on AIX is commonly called a fileset, though the word package
is used when discussing a compiled Red Hat Package Manager package that can be installed on
AIX. These filesets are gathered into larger Licensed Program Products that can be maintained
individually on the server, like an installation of IBM DB2. But you can view any piece of software
registered with the ODM using the lslpp command, checking its level and all associated files.
Solaris to AIX

Page 4 of 11

ibm.com/developerWorks/

developerWorks

Hands on with patching


IBM sometimes releases Authorized Program Analysis Reports (APARs) that fix particular bugs.
You can view APARs using the instfix command, but this command has a second feature, as
well: It tells you whether specific filesets are missing from a particular TL. Use this procedure on a
server after installing newer operating system filesets:
1. Check the server's operating system level by running oslevel s.
2. Run the instfix i | grep ML command to determine whether any TL has missing filesets.
3. Run the instfix -i | grep _SP command to determine whether any SP has missing filesets.
4. If any TLs or SPs have missing filesets, run instfix -icqk $TL | grep :-: to see which
filesets are missing.

File systems
I was thrown off when I first started learning Solaris and heard that out of the box, I could make a
maximum of seven file systems (slices) per disk and that once those file systems were carved up,
their sizes were pretty well fixed. AIX is much more versatile, in part because of how its Logical
Volume Manager (LVM) works.
Figure 1 shows a series of disks with small cylinders. One cylinder, labeled Physical Volume, has
a grid drawn on it to represent the individual physical partitions (PPs) on it. Above the physical
volume (PV), three other cylinders are grouped by an ellipse showing how they together represent
a volume group (VG). A rectangle running across the three disks shows a logical volume (LV)
placed across the three disks.

Figure 1. LVM on disk

In AIX, disksknown as PVsare divided into smaller homogeneous blocks of storage called
physical partitions. These PVs are gathered logically into VGs; within VGs, LVs are laid down
across sets of PPs to build structure onto the PVs. If there is disk mirroring, they may be mapped
to one LPs that points to two or more PPs. Then, a JFS or an Enhanced JFS (JFS2) is established
on the LV, tracked with the /etc/filesystems file automatically, and ready for use. Once they are
mounted, they can be dynamically grown, shrunk, or modified, and adding more space to the VG
is as simple as adding another PV. This is similar to how Veritas Volume Manager can provide disk
management tools but comes integrated with the AIX operating system at no extra charge.
Although Solaris does let you customize your file system layout, even going as far as having
one root file system to contain everything, AIX comes with several default file systems for more
Solaris to AIX

Page 5 of 11

developerWorks

ibm.com/developerWorks/

granularity: /, /usr, /var, /tmp, /home, /opt, and /admin. However, in AIX, the /tmp file system is not
the same thing as swap or virtual memory. Instead, such memory is managed by paging space,
which resides on one or more independent LVs.

Hands on with file systems


The AIX grep command has an interesting p flag that lets you search for information according
to paragraphs rather than just for lines. This flag is useful when searching through the /etc/
filesystems file for specific file system information, because it pulls back just the stanzas pertaining
to the file system you want to examine. Try the following procedure:
1. Pick a file system like /home, and use grep p to gather its information out of /etc/filesystems.
2. Examine the same file system with the lsfs command to see what information is similar and
different between it and the grep output.

Services management
With the advent of the most recent version of the operating system, Solaris 10 took a page from
the play book of how other operating systems like Linux manage services and created the Service
Management Facility. Using commands like svcs and svcadm, this concept was brought in to
handle things like inetd and NFS with simple start-and-stop directives.
AIX uses the System Resource Controller (SRC) to accomplish the same goals. This daemon,
srcmstr, which gets kicked off by the /etc/inittab file at boot time, controls various services by
breaking them into groups, subsystems, and sub-servers. For example, the tcpip group contains
the inetd subsystem, which manages the Telnet sub-server.
Figure 2 depicts a series of concentric circles illustrating the hierarchy of srcsmstr groups,
subsystems, and sub-servers. The largest circle is labeled with the encompassing srcmstr
process. Under that, as an example, is the tcpip group. Within that is a circle labeled subservers,
which has some of the various tcpip sub-servers, such as snmpd and named. Finally, the
innermost circle has subsystems of inetd, including Telnet, FTP, and RSH.

Solaris to AIX

Page 6 of 11

ibm.com/developerWorks/

developerWorks

Figure 2. The srcmstr in action

These individual groups can be found in configuration files in /etc, typically with an rc. prefix, with
a corresponding entry in the inittab or something called from the inittab, such as /etc/rc.tcpip,
managing those services just described.
However, since AIX 5.1 came out, it is also possible to start and stop some services and
applications similar to the System V model used in prior versions of Solaris. The /etc/rc.d directory
structure can be used to put in S and K files at the various run levels listed to bring up or down
processes. But AIX does not adhere quite as strictly to the run level model as Solaris or Linux, so it
is important to place any custom start and stop scripts in the right places.

Hands on with services management


There are many ways to skin the proverbial cat for starting processes at boot time. In the past,
many administrators would add an rclocal entry to the inittab to kick off a local, home brew /
etc/rc.local script to load applications. But since AIX 5.1, it has become possible to make your
own service kicked off by srcmstr for applications such as SSH or Samba by using the mkssys
command. Try the following process:
1. Install a third-party application like SSH to start or stop scripts. Note where the executables
are located.
2. Use the following command to set up the SSHD daemon to have it run as root, with standard
signals for stopping the process (15 for normal, 9 for kill), and be a part of the test group: /
usr/bin/mkssys -s sshd -p /usr/sbin/sshd -u 0 -S -n 15 -f 9 -G test.
3. Start the subsystem by running startsrc s sshd.
Solaris to AIX

Page 7 of 11

developerWorks

ibm.com/developerWorks/

VIO and WPAR


Recently, virtualization has become a big buzzword in the UNIX field, with companies trying to
determine how to get the most bang for their buck. In Solaris, this push led to the introduction of
zone technology, where a global zone server would have several zones underneath it that contain
their own micro-partitions running as independent servers. Drawbacks of this technology include
how the zones use disk space from the global zone itself, and the operating systems have to be
the same level, with any patching that warrants a reboot affecting all zones.
In AIX, virtual I/O (VIO) expands on this idea. The main premise of VIO technology is to create
specific servers on a piece of System p hardware that aggregate and use the underlying disk and
network resources, serving them out to all of the AIX servers in a passthrough function. With VIO,
the individual LPARs can all be at various operating system levels but use the Fiber Channel,
SCSI, and Ethernet resources concurrently. Building two VIO servers into the same System p
hardware can provide redundancy in case one component fails.
Figure 3 depicts two sets of computing environments, illustrating Solaris zones and AIX VIO. On
the Solaris zones side, you can see how the servers are held up by the global zone and that one
outage there will make all of the resources unavailable. On the AIX VIO side, you can see how
the VIO servers act as a passthrough for the LPARs and that an outage in one VIO server will
not affect the larger architecture. Similarly, it's clear how the LPARs can be at disparate operating
system levels, unlike the Solaris zones.

Figure 3. Solaris zones versus the AIX VIO

Similar to zones, IBM introduced WPARs in AIX 6.1, which allow for the creation of mini-servers
that run under the auspices of a larger system. But these WPARs can also be configured to
traverse multiple AIX servers across System p hardware instead of being stuck on one piece of
equipment.
Solaris to AIX

Page 8 of 11

ibm.com/developerWorks/

developerWorks

Hands on with VIO


A specialized superuser, padmin, is used for most aspects of setting up VIO servers. However,
although the commands available to the padmin user look the same as many of the root-level user
commands, they often have a vastly different flag or word structure. Perform the following steps:
1. Log in as padmin, and run the help lsdev command. Note the flags that are available.
2. Become root by running oem_setup_env (notice that you don't use the standard su command).
3. Run man lsdev, and check the flags available. Compare the flags between these two
commands: Note that padmin is far more verbose.

Conclusion
Although picking up another operating system takes time to learn all its nuances and particulars,
this article provide background to put Solaris administrators more at ease when migrating to AIX
servers. You should now have the base-level foundation to help you understand the key concepts
and main differences between the two operating systems.

Solaris to AIX

Page 9 of 11

developerWorks

ibm.com/developerWorks/

Resources
Learn
Read more about the HMC in the Hardware Management Console V7 Handbook.
If you're an RHEL 4 administrator, you will find the developerWorks article, AIX tips for
RHEL4 administrators (Christian Pruett, November 2009), helpful.
Find helpful VIO tips and tricks in the developerWorks article, The VIO cheat sheet (Christian
Pruett, July 2010).
Check out the UNIX Rosetta Stone.
AIX and UNIX developerWorks zone: The AIX and UNIX zone provides a wealth of
information relating to all aspects of AIX systems administration and expanding your UNIX
skills.
New to AIX and UNIX? Visit the New to AIX and UNIX page to learn more.
Technology bookstore: Browse the technology bookstore for books on this and other
technical topics.
Get products and technologies
Try out IBM software for free. Download a trial version, log into an online trial, work with a
product in a sandbox environment, or access it through the cloud. Choose from over 100 IBM
product trials.
Discuss

Follow developerWorks on Twitter.


Participate in developerWorks blogs and get involved in the developerWorks community.
Get involved in the My developerWorks community.
Participate in the AIX and UNIX forums:
AIX Forum
AIX Forum for developers
Cluster Systems Management
Performance Tools Forum
Virtualization Forum
More AIX and UNIX Forums

Solaris to AIX

Page 10 of 11

ibm.com/developerWorks/

developerWorks

About the author


Christian Pruett
Christian Pruett is a senior UNIX systems administrator with more than 14 years of
experience with AIX, Sun Solaris, Linux, and HP/UX in a wide variety of industries,
including computing, agriculture, and telecommunications. He is the co-author of two
IBM Redbooks on AIX, has served as a UNIX book review for OReilly Publishing,
and has worked on several of the IBM AIX certification exams. He resides in Colorado
with his wife and two children. You can reach Christian at pruettc@gmail.com.
Copyright IBM Corporation 2011
(www.ibm.com/legal/copytrade.shtml)
Trademarks
(www.ibm.com/developerworks/ibm/trademarks/)

Solaris to AIX

Page 11 of 11

Vous aimerez peut-être aussi