inspection of your ServeRAID battery. We have seen a number of installations where the ServeRAID

adapter has failed due to either a depleted or defective battery. This has resulted in unscheduled outages

and extreme high levels of stress.

IBM Retain Tip H001648 states: “ Users should install and monitor battery conditions using the IBM

ServeRAID Manager application for the following IBM ServeRAID controller models: 4M, 4H, 4Mx, 5i, 6i,

6i+, 6M, 7k, 8i, 8k, and 8s. All have cache backup batteries of which the battery status can be checked on

the controller properties Status tab.

Users should install and monitor battery conditions using the IBM MegaRAID Storage Manager (MSM)

application for the following controller models: MegaRAID 8480, IBM ServeRAID MR10i, MR10is,

MR10ie, MR10M, MR10k, and M5015. If installed, the cache backup battery status can be checked on

the battery backup unit (BBU) properties tab.”

We emphasize the need to visually inspect the battery. We have seen a number of instances where the

batteries had swelled prior to failing. IBM Retain Tip H191476 addresses this and provides guidance on

replacing the defective part.

http://www-947.ibm.com/systems/support/i/planning/software/i5osschedule.html

This morning, I just had an email exchange with a customer regarding AIX 5.3, and the looming EOL date. This particular customer was somewhat caught off-guard that AIX 5.3 would be going EOL quite so soon.

The most current Releases are: AIX 5.3 TL12 SP5, AIX 6.1 TL7 SP1, AIX 7.1 TL1 SP1.

If you cannot leave 5.2 or 5.3, and wish to maintain support, it is allowed to run, fully supported, as a WPAR under AIX 7.1. Need more information?

http://publib.boulder.ibm.com/infocenter/aix/v7r1/index.jsp?topic=%2Fcom.ibm.aix.wpar%2Fwpar-overview.htm

To check your software products and their position in the software lifecycle, IBM has this page:

http://www-01.ibm.com/software/support/lifecycle/

Don’t be caught by surprise. It’s been a good run, but much improves in 6.1 and 7.1 particularly with respect to harnessing the capabilities of our newer processors.

Can we at ThinkASG help you with modeling or testing migration to AIX 6 or 7? Our BPIC has two Power7 servers for use. We commonly use our resources to assist with upgrades, testing, resource virtualization and performance modeling.

AIX 7.1 TL1 – and oddly enough, SP1 for that TL was released on the same date, October 17th.

For the many of you that have been waiting for TL1 before beginning with AIX 7, this is your signal.

I have worked with three customers this week that had significant issues in production with this particular HMC Code. Fortunately, none of the customers experienced any downtime, merely anxiety and inconvenience.

If you have, or are planning to upgrade to HMC 7.7.3:

• Get it to Service Pack 1
• Get all available PTFs applied. (list below)
• Do not stay logged in to the HMC. Do your work and log out.

When I last spoke with a support rep, I learned that IBM is very aware of the issues and working on fixes to the remaining issues.

All that said, HMC 7.7.4 was released today. I will be applying that in our BPIC later this week. I’m hopeful that some of the bumps and bruises from 7.7.3 are gone in 7.7.4.

Service pack/Patch Details:

Service Pack 1 is MH01272

Mandatory Fixes:

List of fixes for MH01272

* Fixed a problem where Partition Activation was failing with error B2008105 – SRC.

* Fixed an issue whereby the following error is reported during the creation of a Shared Memory Pool and the adding of a paging device to the pool: “HSC03670 – There was an error accessing the management console persistent storage area”.

List of fixes for MH01281

An issue that caused the repair of the PU Book DCA in a 9119-FHA or 9119-FHB to fail with the error “BPA HARDWARE INTERFACE COULD NOT BE ESTABLISHED”, and then forced the DCA to be replaced non-concurrently. This fix allows the BPA interface to be established and for the repair to continue concurrently.

List of fixes for MH01282

Two issues that can cause the managed system to go to recovery state.

The way I read this, it appears you have to be doing something somewhat specific to get the problem to occur.

Review this for AIX 5.3:

http://www14.software.ibm.com/webapp/set2/subscriptions/pqvcmjd?mode=18&ID=5577&myns=paix53&mync=E

Review this for AIX 6.1:

http://www14.software.ibm.com/webapp/set2/subscriptions/pqvcmjd?mode=18&ID=5582&myns=paix61&mync=E

Question:

Do you know of a study of like 2.5” 10K drives against 3.5” 15K drives? (2.5” 10K 600GB vs. 3.5” 15K 600GB drives) I keep getting like answers, but no real documentation to back it up.

Answer:

Before I begin, I would like to point out that the disk drive is not 100% responsible for the performance of a storage subsystem. You cannot predict the performance of a storage subsystem by the drives alone. Meaning that two different subsystems, with the exact same drive type and count will provide different performance levels. The “secret sauce” is in the controller firmware, and how it handles the I/O’s from the host. This is how the DS3500 can get about 308 IOPS/drive on the SPC benchmark (attached). I’m not saying that a system with faster drives will not perform faster, what I’m saying is that there is a lot more that goes into the formula than just the drive.

For the proposed question, I’ve attached a couple of Seagate Product Manuals. I’ll warn you now that they contain a ton of interesting information with respect to the functional operation of the drives (or at least some may find them interesting).

The corresponding web sites are:

Seagate Cheetah  http://www.seagate.com/ww/v/index.jsp?name=st3600857ss-chta-15k.7-sas-fips-600gb-hd&vgnextoid=8ca77f9853dc9210VgnVCM1000001a48090aRCRD&locale=en-US&pf=1#tTabContentSpecifications

Seagate Savvio  http://www.seagate.com/ww/v/index.jsp?name=st9600004ss-savvio-10k.4-sas-6gbs-600-gb-fips-hd&vgnextoid=ef4c68da2cdc9210VgnVCM1000001a48090aRCRD&locale=en-US&pf=1#tTabContentSpecifications

For a little quick math: The formula I use is IOPS = 1000 (ms in a second)/(avg seek + avg latency). I’ll ignore the transfer time, as IOPS are small.

IOPS Savvio = 1000/(3.8+3.0)=147.

Note, I believe the 2.5”, 300GB, 10Krpm drive is what used for the attached SPC-1 benchmark with 308 IOPS. Yes, that is more than twice the expected average performance on a per drive basis. That is the effect of the controllers “secret sauce” with caching, sorted writes, stripe-aligned volumes, etc.

IOPS Cheetah=1000/(2+3.4)=185.

So, as you can see, for half the power (we all want to be green) and half the footprint, you get about 80% of the performance in this example.The one interesting thing I noticed is that the MBTF for the 3.5” drives is 1.8Million hours, and the 2.5” MBTF is 2Million hours.

than 1,200 IBM software products. FLRT now supports Software Group products running on AIX, IBM i, Linux on

POWER, z/OS, Linux, Mac OS, UNIX and Windows. FLRT provides cross-product compatibility information and

fix recommendations on IBM products. FLRT can be useful when you are planning to upgrade key components

or you want to verify the current health of a system.

www-304.ibm.com/webapp/set2/flrt/

• AIX Version 7.1
• 0.5 Power 7 CPUs
• virtualized as two CPUs
• 1GB RAM

Modest resources, at best, and I was hopeful it would be sufficient for my needs.

After having it up and running with:

• Ganglia Client (reporting its own characteristics)
• Ganglia Server
• Ganglia Web Agent
• NFS Server
• A persisting 10MB/second file copy job across NFS

CPU charts indicated to me that I had over-powered the LPAR by well over 100.0%.

I used Dynamic LPAR tools to decrease CPU core to 0.2 Power 7 CPUs, while keeping the resource
virtualized as two CPUs to the operating system. Looking five minutes later at charts I realized that
this was a well-done trimback of resources, and the system was now periodically spiking CPU, while
still maintaining a good level of idle availability.

Thus, a simple glance for a few seconds at a chart allowed me to free 0.3 CPUs from my environment
for other uses.

What impact would 0.3 CPUs have in your environment? What if this was done ten times over, or on
systems ten times the size of my little Apache/application/NFS server combo?

Considering the combination of costs in an enterprise computing environment, saving entire CPUs can
trigger considerable savings. What is the potential impact for you?

Summary of Ganglia capabilities:

• Monitor many large AIX clusters to monitor performance across large clusters of machines.
• Data is displayed graphically on a website, includes configuration and performance statistics.
• Monitor a group of logical partitions (LPARs) on a single machine – these just look like a cluster to Ganglia.
• Not just limited to AIX, which makes it useful to heterogeneous datacenters.

Ganglia Graphs and How To

Ganglia Home Website

Ganglia for AIX and Linux on POWER Binaries

Contact thinkASG for assistance with installing and configuring the Ganglia monitoring solution.