Archive for February, 2007

http://h18023.www1.hp.com/support/files/lights-out/us/revision/6147.html

The firmware update, Version 1.82, contains critical bug fixes and is the minimum version required. Please perform the update at your earliest convenience. Click on the underlined Version to view more details on the fixes.  

You can upgrade the firmware remotely and online easily, I posted an example script. I’ve done this before without issue, it’s worth doing to fix the bugs and avoid any issues from a support angle in the future, think about the system firmware as well, it should be updated at some point, maybe your next security patching process?  Or when the system is down for an upgrade/maintenance.

Also, remember the array firmware, this can be quite important but for the older legacy compaq proliants (1850/6500/2500/1600) you might find the individual disks might need their firmware upgraded, and that can involve the 16 floppy disk scenario. On anything G1/G2 and above you’ll be fine typically, apply firmware ILO, system, array and reboot.. Do test it first though, and not on production until you’re happy.

http://www.prdomain.com/companies/I/IBM/newsreleases/200722739266.htm

IBM Maintains Leadership in Blade and UNIX Servers  
26 Feb 2007 , Armonk, NY : IDC reported today that IBM (NYSE: IBM) continues to hold the number one position in worldwide server revenue share with 32.8 percent revenue share for 2006 [1].

IBM’s leadership position in global server revenue in 2006 was augmented by noteworthy revenue growth in its System z™ mainframe business [2].

In 2006, IBM introduced the System z Business Class mainframe, teamed with industry leaders such as Oracle and SAP to drive innovation on the platform, and extended its customer base and drove new workloads onto the platform through specialty processors for Linux® and Java™ workloads.

IBM continue to offer a competitive, scalable and adaptable blade infrastructure solution, I know their x86 and the power range of blades have been attractive alternatives on the blade market, good to see them doing well. The more choice the customers have, the more chance that we will be able to choose and find a solution comprised of blade technologies. Regardless, as long as IT delivers, what it runs of is of little significance in the grand scheme of things.

http://www.hp.com/hpinfo/newsroom/press/2007/070227a.html

HP Reduces Customer Administrative Costs and Wait Time with Blade Network Virtualization Technologies

PALO ALTO, Calif., Feb. 27, 2007 On the heels of taking the No. 1 position in factory revenues and units shipped for blade servers,(1) HP today introduced virtualization technologies that simplify the connectivity and management of its HP BladeSystem c-Class architecture, the company’s core solution for the next-generation data center. HP also announced the launch of its first workstation blade for the c-Class blade portfolio, the HP ProLiant xw460c Blade Workstation, to provide users with increased performance and scalability.

Dramatically simplifying network connectivity and common server management tasks across the data center, the HP Virtual Connect Ethernet and Fibre Channel virtualization modules enable administrators for the first time to wire just once and then add, replace or recover servers on the fly. In addition to lowering capital and operating costs, the modules deliver simplified management capabilities that reduce the time it takes to execute these actions from hours or days to a matter of minutes.

The XW460c blade workstation does look fantastic, the HP trader workstation range, the XW6000-9400 have been infamous in the banking sector as industrial strength, I’m sure HP will offer the same with their blade workstations, the technology has been tested now, one of the banks is already using them.

Virtual connect looks very interesting and it’s worth checking out their web seminar or reading the text for the press release, though http://h71028.www7.hp.com/enterprise/cache/459411-0-0-0-121.html does explain it very well, it aims to use the technologies available to reduce the lead time in deploying infrastructure which does look very good and I am very impressed. Will need to read the whitepapers to find out more.

Inefficient use of resources

Follow the sun allows a more dynamic IT support model, one where we can share at the very least first and maybe second line support overnight allowing your local team to be on call, but at the same time not be up all night unnecessarily if the issue can be dealt with by whoever is working at that part of the business day.

Poor Global Support

Often a system might have dependencies on systems in other locations, for example, the London batch might be dependent on the Singapore system being available, often you might find that the application teams have access to the systems in Singapore, but the infrastructure don’t, so when the system goes down, someone in Singapore has to be paged, causing delays to the resolution time. If I had the password/access, I might be able to log on and check what’s going on, do my best to stabilize the platform for the time being.
 
Everybody thinks local

IT thinks differently in different locations, which makes deploying systems that bit harder, the fact that Singapore might use Hitachi storage, or they don’t buy HP, they buy Dell, makes the support, the project implementation that bit harder, and standards might vary between sites. By bringing the IT teams together in a follow the sun model, we can’t change everything and make everything equal, but we can move towards a standardised approach, an agreed way of building a cluster, how we deploy, and build blades. The idea being that if Capital Markets request a blade in London, it should be as close to identical as the blade in Singapore.

 

http://web.mac.com/martinmacleod/iWeb/Site/Blog/bladecab_layout.pdf

As you deploy blade enclosures, keeping a note of what the layout looks like, and what they are being allocated too is important. I’ve put together a basic example for a deployment, consisting of 48 HP blades, note the same would apply to IBM/Sun/Dell though the number of blades might differ. The diagram gives you an example of the cabinet layout for BL30p/BL35p blades.

Try and operate on an enclosure basis, allocating individual blades to a task can make administration that bit more difficult and couple that with naming conventions and the game might be over. Remember, that you may have to take down a cabinet for maintenance, trying to minimize the number, the range of applications will make this easier, or at the very least sharing that cabinet between a business line, suite of applications.