Remote monitoring and management/administration

Basher · 14 Oct 2006 at 15:59

Hi guys,

As part of my final year at Uni, I'll be looking at depth into remote monitoring, management and administration software, both in and out of band. I'll be looking at wether technology has reached a point whereby companies can succesfully implement completely lights out remote data centres.

What I'd like form you guys is your opinion on this, do you have any opinions/experience with associated technology, what do you think? Do you believe tehnology is good enough yet? What are your opinions on the iLo vs KVMoIP debate?

Any thoughts and opinions will be great.

Cheers,

James

rpstewart · 14 Oct 2006 at 16:23

You'll never be able to have completely lights out data centres simply because the needs of customers are always changing and as a result there are times when you need to physically go into the box and add new netwok cards, CPUs, SAN cards etc as the customer's needs increase. Doing that however is increasing rare as most of our machines get plumbed in with all these things fully kitted out now. The advent of "golden key" systems is reducing this further - the customer orders a 4 way machine but a 16 way box is supplied and then the customer pays to upgrade to 6-, 8-, 12- or 16-way as and when they see fit. The upgrades can be done online without interruption to service rather than the 4-8 hour downtime that a physical upgrade could take. SAN based storage has the same principle, new enclosures and disks can be added into the SAN without disruption and allocated around the data centre without the need to disrupt operating machines.

The software has been pretty much there in the midrange (UNIX, Solaris, AIX etc) and mainframe markets for years - as long as the network stays up a telnet or ssh session will run for as long as you like. The application software tends to be much more stable in these environments too, I have instances of DB2 that I look after which have been operating continuously since we moved to v8 over 18 months ago.

I think one of the most telling things about the move to remote data centres is that, with the exception of the odd piece of development kit, I haven't physically set eyes on any of the servers I've worked with in the last 3-4 years!

Basher · 15 Oct 2006 at 14:29

Cheers, thanks for the input. What exactly do you mean by 'golden key systems'? It's not a term I've heard used before.

Thanks again,

James

rpstewart · 15 Oct 2006 at 14:40

It's basically a system with built in upgrades. The system is delivered with a full complement of CPUs, RAM etc but the customer only pays for a smaller system. Should the customer then want to upgrade all they have to do is turn the golden key (ie pay for the upgrade) and unlock the hardware that's already in place. The server OS then dynamically picks up the new hardware and continues without even needing a reboot.

Basher · 15 Oct 2006 at 15:47

rpstewart said:
It's basically a system with built in upgrades. The system is delivered with a full complement of CPUs, RAM etc but the customer only pays for a smaller system. Should the customer then want to upgrade all they have to do is turn the golden key (ie pay for the upgrade) and unlock the hardware that's already in place. The server OS then dynamically picks up the new hardware and continues without even needing a reboot.

Hi, thanks for the information. Is there another term for this? I can't seem to find any information on it (by searching for golden key)?

Regards,

James

rpstewart · 15 Oct 2006 at 19:37

I've only ever heard it referred to as a golden key. You might find some more info on it in the Utility Computing section of IBM's OnDemand strategy

EDIT: Looking at the site the Autonomic Computing section is what you're looking for in the context of your original post.