Hi there,
Its a bit of a conundrum this UCS system, so here is my 50 pence on the system against the HP alternative as currently BladeMatrix is its only competitor.
Sorry if i waffle a little here
Fore-word: For the record I actually like Dell rack-mounts.
The first thing to keep in mind is that Cisco and HP currently have a strained relationship, stemming from the fact that HP started selling Networking kit and then Cisco decided to perform a two finger salute and started to invade HP's Datacentre Server stronghold with UCS.
This is important to understand if you are currently a HP server shop who has just invested big money in a Cisco Nexus setup because it is not good news, the industry needs them both playing ball with each other.
Anyway, as the HP Bladematrix solution uses a storage automation component (using their EVA solution), the direct competitor to this is actually EMC VBlock and not Cisco UCS however, for the purposes of a shorter post I will ignore VBlock and talk about Cisco UCS vs. BladeMatrix (without EVA)
It is important to understand at this point that, in my humble, neither solution (HP BladeMatrix or UCS) currently holds all the card's however, it's the HP solution which is technically the weaker out of the two and I will describe what I see as the current advantages for UCS.
Firstly, the Cisco UCS solution is a clean sheet design built specifically for Virtualisation and Convergence of IP and Storage Networks. This means that the blades can be configured with large amounts of memory and the whole Architecture is based around very very fast IP Networking, provided by Cisco.
I will leave to one side the benfits of 10GbE to the Blade, both solutions either support it now (UCS) or will do in the very near future (HP) so lets leave that to one side.
One key differentiator however, is around convergence of Storage and IP Networks. UCS and Cisco Nexus support for converged Networking is there now, the HP solution is based around the oldskool seperate FC+IP setup. There will be a convergence path, its just not there right now.
If you are not ready to converge completely for various reasons (older network, not confident etc.), because of the flexibility of the UCS package, you can hook up the UCS to your existing fibre-channel network using the packaged Cisco 6000 series fabric extenders, this avoid's the requirement for having a Cisco Nexus network to plug the UCS systems into in order to obtain this convergence of Networks (Storage & IP) and crack on without the Nexus network being in place.
There is additional cost there in buying the extenders, but keep in mind that the cost for these switches is offset by not needing to buy chassis modules, like the HP.
Where this convergence route may be seen as a dis-advantage or vendor lock-in (because you need a pure Cisco environment) it would be wise to point out that while the HP alternative can support 10GbE and of course Fibre Channel, it supports it only with HP Networking kit, so the same criticism can be levelled at both solutions. The only difference is that UCS is built for the future, you will need to junk at least some of your HP Networking modules and mezzanine cards to upgrade to a converged network later.
The UCS solution scores here because as soon as you are ready to remove your Fibre Channel Network and replace it with a converged Nexus solution, you won't have to worry about junking any of your blade equipment.
The other current major advantage Cisco has is that the current VMWare V1000 switch does not work with the HP BladeMatrix solution, it is un-supported. So again, if you have just built a multi-million pound Nexus network and expected this to work with the HP solution, you are out of luck.
This is a major head-ache if you have a large VMWare environment. Your Network team will be unable to manage the Distributed Switch like any other Cisco switch.
Both solutions support some sort of stateless, boot-from-san approaches and both support some sort of Host Profile build system although again, as the UCS is a clean sheet design, it just seems to work better.
So, Cisco is a generation ahead of HP on the hardware side, for now.
Where it gets a bit fuzzy for UCS, is on the software automation side.
Again, because the Cisco solution is a clean-sheet design the software which provides the bare-metal installation of the OS onto the Blade is just better than the HP solution, not because Cisco wrote it but, they quite cleverly aligned themselves with BMC Bladelogic, which as far as I am concerned is probably the best system on the market for this type of thing.
BMC Bladelogic provides the base OS installation onto the UCS Blades and BMC Atrium Orchestrator provides the automation system this allows you automate hardware provisioning tasks with Bladelogic and VMWAre itself.
So far so good, however there is a bit of a problem here.
And the problem is that BMC don't have a stellar reputation for being particularly good value for money. If you want to say, expand this system into the full "Self Service" automation nirvana you probably want it to be, you are going to have to pony up a significant amount of money to BMC in order to realise this dream.
The BMC software is fantastic at the moment, they seemed to acquire lots of good start-up's and have done some great work in trying to integrate them but expect to pay $$$
It is worth noting at this stage that if you don't really worry about a fully automated system from service desk ticket to ful-fillment, you could probably get away with just using VMWare director to automate and control the Virtualisation layer and obtain your service nirvana that way.
I do feel that in order to keep BMC and VMWare happy Cisco may have opened up a gap for HP to operate in and expose on the software automation side.
Because, as cobbled together and convoluted the HP system looks and feels, what it does provide is an out-of-the-box single vendor software solution, with a self-service console. The GUI for orchestration is easy to use and you do not need to be a programmer to get the best out of it.
One problem with the HP system is that it appears to be lots and lots of bits strung together with (what used to be) Opsware at its heart. The BMC system is proven and can be upgraded to automate pretty much anything and while this system is mature, the HP BladeMatrix system is currently in an early stage of maturity in comparison.
I also have to question how easy it is going to be to troubleshoot BladeMatrix automation problems without specialist assistance from HP.
All said, the HP software solution probably hits the spot where small to mid-sized customers are currently feeling pain and it probably beats UCS in this regard. Not because it is technically better but because it ticks a couple of key boxes without any additional investment.
I also noticed that the hardware support renewal costs for UCS are significantly lower than for HP, this needs to be acknowledged and factored into the business case.
Cheers.
![Shaz]sigh[](/styles/overclockers/avatars/4.png){kind=avatar}
) and I prefer the direction HP is taking around converged infrastructure and the network as a component of this. 
