[Discuss] Dell Openmanage vs Supermicro IPMI?

[blu at nedharvey.com (Edward Ned Harvey (blu))]

> From: discuss-bounces+blu=nedharvey.com at blu.org [mailto:discuss-
> bounces+blu=nedharvey.com at blu.org] On Behalf Of Scott Ehrlich
> 
> We are looking at a more expensive quote for a Dell PowerEdge Server
> with the option of OpenManage software, and a quote at half the cost
> for a non OEM system from a VAR using a SuperMicro board and
> off-the-shelf parts.
> 
> In addition to wanting a server with a lot of storage, we also want to
> remotely monitor the system's components, leading to OpenManage or a
> more generic IPMI option.
> 
> I've had mostly exclusive experience with Dell OpenManage on Dell systems.
> 
> What are your experiences with OM vs IPMI?   Is it purely name and
> cost (Dell vs off-the-shelf), or is there more that I should know or
> be aware of?

They're dramatically way different.  That's not to say one is always hands-down better than the other - it's only to say they're dramatically different.

Dell OMSA can be accessed directly via ssh, https, or other protocols, or you can centrally manage with the OMSA management station.  It *does* provide you with some Dell-specific functionality such as reconfiguring the RAID card remotely (occasionally useful).  But most of the value it provides could be done with IPMI and some monitoring software.

Basic IPMI can be very difficult to secure - And only provides basic functionality such as power cycling.  I have found that every manufacturer implements it differently; if you find a process that works for Dell servers (using IPMI instead of OMSA) that same process doesn't work on Sun servers, or HP, or SuperMicro.  They're all different, but they all have *some* way of working.

My advice is to only buy the big name brand server with super warranty and stuff, when you have an individual machine that is mission critical.  My advice is to buy the generic hardware whenever you're going to use a compute cluster where you can plan for and afford 10%-20% of the machines to be down at any given time (although something like 5% is more typical.)  By buying the generic hardware and either a basic warranty or no warranty, you can get 2x as much stuff for the same price, which is cool when you scale your environment, but not cool if any individual machine is mission critical.