DELL and CISCO firmware
management philosophies are very different.
DELL has server oriented
approach (similar to HP) and CISCO has network centric approach.
DELL System and Firmware
Management
DELL System and Firmware
Management Approach is year by year better and better but in my opinion still
not optimal. But the future is bright and happy J
We have lot of possibilities how
to do firmware update and unfortunately sometimes you have to test all of them
to be successful L
12-th generation of servers is
far the best because lifecycle controller is significantly faster and less
problematic than in 11th generation.
I don’t want to go to deeply
into specific firmware update problems – and usually there are some ;-) - so
I’ll keep it in more general.
CISCO System and Firmware
Management
CISCO UCS has single management
software for servers embedded in the hardware UCSM (UCS Manager). It is running
inside network interconnects (Fabric Interconnects) and because they are two
interconnects it is in high availability cluster (active/passive). UCS Manager
allow you to do all UCS configurations and also firmware management of all
components (Server adapters, Server BIOSes, IO Modules, Fabric Interconnects
and UCS Manager itself).
CISCO release firmware packages
which must be downloaded into UCS and these firmware’s can be applied. Upgrade
order is very important – starting from IOMs, then Fabric Interconnects and
lastly UCS Manager.
Server and server adapter
firmware management can be included into server profiles. Server profiles is
something like AIM personas. It is a logical representation of the server and
BIOS + firmware versions can be specified there. When Server profile is applied
(associated) to the server then BIOS + firmware is upgraded or downgraded as
defined in the profile.
Server upgrade procedure is done
out-of-band and server cannot run operating system – therefore maintenance
window has to be planned. It takes a while. Internally it works over PXE boot.
Server is automatically reconfigured to boot over PXE where PXE and TFTP is
provided internally by UCS Manager. Upgraded server boot special linux
distribution (CISCO call it PNU linux) and firmware packages are applied in
this temporarily running linux system. After upgrade the server boot
order is changed back and server boots normal operating system.
COMPARISON
Both firmware management
approaches are totally different. CISCO has centralized system leveraging
internal PXE/TFTP where DELL has distributed system where lot of lifecycle
controllers are orchestrated by some 1:many management software.
When I work for CISCO lot of
customers were really scare to do UCS upgrade by them self. I can understand it
because CISCO UCS is not simple system. CISCO UCS is unified system and when
you make mistake during fabric interconnect upgrade you can be in troubles.
Therefore customers usually engaged CISCO Advanced Services or certified
partners.
When I work for DELL Services I
had also several engagements for firmware upgrades because DELL customers are
not aware about OpenManage framework and various firmware possibilities.
If DELL customer want to do
firmware management by them self I usually do 3 day System Management workshop
engagement to explain them practically the architecture and system management
possibilities.
CISCO advantage
·
Unified and
centralized firmware management
·
Firmware can be
defined in Service Profiles
CISCO disadvantages
·
Centralized and complex
system – therefore customers are afraid to do upgrade by them self
·
Proprietary system
even inside using standard protocols like PXE/TFTP
·
Longer server
downtime – I don’t know how it is today but 3 years ago CISCO hadn’t operating
system update packages for BIOS and firmware (something like DUPs) –
disadvantage mitigation: they expect some form of cluster to eliminate
downtimes
DELL advantage
·
Advantages of
distributed system – if one server upgrade fails it doesn’t impact whole
system
·
Dell Update Packages
(DUP) which can be applied via running operating system – OMSA
·
Out-of-band upgrades
via lifecycle controller – firmware staging and application after next server
reboot
·
Open system from
management point of view – WS-MAN, racadm
·
DELL disadvantage
·
Lot of software
components customer must be aware (DELL Repository Manager, Open Manage
Essentials, Lifecycle controller, CMC, …) – but it is necessary to
support all environments
·
Sometimes it doesn’t
work as expected and you have to use another tool or upgrade Lifecycle
controller to higher version and so on – it is much better on 12th
server generation and iDrac 7 and OME 1.2+
Hopefully we will do continuous
improvements in this area.
The best and most optimal
DELL Firmware management strategy really depends on customer environment.
It depends on following:
How many servers do they have?
Do they want to use 1:many
firmware management like OpenManage Essentials, Altiris, MS System Center,
VMware OpenManage Integration?
Do they want to integrate it
with some existing system management (Microsoft, VMware) and
configuration management?
And we have to show to our
customers how it works. Think about Proof of Concepts.
I understand benefits of both
approaches and nobody can say exactly one is better than other. As always – it
depends.
No comments:
Post a Comment