HP BLc7000 Onboard Administrator Option 412142-B21#0D1 Folheto

Códigos do produto
412142-B21#0D1
Página de 4
The team can set power and cooling thresholds for the 
highest level of performance or the most efficiency, or 
initiate cooling and control cooling levels automatically 
to react to and remove heat. Without this ability to 
have zoned cooling, Continental would have to pay 
for cooling the largest load all of the time.
Clustering saves $490,000 dollars more
Working with Microsoft, Continental is deploying 
HP ProLiant server blades in eight-node clusters for 
high availability. “If we lose one of the physical nodes, 
then all of the VMs on it would be able to failover to 
the other seven nodes in the cluster without causing 
any of them to become over-utilized,” Wilson explains. 
The team manages the clusters using Microsoft System 
Center Virtual Machine Manager (VMM) 2008. 
“VMM monitors the host state and the VM state,” notes 
Wilson. “If the host server crashes, the VM is moved to 
another node. So we have highly available VMs.”
As a next step, the team plans to cluster the VMs 
themselves. “That’s another layer of high availability 
on top of clustered hosts,” observes Wilson. “If a 
service stopped or was hung in one of the VMs, then 
that service would failover to the other VM—totally 
independent of the physical host.”
The new architecture makes it easy to offer highly 
available services. “In the past, we would cluster two 
physical servers for a particular service out of fear of 
hardware failure,” Wilson observes. “Now we don’t 
have to. Project owners are okay with having just one 
VM instead of two physical servers because they know 
that it’s a highly available VM.”
Projected savings from the new clustering architecture 
are considerable. When deployment is complete, 
the team will have been able to avoid 20 hardware 
clusters and save $475,000 in hardware and software 
costs and an estimated $15,600 in annual server 
management costs.
2
 
Reducing planned downtime 50 percent 
Server management overall will be simpler, the team 
reports. One reason is that Hyper-V virtual servers 
are easier to take care of than physical servers. “The 
amount of time we take to patch servers has probably 
been cut in half,” notes Wilson. “Usually when you 
patch a server, you have to reboot it, and that often 
takes longer than anything else—about four or five 
minutes. With the Hyper-V servers, a reboot takes 20 
seconds or less.”
Multiply that by hundreds of servers. “We’ve seen 
a big change in the window we need for monthly 
patches, software installations and things like 
that,” Wilson adds. “Since deploying Hyper-V on 
HP BladeSystem, we’ve had a 50 percent reduction 
in monthly planned downtime for maintenance, from 
10 hours to 5.”
Deploying 18 times faster
Management is streamlined in other ways, says 
Wilson. “Systems Center Virtual Machine Manager 
has wizards in it that make it easy to do a physical-
to-virtual (P-to-V) migration with our legacy physical 
servers,” he notes. “And we can deploy a virtual 
machine in 20 minutes or less. Deploying a physical 
server takes six hours easily.”
2
  12 hours of annual management per 20 (2-node) clusters at estimated 
$65/hour = $15,600.
“Since deploying Hyper-V on 
HP BladeSystem, we’ve had a 50 percent 
reduction in monthly planned downtime for 
maintenance, from 10 hours to 5.”
Richard Wilson, Manager, Enterprise 
Engineering, Continental Airlines
3
Hardware
• HP ProLiant BL460c G5 server blades with Quad-Core Intel
®
 Xeon
®
 
Processor E5450 
• HP BladeSystem c7000 Enclosures
Software
• HP Insight Control Environment for BladeSystem
• HP Onboard Administrator
• Microsoft System Center Virtual Machine Manager (VMM) 2008
Operating system
• Microsoft Windows Server 2008 Datacenter Edition with Hyper-V
Services from HP
• HP service and support
Customer solution at a glance