Definition
Planned downtime is caused by the decision of an administrator or
service person. Examples of planned downtime are hardware or software
reconfiguration, or an off line upgrade or patch procedure due to system
limitations.
Customer problem
In today’s NonStop eBusiness computing environment, planned
downtime activities contribute to over 60% of system downtime, and software
has become the predominant opportunity for improvement!
HP's answer
HP is cognizant of this trend and is solving this dilemma in real time. In August 2003, DH Brown published its review of current UNIX cluster offerings, "Superior
TruCluster Technology Combined with AdvFS Yields High Availability and Lower Infrastructure Cost of Ownership". DH Brown concluded, "HP
TruCluster Server software Version 5.1B exceeds the high-availability expectations of clusters, offering a Cluster File System (CFS) with
a shared root directory and other advanced features that ensure the easiest cluster management." Further, DH Brown notes the unique cost-savings
characteristic of TruCluster Server software: “… once the cluster is established, it manages with a “fixed management cost” roughly
equivalent to a single system. The more nodes, the greater the advantage to the IT infrastructure.”
Here's how Tru64 UNIX keeps planned downtime to a minimum, making
your AlphaServer system the most reliable UNIX platform on the market:
Managing multiple systems as one:
Using TruCluster Server software to manage your multiple systems, you
can perform routine maintenance tasks without any downtime. Just take
down one node while the rest of the cluster is running and perform
the operation once for the entire cluster. You're saving time with
minimum effort and keeping the availability at the service level you
agreed to.
Rolling upgrades: Install patches
on a cluster without disrupting operations. Individual members of the
cluster which are not being patched initially, are set up so that changes
to the shared cluster file system do not impact the overall availability
of the cluster. The cluster member being patched can be removed from
the cluster, patch(es) applied, member re-booted, and tested with the
end user's application to verify that the patch(es) can be applied
to the other members of the cluster. All this is being done while your
applications continue to run on the remaining members of the cluster.
Online tuning: Tune system parameters
as needed without rebooting the system and optimize your applications
performance. It's fast, it's easy, and it keeps the level of availability
as high as you need it while boosting performance.
Multi-user mode dupatch:
Install patches on a system while applications are still running and
available to users (in multi-user mode). Patches can be installed remotely
thereby guaranteeing that the network capabilities of the system cannot
be disabled during the patch operation. The integrity of the running
system is not compromised while installing patches. More specifically,
patches will not overwrite files currently in use. Applications running
on the target system operate while patches are being installed. Patches
can be backed out and/or removed. dupatch is capable of running in
a single system configuration as well as in clusters.
On-line Insertion and on-line removal:
Add capacity to the system, remove components for upgrades, alter system
configurations, and replace components while the system is running.
Components may be storage devices, power supplies, CPUs, and so on.
|
