home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Power Tools 1993 November - Disc 1
/
Power Tools Plus (Disc 1 of 2)(November 1993)(HP).iso
/
3khw
/
50916983
/
50916983.txt
< prev
next >
Wrap
Text File
|
1993-03-22
|
11KB
|
206 lines
White Paper
SharePlex/iX
March 1993
Introduction
In organizations that rely on computer processing to support
their business, the need for greater system uptime and data
availability is steadily increasing. The HP 3000 is
Hewlett-Packard's premier computer platform for worry-free,
business-critical computing, and delivers the most effective
solution for environments that require high system and data
availability .
One of the features that makes the HP 3000 the preferred
platform by commercial customers is the ability to configure
systems in a loosely coupled or clustered manner. This is
achieved through a product offering called HP SharePlex/iX.
This paper will explain what SharePlex/iX is and how it
benefits HP 3000 customers.
What is a cluster?
SharePlex/iX is a clustering solution for the HP 3000. But
what is a cluster? In the computer industry, there are a
range of solutions that try to pass as a "cluster" that vary
dramatically in functionality. The least functional would
be a configuration of 2 to 3 systems with switchover
capability, for warm standby in the event of a system
failure. The other end of the spectrum, where SharePlex/iX
resides, is a multi-node computer system which has:
o a single-system view from the perspective of its users,
programmers, operators and administrators
o provisions for enhanced availability
o cluster-wide operations and management features
o shared, cluster-wide facilities for print queues, batch
queues, file systems, and peripherals
o a graceful incremental growth capability
o a flexible configuration through interconnect and topology
options
Benefits of clusters
The benefits of clusters vary from providing high data and
application availability to protecting customer investment
in existing systems. The following details some of those
benefits and explains how SharePlex/iX delivers them.
Figure 1. SharePlex/iX: Local and wide area cluster
Higher application and system availability
By providing features to protect data in the event of a
system or disk failure, clusters have traditionally provided
the benefit of increasing application and system
availability.
However this doesn't go far enough. The need still remains
to be able to recover from unexpected disasters that make
data unavailable. These can include a local datacenter
disaster, such as an outbreak of vandalism, a regional
disaster, such as a power outage in a section of New York,
or a widespread disaster, such as an earthquake that impacts
a very large geographic area. Businesses are increasingly
aware of the need for a Disaster Recovery Plan (DRP) that
enables them to resume access to their data and other
computing resources. Sales losses from downtime can run as
high as $70,000 a minute and a long outage can cripple or
even destroy a company's ability to do business after the
event. Though often overlooked, DRPs are especially
critical to companies that are dependent on the availability
of their systems and data in order to do business.
Figure 2.
While some people rely on Fault Tolerant systems for this
protection on a single site, SharePlex/iX configurations
protect data even in the event of a natural disaster that
completely destroys an entire node. SharePlex/iX
incorporates data and application shadowing features than
replicate the data in another distant location. In the
event of a disaster, the data is preserved and is still
accessible by other parts of the company that need it. And
SharePlex/iX can be configured without geographic
limitations, anywhere in the world.
In addition to the obvious high availability benefits of
disaster recovery, SharePlex/iX has more subtle ways to
increase data and application availability. Customers can
switch application processing from one system to another
during particularly heavy periods of processing. For
example, in order to improve application performance and
response time, a company might choose to do its year-end
number crunching on a larger, more powerful system than its
finance department usually uses.
This level of flexibility enables customers to tailor their
information systems to their business needs, rather than
making the business adhere to the computer system
limitations.
Figure 3. SharePlex/iX: Corporate wide cluster
Another example involves the ability to send print jobs to
any printer on the network. Aside from the obvious benefit
of sharing system printers, this feature is capable of
keeping a copy of the print job on a local system until it
receives confirmation that the job has successfully
completed on the remote system. If the remote printer
fails, the job can be automatically resent, or routed to
another printer.
SharePlex/iX is completely compatible with other HP 3000
high availability features that provide automatic system
restart in the event of a system software failure, system
switchover in the event of an SPU failure, and disk
mirroring. All of these features increase application and
data availability even further for SharePlex/iX cluster
users.
Investment protection and growth-increment flexibility
SharePlex/iX is the only highly functional cluster solution
offered by a major computer vendor that is supportable
across both CISC and RISC platforms. This enables HP 3000
customers who have a mix of MPE V and MPE/iX systems to take
advantage of clustering benefits across their entire
enterprise. Customers need not discard older HP 3000 models
in order to get the functionality of a cluster (this is an
issue for DEC VAX Clusters, which are only fully supported
on the older CISC platforms).
SharePlex/iX protects customer investment by enabling an
overloaded CPU to move an application or an application
module to another system, improving performance on the first
system. No applications changes are necessary. So if a
system in Houston were near its processing limits, an under-
utilized system in Chicago could offload it. The Houston
users would never know the difference. And the customer
would be getting a better return from his investment in the
Chicago machine.
Clustering capability provides processor growth flexibility
in the same manner: customers only needing a small
incremental amount of processing power for a particular
application have more choices. They can add a small system
to the cluster, or perhaps just take advantage of some
excess processing power on another system in the cluster.
In the above examples, SharePlex/iX is most appropriate when
the configuration consists of multiple applications or
databases that need to occasionally share data. If the
configuration consists of one large application or group of
applications that heavily access the same database,
upgrading the processor, rather than spreading the workload,
would be more efficient.
SharePlex/iX enables workload balancing by integrating
several key technologies, most notably a product called
NetBase and a product called HP OpenView System Manager.
NetBase consists of a set of modules that provide data
shadowing, file access, spooling, and process management
features. These features facilitate the kinds of workload
balancing in the examples above. Since applications need
not be rewritten or adjusted to take advantage of NetBase,
it is an easy and cost effective technology to implement.
Figure 4. NETBASE: Primary SharePlex/iX enabling technology
Operational ease-of-use
SharePlex/iX clusters include software that makes the
cluster look like one system to the end user and
applications, so no retraining or application adjustments
are necessary. SharePlex/iX configurations are managed with
HP OpenView System Manager, which is easy to use even for
operators with very little training. Many functions that
would normally require human intervention are automated, and
operators are alerted to events that require their attention
via simple icons that light up and change colors. Even
clusters with nodes that are dispersed worldwide can be
easily and centrally managed this way.
Figure 5. HP OpenView System Manager for SharePlex/iX
NetBase software has been enhanced to integrate with HP
OpenView System Manager so the centralized console is
alerted if any event occurs involving any of the NetBase
modules. This greatly simplifies cluster management.
Figure 6. Monitoring distributed HP 3000s
Comparing SharePlex/iX to other vendors' clustering
solutions
The Aberdeen Group, in their report, Clustering: An
Alternative Growth and Operations Path, compares the HP 3000
SharePlex/iX solution to other vendors' clustering
solutions. Digital Equipment Company did a lot to make
clusters a viable computing solution in the 1980's, and it
scored highly in the Aberdeen report from the perspective of
the old CISC-based VMS platform. DEC's new RISC platform
does not support the clustering capabilities of the old VAX
VMS platform. Thus the benefits of having a fully
functioning cluster with RISC-based hardware are not
available to DEC customers. In addition, VAX Clusters
require expensive specialized networking to overcome
geographic limitations. SharePlex/iX does not have any
geographic limitations, which is an important feature for
true disaster recovery, and it utilizes less expensive
standard networking to keep costs down.
The HP 3000 was rated higher than all the other cluster
solutions evaluated, including IBM's Sysplex (MVS), the
RS/6000 solution, and several other UNIX* solutions. The HP
3000 is the clear leader in cluster solutions on a RISC
platform, without geographic limitations. SharePlex/iX
offers flexible configurations that increase the
availability of customer's data and applications, which is
just one reason people choose the HP 3000 for worry-free,
business-critical computing.
For more information:
o SharePlex/iX Clustering for the HP 3000 5091-7037E
o Clustering: An Alternative Growth and 5091-7103E
Operations Path (Aberdeen Group reprint)
o HP 3000 High Availability White Paper 5091-6982E
Technical information in this document is subject to change without
notice.
(c) Copyright
Hewlett-Packard Company 1993
All Rights Reserved. Reproduction, adaptation, or translation without
prior written permission is prohibited except as allowed under the
copyright laws.
Printed in USA GD0393
5091-6983E