A High Availability solution demands that all tiers of an organization "buy into" the benefits as well as the costs of such a solution. True or False?

HP's Scalable Computing Architecture ( SCA ) can be considered an implementation of the Massively Parallel Processing ( MPP ) architecture. True or False?

OnLine Addition and Replacement ( OLA /R) is a key technology in HP-UX 11i to minimize the impact of the failure of a PCI interface card. True or False?

Service Level Agreements in a High Availability cluster should include performance- related metrics in the event that multiple applications need to be run on a single node in the cluster. True or False?

A Serviceguard cluster must contain at least one heartbeat and one standby LAN. True or False?

True.

True. SCA is an MPP system because there is only one instance of HP-UX running across all nodes.

False. Utilizing multiple interface cards is the only way to minimize the impact of the failure of a single interface card. OLA /R avoids the necessity of shutting down a system in order to replace a single card.

True.

True. An attempt to configure a Serviceguard cluster will fail if either of these requirements is not met.

Planned outages can render a system unavailable for many hours, even days, e.g., to perform hardware updates. Service Level Agreements commonly do not include planned outages in the calculation of overall availability figures. Why?

I have 10 tape- drives that individually have an MTBF of 400,000 hours. How many tape drive failures would I expect in a one-year period?

HP-UX Trusted Systems offer several additional security features. Why would Trusted Systems be considered a good inclusion in a High Availability clz`uster where many users access individual systems and applications? Also explain why HP-UX Trusted Systems may be considered a poor choice in the same situation.

Should client PCs in a High Availability cluster support dynamic routing? If not, why not? Is the Router Discovery Protocol sufficient? Are there any features of RDP that we need to consider?

Looking at Figure 24-6, could this be considered a valid High Availability Serviceguard cluster?

Figure 24-6. A valid ServiceGuard cluster?

Availability figures are commonly calculated based on the time the system has been operating satisfactorily. This commonly does not include planned outages because we can foresee them and plan around them, i.e., move the required Service to another node in our cluster. In this way, the overall availability of the Service that a system provides has not been affected, even though the individual system has been down. Commonly, users are not concerned about individual systems but overall application/service availability.

The Annualized Failure Rate is calculated by taking:

Hence the AFR for my tape drive =

This equates to 2.2 failures per device per year. For 10 tape drives , we would expect 22 failures per year.

The additional security features that HP-UX Trusted Systems offers can be used to secure and monitor individual hosts . From a purely high-availability perspective, this is good insofar as we need to ensure that a single node is not compromised, and if so, we know about it as soon as possible. From a clustered viewpoint, HP-UX Trusted Systems offer a challenge insofar as ensuring that usernames and passwords are consistent across the entire cluster. Currently, HP-UX Trusted Systems cannot be integrated with an LDAP Directory Server, although it can be integrated into an NIS+ environment.

Yes. PC clients should support dynamic routing because a router in itself can be seen as a Single Point Of Failure in a high-availability network. By utilizing dynamic routing, the user PCs can have their routing tables updated with a new default route when necessary. The Router Discovery Protocol is sufficient in a small- to medium- sized network. However, we will need to ensure that the PC client operating system supports RDP and ICMP redirects.

This not a valid ServiceGuard cluster because a data/heartbeat LAN and a Standby LAN must be part of the same network and connected via a hub/switch. The use of a router to connect two segments of the same subnet is not a valid configuration.