Question: How do I choose the most appropriate HACMP/XD replication technology?
Answer: This is predominantly determined by the distance between your production and backup sites.
If the sites can be covered by a common storage area network—say, adjacent buildings in the same campus—then the LVM split site mirroring function described in the HACMP Administration Guide is most appropriate, providing the best performance at no additional expense.
If the sites are within the range of PPRC (roughly, 100km) and compatible ESS/DS/SVC storage systems are used, then one of the HACMP/XD: PPRC technologies is appropriate.
Otherwise, consider HACMP/XD: GLVM.
Question: What's involved in moving a Service IP Address between sites?
Answer:
HACMP/XD 5.4 provides the ability to move a Service IP Address between sites, by performing IP address takeover on the links between sites. On prior releases, this operation would have to be provided through customization.
On any release, it is a customer responsibility to ensure that the operation of moving an IP address between sites is actually meaningful. That is, once the IP address is moved, will the customer's networking infrastructure route packets to it. This may happen automatically; or it may require updates to the configuration of the customer's routers. No general guidance can be given, since there are a large number of possible networking infrastructures, each with their own way of handling this.
Question: Which storage subsystems are supported with HACMP/XD: GLVM?
Answer:
In general, any storage subsystem supported by HACMP is also supported with HACMP/XD: GLVM.
Additionally, since HACMP/XD: GLVM is not strongly dependent on the storage hardware type, any storage subsystem that works with AIX 5L is very likely to work with HACMP/XD: GLVM. If the disaster recovery cluster has only a single node at each site, so that local fallover is not required, then any storage subsystem supported by AIX 5L LVM will work with HACMP/XD: GLVM.
Question: Does HACMP/XD require a non-IP link between sites?
Answer:
A non-IP link is not required, in the sense that the cluster will function correctly without it. Non-IP links are recommended in an HACMP cluster for two reasons:
Having more than one link prevents a single link failure from being interpreted as a node failure
Having a non-IP link prevents temporary congestion on the IP link(s) from being interpreted as a node failure
These reasons are still germane in an HACMP/XD cluster. A customer using only a single, IP-based link between sites should understand their exposure.
A non-IP link between sites can be created using a leased line and modems.
Question: Can other vendor's storage replication, such as EMC's SRDF, be used with HACMP/XD?
Answer: Currently, HACMP/XD has no provision for storage replication other than that provided by HACMP/XD: PPRC and GLVM.
Question: Does HACMP/XD support multiple storage units per site?
Answer: This is supported - and is transparent to - HACMP/XD: GLVM. It is not currently supported by HACMP/XD: PPRC, but is planned to be supported in a future Service Pack.
Question: How do I properly size my networks for IP-based mirroring?
Answer:
Tools such as gmdsizing and vmstat must be used with benchmarking to determine the aggregate write rate for the disks to be mirrored. The networks should provide that bandwidth to avoid significant impact to performance. Refer to the HACMP Downloads page for the gmdsizing tool.
Question: Why does IBM offer two different IP-based mirroring technologies?
Answer: Both HACMP/XD: HAGEO and HACMP/XD: GLVM offer IP-based mirroring and high availability function. HACMP/XD: HAGEO represents technology which has been available for a number of years. HACMP/XD: GLVM is newer technology which utilizes the proven function of AIX LVM.
Question: How do I implement IP-based data mirroring without high availability?
Answer:
AIX 5L Geographic Logical Volume Manager (GLVM) is a function available in AIX 5L V5.3 which supports the mirroring of a customer's data to a remote site. This remote mirroring is provided in AIX 5L at no extra cost and without the high availability functions provided by HACMP/XD: GLVM. This standalone GLVM function in AIX 5L can be used by customers who want only to keep a remote copy of their data without the automated failure detection, fallover and restoration functions of HACMP. A standalone implementation of AIX 5L GLVM can be migrated to an HACMP/XD for GLVM implementation when the customer is ready to move to automated high availability. For more information see:
http://www.ibm.com/systems/p/os/aix/whitepapers/aix_glvm.html
|
