Recently I worked with a customer who was performing a Disaster Recovery test by restoring a huge inventory of tapes. After many hours, one of the tape read processes errored out. That meant they had to start over, trying to recover from the previous set of tapes. Again, after several hours, an error message announced the halt of the operation. Needless to say, these folks are re-thinking their recovery strategy!
If this customer had been facing an actual emergency, they might not have been able to get their IT operations running, or at best, they might have eventually become operational, but probably would have lost days worth of their valuable data and certainly would have missed their SLA targets.
In recent times, some companies are abandoning tape technology in favor of replicating their data to a remote site. The ability to have 100% of your data replicated and immediately ready to access at the DR site is quite attractive. Modern storage subsystems often have that capability built into them, however, the storage subsystems need to be nearly identical at both ends, as the replication technologies from different vendor’s equipment won’t work together. To solve that problem, you could insert devices like IBM’s SAN Volume Controllers, but that is also an additional expense. Sometimes, the solution becomes cost prohibitive because, in addition to the costs for the data links, there may an additional cost to activate the replication technology in the storage subsystem.
At Volta, we’ve been thinking about customers who are caught in these kinds of dilemmas and are working to find the most cost-effective solutions for them. In this case, the most obvious was host based replication, so we decided to look a little deeper into it.
AIX has the ability to replicate data using the Logical Volume Manager (LVM), (that we Administrators know and love), to a remote site, either synchronously or asynchronously. That ability is available to anyone who has two AIX systems, simply by adding a couple of additional file sets from the installation media, and of course configuring them to do so. The file sets are the Geographic Logical Volume Manager (GLVM) client and server pieces that extend the capability of LVM across geography using IP protocol. It doesn’t matter what kind of storage is being used, or if it is different at each end. That gives customers the flexibility to use different tiers or different vendors for their backend storage. Additionally, the functionality is fully supported by IBM’s AIX technical support, as well as, with PowerHA Enterprise Edition, should the customer eventually decide to fully automate failover to a remote site. And, best of all, there’s no additional cost to use the replication technology, because it’s supplied with and supported under your AIX license and Software Maintenance Agreement.
AIX Administrators are typically very well versed in the use of the LVM, but usually not as familiar with GLVM. To make it easy, we have developed easy to use menus and scripting that can implement geographic mirrors using GLVM. Our scripts also help you bring up the remote copy after a failure of the primary system, as well as, to reintegrate the primary system back into the mirror when you are ready, which resynchronizes the mirror back to the primary site.
To help you visualize this offering, we have developed a video. It runs 23 minutes and is a demonstration that we recorded using systems in our lab. The demo opens by showing how we set up the mirror and add data to the storage from the primary site. Then we halt the primary system and make the storage available on the remote site, where we add additional data. Lastly we restart the primary system, reintegrate the mirror copies, and make the storage available back at the primary site, where we can see all of the data stored to that point in time.
This video can be found at http://vimeo.com/238672883
Thank you very much for taking the time to watch this video. We hope you find it valuable and please feel free to contact us if you have any questions or would like more information. Read another internal article on this subject here.