IBM PowerHA SystemMirror for i was first introduced in 6.1 and has received continual updates and new functionality since then. As PowerHA has grown, new features such as the recent HyperSwap technology continue to bring even more high availability and disaster recovery options to meet the ever-changing and diverse needs of businesses around the world.
PowerHA now offers solutions that fit customers both at the largest end of the spectrum with the largest Power Systems and latest storage area network (SAN) technologies as well as those with just one or two cores on internal storage.
I recently joined Steve Finnes and Tom Huntington on a webinar discussing the benefits of PowerHA in depth. I encourage you to listen to the reply and learn how the right solution can strengthen your recovery plan. For now, here’s a bit of an introduction to the technology so that you don’t arrive emptied-handed.
The Foundations—Hardware Replication
At the core of all PowerHA technologies lies the concept called shared storage clustering. Shared storage clustering allows you to exploit hardware replication, which takes advantage of the fact that the system itself knows how to read and write data to and from its disk units. So, if the contents on the disk units in the cluster are replicated in real time to other systems in an identical way, the other systems will be able to read and manipulate the data in the same way when they assume ownership of the production applications.
All this data sharing and replication happens a low level. It’s an extension of a standard operating system paradigm extended to multiple systems. As a result, the solution becomes very easy to manage—the data is either replicating or it’s not. The set of disk units PowerHA shares and/or replicates is called an independent auxiliary storage pool (IASP).
IASPs have been around for years, but there’s always been a bit of a stigma around them. Steve Finnes helps to clear up some of the confusion around IASPs in this interview.
Business Availability Objectives
The key to any solution around high availability or disaster recovery is to start the discussion with business-level objectives. Two terms you’ll often hear for describing these objectives are recovery point objective (RPO) and recovery time objective (RTO).
- Recovery point objective (RPO): When there is an outage, how much data can be lost? Is it acceptable to go back to a week-old tape backup? Or is it required to have a solution that provides more recent data of only a minutes or hours?
- Recovery time objective (RTO): When there is an outage, what is an acceptable amount of time for the systems to be down while still meeting service-level agreements?
The example in Figure 1 shows a graphical representation of RPO and RTO. By starting with these business requirements, you can then move toward finding the solution that meets your needs.
Technologies for Every Outage Type
PowerHA is really a family of solutions that can meet the needs of various types of hardware, RTO and RPO targets. Below are some of the key technologies and recommendations for use:
Geographic Mirroring
Geographic mirroring is replication performed by IBM i. It can be used with any storage (internal or external) and can be performed synchronous or asynchronous.
Synchronous geographic mirroring keeps both copies of the disks identical for a very low RPO, but has distance limitations.
Asynchronous geographic mirroring has a higher RPO and RTO, but allows for greater geographic separation between systems. The asynchronous option is typically used with internal storage and less than a few terabytes of data.
LUN Level Switching
LUN Level Switching moves a single set of disk units from one IBM i to another to protect against an IBM i outage. It provides a very low RPO and RTO for IBM i outages, and is often combined with GlobalMirror to give protection against a storage level outage. LUN Level Switching happens within the same data center for local high availability.
MetroMirror
MetroMirror is synchronous replication performed by the SAN storage. It gives two copies of the data and provides a very low RPO, but is limited by distance. Typical setups only go across a metro area before performance becomes a concern.
GlobalMirror
GlobalMirror is asynchronous replication performed by the SAN storage. It allows for greater distances than MetroMirror, but with a higher RPO and RTO. Customers use this type of PowerHA to replicate data across oceans.
HyperSwap
HyperSwap protects against a storage device failure across two or more SANs with minimal impact to running applications. It is typically combined with LUN Level Switching to offer protection against both storage device failures and IBM i outages.
FlashCopy
FlashCopy is a very quick, point-in-time snapshot of the data. While not official part of the PowerHA family, it is used to significantly reduce the backup window. FlashCopy also requires SAN storage.
Figure 2 shows the various types of outages that the PowerHA technologies can protect against.
Combining Technologies
Even though the individual members of the PowerHA family of technologies provide various levels of protection, many of these technologies are designed so that they can be easily integrated together to form an even more robust high availability solution.
Figure 3 (image not available) shows an example of combining three of these technologies: LUN Level Switching, GlobalMirror and FlashCopy.
- LUN Level Switching between IBM i A and IBM i B provides a very low RPO and RTO for outages of a single IBM i system at the local site. It’s useful for PTFs, OS upgrades, firmware upgrades, etc.
- GlobalMirror, while it has a higher RPO and RTO than LUN Level Switching, gives protection against outages that LUN Level Switching cannot help with, such as a fire at the local data center.
- FlashCopy greatly reduces the backup window, and can do so without needing to pause or stop the replication for the duration of the backup.
Solutions that Grow with Your Business
Despite the variety of flavors that make up the PowerHA family, the way you manage the various technologies remains largely the same across the board. This consistent with the core concept of the IBM i platform itself: the technologies that power high availability will change, but the way you manage those technologies should not have to change.
This has allowed businesses to start with one technology and grow their high availability solution over time to incorporate new technologies with a minimal administrative learning curve.
Again, join Steve, Tom, and me on the webinar to learn how PowerHA can simplify HA/DR management—including role swaps and day-to-day monitoring. Knowing your options with these strategic solutions is crucial.
This blog was written by Brian Nordland. Brian is the technical lead for PowerHA SystemMirror for i.
This blog post was originally published on IBMSystemsMag.com and is reproduced here by permission of IBM Systems Media.