IBM i Data Replication: Hardware vs. Software
There are a lot of options out there for IBM i data replication. We’ll help you weigh the options as your put together the pieces of your high availability plan. Some things you’ll want to consider as you start out include:
- Should I invest in hardware or software replication?
- Do I buy another server, buy a CBU, or rent a partition?
- Do I have my own secondary data center, use colocation, or have my solution hosted in the cloud?
- How often do I want to role swap?
- Do I hire a business continuity expert to help me?
You’ll find help for some of these considerations at the links provided. We’ll tackle the topic of hardware replication vs. software replication in this post. Software and hardware replication do have different capabilities that sometimes work well for one organization and not so well for others. It depends on your needs, the level of software, speed of communication, your application, and hardware configuration.
How Does Hardware Replication Work?
On IBM i, PowerHA is the product offering for hardware replication—although, technically, it’s a software and hardware combo solution. With hardware replication, the system or storage area network (SAN) replicates data by disk sector whenever it changes from one system to another. The PowerHA software piece helps to establish the cluster of IBM i systems participating in the replication process.
Pros/Cons of Hardware Replication
Ongoing maintenance is minimal. With PowerHA, all libraries and all directories in your independent auxiliary storage pool (IASP) are automatically replicated from one IASP in the cluster to another IASP in the system cluster. There is less day to day monitoring, but it’s still not advisable to set it and forget it.
You have the option of using internal disk or SAN. Internal disk does have a recommended threshold of around two terabytes of disk space. If you are over that, you should consider SAN to use PowerHA or you will need to look at software replication.
Very little (if any) scripting is required. The PowerHA software piece helps you set up clusters and move from server to server. Your downtime on the role swap is related to the time it takes to verify the IASP and to attach to the secondary system. Role swaps are generally a bit easier, and I’ve observed that PowerHA customers are role swapping more frequently.
Some objects on IBM i cannot go into the IASP. Like profiles, for example. In an IBM i server, you need to separate your business data from the operating system (i.e., SYSBASE). You do this by restoring libraries and directories into an IASP, essentially creating separate disk volumes.
You need to install your application in an IASP. This takes some planning, but IBM or its business partners can help you with this activity. PowerHA has a software component called Admin Domain that replicates user profiles and configuration objects from one SYSBASE to another. So, once your data and application are in an IASP, the information is automatically replicated.
How Does Software Replication Work?
Software replication uses a technology called journaling to record changes to your data. Most software replication solutions, like Robot HA, also use an extension of the technology called remote journaling. When data is changed on the source server, the local journal records the change and then sends the changes automatically to the target server.
You use your software replication solution to create the rules on the source system for which libraries and IFS directories should be journaled. Then, on the target system, the tool recognizes the journal changes and applies them to your database on the target.
Not all objects are journaled, so software replication solutions have another process for recognizing objects that are not physical files, data queues, data areas, or the IFS to ensure that changes to these objects or new objects are sent to the remote server.
Software replication finds out about new objects by reading QAUDJRN on the local system or by using Save Changed Object APIs. It then uses traditional save and restore to keep source and target servers consistent.
Pros/Cons of Software Replication
It only sends what has changed down the communication line. So, if you have a smaller budget or can’t otherwise establish a quality communication connection between your two servers, software replication is a great option.
It’s more flexible. With software replication solutions, like Robot HA, you can determine what you want to replicate and create the rules.
Your target server is fully active. With hardware replication, the target IASP where your data lives is not active. With software replication solutions, you can back up or even query the data on your target server while it is being replicated.
You must build scripts for doing the role swap. Many organizations struggle here because it involves shutting down your application and then starting it up on the target.
It requires more monitoring and periodic verifications. This ensures that you have all the data you want. With hardware replication, you add the library or directory to the IASP and it will be replicated immediately. With software replication, organizations are typically monitoring the replication process on a daily basis.
Whether you decide that hardware or software replication is the best fit for your business, we recommend using an automated monitoring solution to keep an eye out for any issues that could threaten your high availability environment. After all, PowerHA sends events to QSYSOPR, too.
While it’s not a requirement to have SAN technology with PowerHA, we recommend it. With this configuration, the SAN is what really does the replication instead the server. SAN also allows you to grow your environment without worrying about exceeding the two-terabyte internal disk limitation.
IBM i Data Replication Synergies
Hardware replication and software replication both work—and they can work well together. When problems arise with these solutions, it’s generally the result of how either was configured. The good news on IBM i is that you have options—valid options—that are being used by organizations like yours every minute of every day.
If you’re like 45% of IBM i shops, you’re still pussyfooting along without a high availability (HA) solution, putting your business at risk. Don’t let budget or resources be your excuse for putting off HA one more day.