PowerHA Updates Focus on Integration and Automation
May 13, 2020 Alex Woodie
Despite the introduction of Db2 Mirror last year, PowerHA remains a critical component ensuring the high availability and disaster recoverability for thousands of IBM i shops around the world. Recent updates, including the delivery of PowerHA version 7.4 in the second half of 2019 and a minor update delivered in March 2020, have focused on bolstering the product’s integration with Db2 Mirror, as well as HyperSwap in the DS8000. Automation, monitoring, and usability have also been improved.
Brian Nordland, the PowerHA technical product manager with HelpSystems (which develops the product on behalf of IBM) recently updated the IBM i community on the state of PowerHA. That includes the two recent updates, as well as a glimpse into new features due out next month. His presentation is one of many available through COMMON and its recent inSIGHT virtual conference, which is worth checking out.
Db2 Mirror, which IBM launched last year with IBM i 7.4, enables two Db2 for i databases to be clustered together over a high speed RoCE link for the purpose of delivering continuous, real-time availability of IBM i data. But due to the RoCE requirement, Db2 Mirror limited to working in a campus-wide environment, which means that it’s not useful for disaster recovery. Plus, Db2 Mirror doesn’t replicate data in the IFS, which is a critical file repository in modern IBM i applications.
The Db2 Mirror integration that IBM delivered with PowerHA version 7.4 in December helps to solidify Db2 Mirror environments, according to Nordland, who worked at IBM Rochester until Big Blue tapped HelpSystems to take over the PowerHA product in 2017.
“Db2 Mirror provides synchronous read and write access to IFS data from two systems simultaneously through a high-speed network adapter called RoCE,” Nordland says. “All of these requests under the covers are forwarded to a primary node that owns an iASP with all of your IFS data. PowerHA provides transparent switching of this iASP during outages and failures.”
Plus, in PowerHA 7.4, HelpSystems greatly improved the speed of iASP vary-ons. That makes switches faster, both for customers just using PowerHA and for those using PowerHA in conjunction with Db2 Mirror.
“PowerHA, when combined with Db2 Mirror, provides near continuous availability of IFS data and also adds the capability to support real-time disaster recovery solutions in combination with Db2 Mirror,” he says. “Once you combine PowerHA and Db2 Mirror for disaster recovery, you not only get protect against these local outages within in the data center, but also a protection against an outrage of the entire data center, as well as regional outage, like a natural disaster.”
Last year’s delivery of PowerHA 7.4 also delivered integration with IBM Copy Services Manager for DS8000. According to Nordland, this provides PowerHA with expanded protection capabilities by hooking HyperSwap and Global Mirror together.
Hyperswap debuted back in 2014 as a high-end data protection mechanism to ensure that IBM i data loaded in a redundant DS8000 arrays are identical and can be swapped back and forth with almost no perceptible lag. Like Db2 Mirror, it focuses on data availability, and it has limited range. But in Hyperswap’s case, the limit is 300 kilometer, which is a bit further than Db2 Mirror can go.
Hyperswap has supported LUN-level switching in conjunction with PowerHA Express Edition since 2016. But it was still limited to providing a local data protection regime. With the recent integration between Global Mirror and Hyperswap delivered via a PowerHA PTF in December 2019, IBM i shops can be protected by regional outages that impact an area greater than 300 kilometers wide.
The integration between PowerHA and IBM Copy Services Manager gives system architects another tool to build a compelling data and application protection regime, Nordland says. “We try to make managing this as similar as possible to our existing technology,” he says. “Global Mirror is managed just as if you have Global Mirror without Hyperswap. Hyperswap is managed just as if you had Hyperswap without Global Mirror. PowerHA handles everything as they relate to each other under the covers automatically.”
Next month, IBM will be shipping an update to PowerHA that furthers the integration story. According to Nordland, users of PowerHA 7.2 and higher will get support for IBM i SQL services in IBM i, which will further bolster the capability of administrators to manage cluster information, administrative domain information, and cluster resource group (CRG) information, Nordland says. “This will ease the integration of PowerHA with other technologies and solutions,” he says.
On the automation front, HelpSystems has worked to bolster PowerHA in several ways. For starters, in PowerHA 7.4, it has worked to automate the setup of the Administrative Domain, which is how the product replicates non data elements that are part of the operating system, such as user profiles and configuration files.
The new ADDCADMRE command simplifies the process of finding all of the IBM i resources that an administrator may want to include in the Administrative Domain. The command has some intelligence to make sure that the administrators doesn’t miss important items, like groups of user profiles.
Once the domain is set up, PowerHA can keep it current by automatically adding resources (such as user profile) to the domain as they’re created on the production box. And when a resource such as a user profile is removed from the domain, PowerHA will automatically propagate that deletion across all of the nodes, Nordland says. This is enabled through the new WRKHAPCY command.
“We understand that not everyone wants everything to automatically be added to the Administrative Domain and synchronized across all nodes,” he says. “So in order to give you choice over the behavior of things in PowerHA, we created something called PowerHA policies. These give you customization over the behavior of your PowerHA environment.”
This not only saves the administrator time, but it reduces the chance of missing a critical user profile that needs to be replicated on the DR box. And it works the other way too – an administrator doesn’t have to worry about an unused user profile creating a security issue on the DR box long after it’s been deleted on the production machine. “PowerHA does all of this for you while reducing the amount of time you actually have to spend with the product,” the technical product manager says.
Reporting goes hand in hand with automation by communicating what the system is doing. To that end, PowerHA 7.4 has been bolstered with new reports that ensure the production and DR boxes are being maintained in compliance with security requirements.
“When run on a single node, it will analyze all of the profiles across all of the nodes in the Administrative Domain, including profiles not monitored by the Admin Domain, to find any that are inactive and disable them,” Nordland says.
Administrators are able to specify how many days constitutes an unused profile, he says. “This functionality is designed to replace the analyze profile activity functionality provided by the operating system so that it works across the HA environment,” he says.
PowerHA admins will also appreciate a new field that displays the current recovery point objective (RPO) for the setup, in minutes and seconds (the RPO is the amount of data that you are set to lose if the production system were to suddenly stop and you had to rely totally on the DR box).
“Now I can easily tell exactly what my RPO is and where my data is and how protected I am,” Nordland says. The RPO feature works only when using IBM Spectrum Virtualize (SVC)-based storage and IBM Copy Services Manager (CSM)-based storage, according to Nordland.
On the usability front, there are several enhancements.
As PowerHA systems get bigger, the number of items stored in the Administrative Domain also gets bigger. To address this crunch, IBM increased the maximum size of the Administrative Domain by 340 percent with the 7.4 update. It can now hold up to 200,000 items.
Previously, admins could specify if an iASP should be varied on or off following a failover. With 7.4, there’s a new command that instructs the system to return the iASP state to the same as it was before the failover.
PowerHA can be configured to automatically failover in some instances. With 7.4, the controls get more granular, as administrators can specify situations where they do not want to automatically failover. It can even be configured on a CRG basis, Nordland says. “We give you lots of control over this,” he says. “You can say ‘Allow failovers win the same site. Allow LUN switches to happen. But disable failover that go across sites, over to my DR site, because there I have a lot more work to do for a failover.'”
HelpSystems, with input from IBM, has delivered a lot of PowerHA functionality in 2019 and 2020. “But we’re just getting started,” Nordland said. All of the new features are the result of customer feedback, and Nordland encourages PowerHA customers to continue submitting more feature ideas and requests for enhancements – the more detailed, the better.