2.1 Gdps -
False. PowerHA manages application-level restarts and logical replication. 2.1 GDPS manages storage block-level consistency. In a mature environment, you run both: GDPS for site failover, PowerHA for LPAR failure within a site.
It introduced native, automated awareness of IBM i’s commitment control. By leveraging the IBM i FlashCopy integration, the 2.1 stack allows GDPS to understand when a database is in a consistent state before initiating a site switch. This reduces the risk of "split-brain" scenarios and logical corruption. Core Components of the 2.1 GDPS Stack When architects discuss a "2.1 GDPS solution," they are referring to five distinct technological layers working in concert: 1. The DS8000 Storage Family You cannot achieve 2.1 GDPS without IBM DS8000 series storage. These arrays provide the underlying Global Mirror (asynchronous) and Metro Mirror (synchronous) links. The "2.1" spec requires DS8000 LIC (Licensed Internal Code) level 7.8 or higher to support the IBM i-specific command sets. 2. The HyperSwap Capability For IBM i partitions, the hallmark of 2.1 GDPS is HyperSwap . Historically, if a primary storage volume failed, the system required an IPL (Initial Program Load). With HyperSwap, the IBM i operating system (via the DS8000 Multipath Driver) instantly redirects I/O to the secondary copy. In a 2.1 environment, HyperSwap is triggered automatically when GDPS detects a storage path failure, without disrupting active jobs. 3. IBM i 7.2 (or 7.3/7.4/7.5) The "2.1" moniker often confuses buyers running IBM i 7.4. The name persists for legacy compatibility, but the feature set requires IBM i 7.2 TR3 or later. Specifically, the SST (System Service Tools) commands WRKDSKSTS and the disaster recovery APIs were enhanced in 7.2 to report back to GDPS about journal latency. 4. GDPS Base Code (V2.1 SP1) The actual software component—GDPS Base—runs on a dedicated z/VM or Linux virtual appliance. Version 2.1 introduced the "Session Manager" for IBM i, which allows a single GDPS console to monitor up to 250 IBM i partitions simultaneously. 5. Metro/Global Mirror Combination The real genius of 2.1 is three-site replication . You can configure Metro Mirror (synchronous, <10km) between Data Center A and B for zero data loss, while simultaneously sending Global Mirror (asynchronous) to Data Center C (over 300km away). GDPS 2.1 automates the promotion of the Global Mirror copy to a Metro Mirror role during a failover. Why 2.1 GDPS Matters for IBM i Administrators If you manage an IBM i environment running Domino , DB2 for i , or JD Edwards EnterpriseOne , you have likely faced the "journal problem." Traditional clustering (like IBM i Cluster Management or PowerHA) requires the secondary node to replay journal entries. If the primary site fails mid-transaction, you might lose seconds of data. 2.1 gdps
For those unable to meet the distance or budget requirements, consider GDPS Global Mirror alone (asynchronous, 2.0 feature set) or modern PowerHA with NVMe replication. However, for the most demanding financial, healthcare, and retail IBM i workloads, remains the gold standard against which all high availability is measured. This article is based on IBM Redbooks, DS8000 documentation, and field experience as of 2026. Always consult your IBM Business Partner for current licensing and compatibility with your specific IBM i release. In a mature environment, you run both: GDPS
But what exactly is "2.1 GDPS"? It is not simply a software version number. Instead, it refers to the integration of (or later) with Global Mirror (formerly XRC) and Metro Mirror (formerly PPRC) in a GDPS environment, refined by the operational standards defined in the IBM Redpaper REDP-5501, "GDPS V2.1" . For the IBM i professional, understanding 2.1 GDPS means understanding how to fuse logical replication with physical storage clustering. The Evolution from 2.0 to 2.1 To appreciate the 2.1 GDPS framework, one must look back at its predecessor. GDPS 2.0 offered basic hypervisor-level orchestration. It could fail over a logical partition (LPAR) from one data center to another, but it relied heavily on manual intervention for IBM i-specific objects (such as journal receivers and user spaces ). This reduces the risk of "split-brain" scenarios and
In the world of enterprise IT, downtime is measured in the cost per second. For organizations running mission-critical workloads on IBM i (formerly AS/400, iSeries, System i), the difference between a minor interruption and a catastrophic business failure often comes down to one acronym: GDPS (Geographically Dispersed Parallel Sysplex). However, with the introduction of tighter coupling between operating system releases and storage replication, the threshold known as 2.1 GDPS has become the new gold standard for near-zero Recovery Time Objectives (RTOs) and Recovery Point Objectives (RPOs).
The bank’s IBM i administrator noted: *"Before 2.1, we spent three hours verifying JRN data after a drill. Now, GDPS presents the secondary as if nothing happened. We IPL and run CHKOBJITG —that’s it." As of 2025, IBM has released GDPS 4.3 and 4.4, but the industry still refers to "2.1 GDPS" as the baseline for IBM i resilience because that version introduced native journal awareness and HyperSwap. Future versions add AI-driven predictive failover and NVMe over Fabrics (NVMe-oF) support, but the architectural principles—synchronous replication, storage-level consistency, and automated orchestration—remain rooted in the 2.1 specification. Conclusion: Is 2.1 GDPS Right for You? If your organization operates an IBM i environment where every transaction matters and distance allows synchronous replication (under 25km), then pursuing a 2.1 GDPS architecture is the most robust business continuity strategy available. It is not a replacement for log-based replication tools, but it is the ultimate safety net for storage and site failure.
False. 2.1 GDPS requires DS8880 or DS8900 series. It will not work with IBM FlashSystem (unless using SVC-based Metro Mirror, which is a different, non-2.1 architecture) or third-party arrays like Dell EMC or NetApp.