IBM Moves its GDPS Solution for Resilience and Disaster Recovery-in-Seconds to IBM Systems

GDPS is a clustered system design that combines IBM Z systems and DS8K storage to deliver high-availability, resilient, disaster recovery solutions.  Today, GDPS is purchased by enterprises that are:

      • Seeking a continuous availability solution for IT processes (a high-availability, resilient, disaster recovery solution);
      • Seeking to address regulatory requirements;
      • Looking for protection against financial loss (lost revenue opportunities, compliance penalties/fines, legal actions, reputation loss); and,
      • Looking for security-related reasons to recover more quickly from security breaches and avoid loss-of-reputation/brand damage from such breaches. For more on this topic, see this Clabby Analytics blog on IBM’s Cyber Vault for a description of an IBM detection/recovery solution designed to help enterprises recover more quickly from outages due to corrupted logical data – the data that includes entities (tables), attributes (columns/fields) and relationships (keys) – the logic that drives a database. Cyber Vault can also help IBM Z customers identify cyberattacks aimed at logical data – and help enterprises respond/recover more quickly from data breaches.

GDPS is now available from IBM Z in the IBM Systems Group

As part of a strategic rethink, IBM has decided that it makes more sense to align GDPS with its IBM Z organization, where it will be under the control of the people who actually build IBM Z environments.  As a result of this strategic realignment, GDPS staff have been reassigned from IBM’s managed services organization to IBM Z. – meaning that future GDPS architecture, design, development and services will be led by IBM Z.

For IBM Z clients, this realignment means that IBM can more closely align GDPS with future IBM Z development – giving the systems group control of GDPS plans, funding, development and execution from now on.  Accordingly, GDPS development will track in lock-step with IBM Z plans.

What does this mean to a typical information technology (IT) organization? 

The big benefit of moving GDPS into IBM Z is that IBM Z developers will be able to develop new features for GDPS and will also be able to expand functionality to meet broader resiliency/disaster recovery needs in the resilience marketplace.  Furthermore, it is reasonable to expect that IBM Z will add new features to ease installation, simplify management and blend GDPS with IBM security offerings to improve security and recovery speed.

How to acquire GDPS

Through an IBM Systems Lab Services engagement, the IBM Z group can now provide defined GDPS configurations through various well-defined pricing models.

As for deployment, GDPS will be deployed by IBM system engineers who help design GDPS offerings, then ensure that those clusters are deployed correctly – and that GDPS customers receive the training and educational services needed to make GDPS customers self-sufficient.

What I think IBM will now do with GDPS

With GDPS available as a solution within IBM’s Systems Group, IBM will likely:

  1. Continue to offer GDPS as a means to create Parallel Sysplex replicated environments with varying levels of recovery times (you pick the elapsed time [days, hours, minutes, seconds] required for IT services to be resumed (known as “recovery time objective – RTO“) – in other words, how long you’re willing to wait until systems are up-and-running. And you pick the amount of time your enterprise can wait for data to be recovered and/or restored [this is known as a “recovery point objective – RPO”]),  and,
  2. Find ways to package GDPS offerings to aid in resiliency, failover and disaster recovery for straight-up (non-clustered) IBM Z environments. This practice is known as “cascading,” the rolling advanced functionality to other systems environments).  If this happens, GDPS functionality could wend its way to other IBM platforms (such as IBM Power Systems) or even to non-IBM platforms.  As an example, consider GDPS HyperSwap failover solutions that allow HyperSwap failover today on non-Sysplex-clustered systems.  (For an example of how GDPS HyperSwap, based on synchronous Metro Mirror, works, see this blog).

GDPS buying trends and buying patterns

To help IBM Z customers address resiliency/disaster recovery needs, Parallel Sysplex technology helps overcome single-points-of-failure within a server environment, Logical Partitions (LPARs), and related subsystems (to do this, IBM runs multiple, failure-isolated copies of programs and data across systems and storage devices).  A primary focus of this Parallel Sysplex design is to ensure application availability.  Over the past twenty-plus years of development, IBM has structured Parallel Sysplex as a single system image to streamline manageability (though it runs across a cluster of several systems); IBM has added Dynamic Session Balancing to up utilization rates, and IBM has made it simple and straightforward to add additional capacity (scalable capacity).  GDPS simply extends these features geographically to remote, mirrored systems and storage (such that if a disaster should occur at one site, computing can failover to a remote site with minimal interruption in service).

For most enterprises, GDPS is purchased as a replication environment.  But the offering provides far more than replication services.  Actionable scripts can be used to initiate actions across a single system or multiple systems.  GDPS can be used to monitor and control (stop, start, for instance) actions across a replicated environment; it can assist in load balancing; it can help in cross-site initial program loading (IPLing); and it can quiesce a system workload (to remove it before a hardware change window, followed by an IPL afterward).

Today, GDPS is deployed in four basic configurations.

  1. GDPS Metro (HyperSwap capable) addresses the requirements of clients using synchronous replication (Metro Mirror) and is tightly integrated into Parallel Sysplex clustering technology.
  2. GDPS Global orchestrates the recovery of systems that are using asynchronous replication.
  3. GDPS Metro Global combines the local, near-continuous availability of GDPS Metro with the unlimited distance recovery options provided by GDPS Global (which can be deployed in three and four site configurations).
  4. GDPS Continuous Availability uses near real-time replication to update data to a second instance of a workload in another location and manage the distribution of work across the two copies.

IBM states that, at present, about 75% of its GDPS deployments are in the financial services segment since this segment has the most stringent CA/DR requirements (this includes banks, securities, and insurance firms).  IBM also has over 200 site deployments in three- and four-site configurations.

As a trend, IBM sees strong interest in GDPS CA (continuous availability) configurations.

Summary observations

GDPS presents a means for enterprises to acquire an advanced disaster recovery/continuous availability solution without having to expend internal resources building their own resiliency/failover/disaster recovery architecture.  By turning to a trusted partner, enterprises can thus avoid writing their own custom code and implementing custom processes to ensure the viability of their computing environments.  What internal development does is increase developmental expenses while burdening the enterprise with long-term support/maintenance issues.

As a side benefit, placing the responsibility of developing advanced clustered systems designs into the hands of trusted business partners allows vendors to build in additional functionality while extending support for GDPS for the long term.  (IBM currently has about 50 developers devoted to expanding and improving GDPS – with deep knowledge of the product to aid in support issues).

On the surface, GDPS may appear to be an expensive solution (consisting of a completely redundant hardware environment as well as related software for failover, load-balancing, etc.) – but it is quite likely that enterprises that chose to build their own Z resiliency environment would spend significantly more on development of that environment, and on long-term support costs, while failing to keep pace with the new features added to the vendor’s product over time.

With today’s regulatory and market pressures to ensure operational continuity and data privacy and security, choosing a mature resiliency environment developed by a vendor with deep industry knowledge, deep product knowledge, and an expansive development/support organization makes more practical sense than going it alone using a custom internal approach.  GDPS offers enterprises a means to build resilient and highly responsive systems designs while placing the burden of developing and maintaining software on the vendor instead of on internal resources.  For enterprises looking to address regulatory requirements, reduce risk, recover quickly, and reduce development and support expenses, IBM’s GDPS should be considered as a means to address these requirements.