The latest on Fault Tolerance: HPE is a key player in the AL4 market according to IDC

Recently, the IDC report on Fault Tolerance as recognized by IDC as Availability Level 4 (AL4) is being highly promoted by HPE. IDC’s updated analysis of HPE, as seen prominently on social media sites liked LinkedIn, in support of SuperDome Flex being given AL4 classification. HPE’s enterprise-class computing mainstay, SuperDome Flex, now occupies a place in this highly prized category alongside offerings from IBM (Parallel Sysplex), Oracle and yes, HPE NonStop. This IDC report can now be downloaded from the HPE web site using this link: Mission-critical platforms deliver continuity in the shift to digital-first strategies

While this is good news for HPE in their pursuit of further differentiating their premier enterprise-class servers nevertheless it has many within the NonStop community wondering if the attention HPE is giving this family of servers detracts in any way from the prominence normally reserved for the HPE NonStop product suite. Complementary or supplementary? That seems to be the question of the day. But what exactly does IDC mean by classifying any system today as enterprise-class and categorizing it as AL4?

True fault tolerance through redundancy in hardware and software: This approach provides continuous data processing, even in the event of the failure of one hardware or software component. Because all data processing (i.e., support for a given application) occurs in a side by-side fashion, the failure of any single component in the system will not affect other sets of computing resources that also support that same application. As a result, computing can continue uninterrupted. However, in the event of a natural disaster or widespread power outage, the workload would still need to be replicated on alternate servers outside the campus, city, or geographic region that experienced the outage.

Shortly after the concept of Availability Levels begun being promoted by IDC, in my post of July 13, 2014, HP NonStop systems’ deployments you might have missed! I tapped into yet another paper from IDC. Written by IDC Group Vice President and General Manager, Enterprise Platforms, Matthew Eastwood, and published in November, 2013, it too can be downloaded from IDC – just follow this link; Mission-Critical Business Applications: The Need for Always-On Servers. IDC’s Eastwood stated that, “When downtime is not an option, organizations are increasingly turning to fault-tolerant systems to keep their business up and running … Downtime of any kind results in a loss of confidence and competitive advantages in the marketplace.”

For the NonStop community this is well understood, to the point where today almost all deployments of NonStop in support of mission critical applications result in NonStop systems residing in two or more locations. As has been reported by NTI, with products like DRNet®/Unified, addressing coverage during any disaster has become second nature for almost every NonStop user and with NonStop qualifying as an Enterprise Class computer on all counts, it seems that perhaps categorizing an ever-increasing population of servers as AL4 is weakening the definition rather than broadening the options available to all enterprises.

Looking into this latest IDC report that features HPE SuperDome Flex, I was shocked by how easily IDC move between generalizations. Take the definition IDC provides for classifying enterprise-class computing and yes, it’s a long list that covers the “12 Tenets of Enterprise-Class Computing”

• Capacity. The ability to take on extremely large workloads without choking on data volumes, transaction volumes, or processing and transaction speeds
• Continuity and compliance. The ability to withstand a wide range of failures at the hardware, middleware, and application levels to the point that processing continues without noticeable interruption or data loss even if the entire site (cloud or datacenter) fails as well as the ability to report comprehensively and easily on a platform’s compliance with regulatory mandates
• Changeability. The ability to flexibly deploy, scale, move, or terminate workloads (This includes the ability to respond quickly and comprehensively to such massively disruptive events as the COVID-19 pandemic.)
• Cognition. The ability to analyze data in real time, including through the infusion of AI and analytics into all levels of the platform stack to achieve better security, greater automation, and increased AI-driven application functionality
• Customer centricity. The required functionality to enable a singular focus on the end user — whether that is a consumer, a business partner, an ecosystem participant, or any other type of user — with all platform capabilities flowing from that starting point
• Cyber counteraction. The ability to withstand or counter and remediate any form of attack on the system, from both inside and outside the organization
• Cloud compatibility. The ability to operate like a cloud (private or hybrid) as well as to seamlessly operate in conjunction with other dedicated or public clouds
• Community-based (open source) software. The ability to support open source software solutions and programming languages to such a degree that developers can freely leverage the platform without requiring additional skills
• Consumption-based cost. The availability of a pricing and/or licensing scheme that is mostly or entirely based on the actual consumption of the compute resources
• Cryptographic capability. The ability to encrypt and decrypt all data, at rest and in flight, without a noticeable performance overhead and with encryption strong enough to withstand current and future code-breaking technologies
• Climate consciousness. The ability to decrease the platform’s climate footprint with each new generation without compromising its performance
• Connectedness. The ability to connect easily and securely to internal and external environments such as other infrastructure in the datacenter, at the edge, or in the cloud as well as with end-user environments such as the web, mobile networks, and industry networks

It is only later in the IDC report that they then acknowledge that “The HPE Superdome Flex, powered by Intel Xeon Scalable processors, delivers on most of the 12 tenets for enterprise class computing, as defined by IDC.” Yes, most of the 12 tenets but then no indication really of what was overlooked. Now, to be very clear, the HPE SuperDome Flex represents a family of successfully marketed enterprise-class computers, they have established a very loyal community of users. In many ways this is no different to what we observe with the NonStop community. Yet, with the initiatives completed by the NonStop development team that visibly align with the 12 tenets of enterprise class computing at no time has IDC ever said that HPE NonStop “almost” delivers enterprise class computing.

IDC writes about HPE SuperDome Flex being:

• Powered by Intel Xeon Scalable processors, delivers on most of the 12 tenets for enterprise class computing, as defined by IDC.
• AL4 Availability on HPE Superdome Flex IDC considers HPE Superdome Flex an AL4 system, delivering fault tolerance.
• HPE is a key player in the AL4 market.

Returning to the value proposition of HPE NonStop systems, as recently as 2019 (and covered by another IDC report this time sponsored by Oracle), it was reported how HPE NonStop was showing signs of increasing its market share following the move to the Intel x86 Architecture:

HPE grew 21.8% y/y to $260.7 million, largely because of a shift in product mix. As businesses continue to shift away from the proprietary Itanium-based Integrity NonStop to the more affordable x86-based NonStop X, Average Selling Price (ASPs) have declined significantly, resulting in savings for their customers and allowing HPE to capture some market share.

Nice one, Oracle. However, here is the tenet that is likely to be the most important tenet of all; the end user experience. How transparent and indeed seamless is the transition to backup resources following detection of a fault either localized as in a process or processor or traumatic and wide spread as in a disaster? As IDC has loosened the qualifications to join the AL4 club as the data now suggests then is there a need for an AL5?

There is logic for IDC to loosen the qualifications as industry analysts living off the revenues of their research reports find it counterproductive to position just one or even two vendors in a classification. The more widespread the participation by vendors the greater the revenue opportunities and this has been the primary operational tenet of all such institutions. No, not all systems are fault tolerant to the point where take-over is performed transparently to the user and seamlessly for IT operations.

When raising the prospect of AL5 with industry experts within HPE, there was little support afforded such an idea. However, what did come out of the exchange was the challenge it afforded, “It would be interesting to take parts of the latest entries into this IDC category of AL4 (an Exadata, SuperDome, any cluster in general) down and ‘see’ if the users are aware something happened:  I think they would know something happened.” IBM may have the lion’s share of the AL4 marketplace with Parallel Sysplex deployment but the secondary challenge here is how many enterprises actually expand the necessary resources to move from simply deploying Parallel Sysplex to capitalizing on its capabilities.

When I first raised this topic with NTI it soon became apparent that the NonStop community is well versed in operational aspects of ensuring fault tolerance at the application level. With DRNet®/Unified deployed and in production use across hundreds of HPE NonStop systems in support of even higher numbers of hundreds of use cases, visibility into this market has helped NTI to continuously adapt to application requirements. No more so than when the applications are in support of financial services where DRNet®/Unified has a strong foothold that continues to expand.

“What is important to realize,” said the HPE industry expert, “is that these IDC reports are very much a collection of estimates” supported by the opinions and observations of the analysts involved. “Perhaps it’s not so much the need to create an AL5 to better reflect the capabilities that HPE NonStop brings to the table so much as there is the need to introduce AL3.5” that is more inclusive of the mix of clustering and virtualization we see entering the marketplace today. However you elect to categorize HPE NonStop systems the most important take away of all is that as of today, no other vendor has been able to knock it off the highest possible pedestal accessible to all those who require am enterprise-class computer that meets all 12 tenets and is fully fault tolerant.

In this regard NonStop stands alone and will continue to do so for as long as we care about meeting the needs of our end users. For this, HPE continues to excel and in so doing, not only sits atop a pedestal alone but does so in a manner that makes any user, any enterprise, any application capable of readily leveraging HPE NonStop systems to the benefit of the enterprise as a whole. 

#NonStopRocks