The ABC’s of an Affordable Business Critical Compute

When it comes to TCO, HPE Mission Critical Systems’ NonStop continues to surprise!

We have always heard the role of NonStop described as being best-suited platform where 100% uptime is mandatory; where the customer’s business cannot tolerate any exceptions; where nothing can be allowed to interrupt the works of the solution and where the application’s failure means the customer’s business fails.

There is a cost when a Business Critical application fails, especially when the failure affects publically used and very visible applications like ATMs or POS devices. An outage or even a brownout – the failure to accommodate transaction peaks – can cost a company more than just the loss of transactions in progress, but also customer confidence, expensive SLA fines or wasted staff time mitigating the failures and responding to the press.

There is a growing sense too that perhaps cloud service providers will make the need for business critical compute unnecessary, at least within the walls of a traditional data center. However, even here there are warning signs appearing that perhaps clouds have begun losing momentum within the enterprise. Perhaps too the hybrid IT model is here to stay:

“Every day, one comes across statements or truisms that people take as common knowledge but that are actually wrong. Darth Vader never says ‘Luke, I am your father’ and no one in Casablanca ever says ‘Play it again, Sam.’ And it looks like we can now add to that list the misconception that all workloads are moving to the public cloud.”

THE MYTH OF THE 100% CLOUD WORLD:
MANY BUSINESSES ARE INSTEAD BOOSTING ON-PREMISES WORKLOADS
May 2020 Jim Rapoza Research Director, IT
Aberdeen Research (as published on HPE.com)

 

Even more telling were the findings of IDC as published in its updated July 2020 Availability report:

“Mission- and business-critical workloads are growing, and business support functions that previously could be run on a low availability tier are increasingly deemed business critical.

“Even as availability and security in the public cloud have greatly improved, true fault tolerance continues to be seen as an on-premises or hybrid cloud capability, not as a public cloud capability. IDC research shows that 38.5% of businesses hosts the highest availability tier on on-premises infrastructure, whereas only 2% of businesses host this tier in a public cloud.”

Worldwide AL4 Server Market Shares, 2019:
Fault-Tolerant systems become Digital Transformation (DX) platforms.
Paul Maguranis Peter Rutten

 

Cloud computing just like Client/Server computing before it, will become ubiquitous in time leading us to acknowledge that the cloud experience is everywhere and the very prospect of delineating any business-critical application as cloud resident or not will be a moot point. Enterprise hybrid IT will be with us for yet another technology lifecycle that appears not to be ending any time soon. New systems will continue to be evaluated on the basis of meeting business objectives at affordable price points and the need for systems running 24 X 7 is becoming more widely appreciated.

The move away from assessing these business critical systems’ purchases in terms of Price / Performance to giving greater attention to a customer’s Total Cost of Ownership (TCO) has been hard to ignore. Such a move has helped shift IT professionals’ perceptions about the true cost of NonStop. This past decade has helped illustrate that there are two very different sides to costs and to the value any system provides, but the most important value proposition of NonStop systems remain. It’s all about uptime, surviving failures and delivering the industry best levels of availability.

 

A is for Availability

When it comes to NonStop its availability and scale-out attributes have made it one of the first out-of-the-box systems satisfying IDC’s (International Data Corp’s) Availability Level Four (AL4). Recently reviewed and renewed in this category for 2020 by the IDC, NonStop outages are transparent to the user where total downtime is measured in seconds per year. Often described as a cluster in a box, NonStop systems are delivered with a software and hardware configuration that work together to prevent any single point of failure from affecting the Mission Critical workload running on the system.

Outages represent cost to a business. A 2019 – 2020 Outage Impact Study conducted by the vendor LogicMonitor concluded that:

“Availability has become our most valuable commodity, yet high-profile outages and brownouts are occurring at an alarming rate.”

Summarizing the results of its most recent survey LogicMonitor confirms what the HPE NonStop team and HPE’s NonStop partners have known for some time; availability is of paramount importance. “80% of respondents indicated that performance and availability were important issues. Downtime is rampant. 96% of global IT decision makers have experienced at least one outage in the past three years.”

Furthermore, the survey revealed, “Downtime is expensive. Companies with frequent outages and brownouts experience 16 times higher costs than companies with fewer instances of downtime.”  System failures these days can also lead to embarrassing public situations for different businesses, especially those in the Finance or Banking industry.

In this context, brownouts are viewed as those conditions where a sudden transaction spike leads to a degradation of response times to where it is concluded to be inoperable and well below any acceptable Service Level Agreement (SLA) metrics. A failure to scale-out in times of crisis can be even more unacceptable than a total stoppage as it may take time before the brownout condition is recognized. Scalability matters just as much as availability and no conversation about TCO can ignore that there is a price to be paid and a risk to a company’s brand whenever a system fails to meet end-user expectations.

The importance of availability to enterprise IT was referenced by IDC in its update on the AL4 marketplace:

“The end of ‘regular business hours,’ with businesses’ applications required to be available to customers at all times, has put tremendous pressure on the infrastructure that supports those applications, allowing for little if any scheduled or unscheduled downtime.”

And in an update to its findings reported in 2014, IDC provided the following:

“The cost of downtime is increasing as businesses become more and more dependent on their infrastructure for daily operations. For 20.7% of organizations, the cost of downtime is $5,000- 10,000 per hour; for 18.4%, it is $10,000-25,000 per hour; for 17%, it is $25,000-100,000 per hour; and for some businesses (1.4%), it is $500,000.”

Irrespective of where your business may fall within this delineation, availability and with it the cost of downtime is arguably the primary reason why enterprises continue to evaluate systems on their propensity for uptime. No business can afford the fallout of disenchanted customers and business partners following an outage of any kind.

 

B is for Bandwidth

Consideration of bandwidth is a reference to the ability of a system to handle devices, transaction volumes and database access in a manner that is readily scalable without the need for any downtime. Most important of all, when it comes to expanding any aspect of NonStop, by planning ahead this expansion can be achieved without significant downtime.

Everything needed to run a business critical application could be selected to create an operational environment on a NonStop system, out of the box. HPE resells many NonStop partner software packages that can be added to the deal for the customer’s convenience. But at the basic level NonStop scales.  And it scales massively. From a small 2 processor system, to a large 16 processor system that can be clustered into large systems of 4080 processors for a total of 24,480 cores with the same system image and application environment.

When compared to the market share of vendors like IBM and Oracle, IDC in its AL4 marketplace findings, reported:

“In 2019, the worldwide AL4 server market declined 5.6% compared with 2018 from $3.68 billion to $3.5 billion in large part because of lower average selling prices (ASPs) for the platforms included.

 “IBM’s share has declined somewhat; HPE, on the other hand, has picked up share; and Oracle has remained relatively steady.

 “As businesses continue to shift away from the proprietary Itanium-based Integrity NonStop to the more affordable x86-based NonStop X, ASPs have declined significantly, resulting in savings for their customers and allowing HPE to capture some market share.

 “This is indicative that the new platform is much more affordable and becoming a strong driver of DX growth for their consumer base.”

When contemplating why NonStop systems today are becoming affordable and acquiring a growing market share, it’s important to recognize that it’s still the NonStop architecture that has served enterprises so well for decades. For the NonStop community this doesn’t come as a surprise but to the broader IT industry, it is material to know that NonStop has the bandwidth of compute, communications and confidence (in security) that they now need.

 

C is for Cost

Once we have come to better understand the benefits of availability, scalability and the overall bandwidth NonStop delivers and realize the contribution NonStop can make to business critical applications, the cost of NonStop enters the conversation. This is not unexpected and yet, even with the shift to commodity hardware and open systems there is still considerable misunderstanding of the affordability of NonStop systems today.

In calculating TCO, there are many models being promoted that can help rationalize a system purchase. What are the constant reference metrics that appear in these TCO models? You will always find reference to purchase prices, migration costs, access to skillsets (in support of implementation and then ongoing management of the system), data center infrastructure costs (including data center space, power and cooling), and all software costs associated with the complete software stack together with the cost of the application that runs on top of everything.

This is where cost of downtime begins to become extremely relevant as most models summarize the above in terms of TCO being the sum of initial costs, operating costs (including premise infrastructure, cost of staff and all software), maintenance, downtime and the ability to satisfy SLA terms and of lesser importance in today’s world of technology breakthroughs, any residual value. Simply stated, this could be represented as TCO being the sum of initial costs plus operating and maintenance costs less residual value around the effort required to keep the system running 100% of the time.

HPE’s NonStop team is delivering systems today where the initial purchase price is considerably lower than the competition.  When it comes to traditional purchases of NonStop systems, a recent example best illustrates the glaring disparity in initial pricing. Comparing a NonStop X NS3 system with NonStop SQL/MX against an Oracle system, the numbers tell their own story. Configured with Oracle RAC for improved availability and deployed on just two Linux servers, each with 4 cores (for a total of 8 cores spanning the pair), comes with a hefty price tag just for Oracle.  In the US, expect to pay approximately US $35K per core per year just for Oracle/RAC for a total of US $280K per year for the system. The hardware and Operating System (OS) in support of Oracle/RAC still require payment.

Meanwhile a four processor, two core (with a total of 8 cores), NonStop X NS3 system with NonStop SQL can be purchased for the surprising total of US $201K per year, including maintenance (based on a five year term license). Unlike Oracle/RAC purchases, single core NonStop X NS3 can be purchased for a lesser amount just as NonStop SQL can be optional. Figures referenced here are based on recent sales and there may be some additional costs to do with final memory configurations and unique services selected by the customer. The difference in the cases however will not be material.  HPE’s NonStop’ offering generally ends up being a third less than the cost of an Oracle/RAC solution.

Surprised?   In the past, it has not always been an apples to apples comparison. NonStop systems come with all the software layers provided on the system, so the total system cost needs to be compared, not just the price of a Linux server with a barebones Linux system with only the OS running on it.  On a Linux system, the OS layer, Security layer, Database, Manageability tools, Replication/DR tools, application etc. all have to be purchased separately. On NonStop everything needed to run the application can be included to create the customer’s environment for the system out of the box.

 

D is for Database

A system with fault tolerance built into it generally requires less time of System Operators and Database Administrators to monitor and to take urgent and oftentimes reactive actions to near failure situations.  We’ve often encounter  Oracle database users with 5 or 6 database admins needed to monitor and manage the Oracle environment vs a similar sized database infrastructure on NonStop where only a single database admin was required. Furthermore, and a direct bearing on the affordability of NonStop, all utilities are available and required services can be performed online with no need to ever stop the underlying relational data bases management system.

NonStop is where data is created and it’s as fresh as data can be – it’s important to have this data adding to current business insight and for business this means ensuring as readily available and easily scalable SQL. NonStop offers a fully functional SQL/MX database created and supported by HPE and the database is fully fault tolerance and scalable as well. Making sure data is always available and at the right time is of paramount importance to business and, according to IDC:

“Making the data available to new customer-facing applications is what transforming AL4 systems is all about. The data that AL4 systems process is of the highest value, often with real-time aspects, single source of truth (SSOT) implications, and in-flight security and data privacy considerations.

 “The most important reason is that these (AL4) systems process many enterprises’ most critical and most valuable data, often with greater volumes than any other data types, and businesses have decided to unlock this data and leverage it for their digital transformation.”

One NonStop vendor who has been actively engaged in helping NonStop applications integrate with the world around them has been Striim with its real-time data streaming platform that supports NonStop. When it comes to NonStop, Striim Cofounder Sami Akbay is quick to point out that:

“At Striim our efforts have been directed at ensuring data created on NonStop can be readily made available across the enterprise. The Striim Platform is ideal for ensuring NonStop data is never siloed and locked away from applications providing greater business insights and there are already NonStop customers realizing value from tapping into the fresh data only NonStop can provide.”

Business insights derived from fresh data, performed in real time, is providing the competitive advantage needed by business in today’s marketplace. Doing so at an affordable price has seen tradeoffs being made in the past, but not any longer:

“Until the mid-2010s, unlocking the data in an AL4 system without spending significant time and resources on moving it off-platform — and potentially compromising the data in the process — was nearly impossible. Businesses were unable to easily perform real-time analytics on their core transactional data …”

NonStop systems in possession of data as it’s being created, in real time, can meet the needs of business and do so at less cost than enterprises may have thought possible. Expanding market share in the all-important AL4 marketplace isn’t an accident and it is the presence of NonStop SQL that is the most visible contributing factor in the affordability of NonStop. Software from HPE NonStop partners allows for making this data available in real time to analytics platforms so analytics can be performed in near real time.

 

E is for Ecosystem

When it comes to those businesses that express some nervousness about finding products, applications and qualified staff, this has eased considerably in recent years, as NonStop has hundreds of partners around the world who provide such products and can offer access to such skills. Of course, HPE provides services through the Pointnext organization and the NonStop partners’ services are complementary, but the mere presence of such a diverse partner ecosystem speaks volumes to just how strongly these partners are committed to NonStop systems. Again, there is no real mystery as to the number of NonStop partners present in the AL4 marketplace given how HPE’s NonStop Partner program has helped find, cultivate and connect these partners with new NonStop users.

According to UK-based TCM Managing Director, Daniel Craig, “There’s no real need in trying to learn NonStop, just leave it to the experts. The NonStop customer gets all the benefits of the NonStop platform with little to no effort regarding management or maintenance.” In the Americas, it was Odyssey Information Services Business Development Manager, Ben Coan, who said, “We are always open to sharing our knowledge / expertise with our clients.” As for large solutions offerings such as those supporting payments processors, OmniPayments CEO, Yash Kapadia, said, “We demystify the entire process so a customer does not have to worry about a new platform. OmniPayments will install the software, take care of the configuration and add automation and dashboards so the system is in self-managed mode. The customer only gets email and alerts via text when things go wrong.”  Today, the NonStop business has enough Service partners to bring thousands of skilled NonStop personnel to our customers.

So whether it is about finding staff or offering a full-service program designed to minimize the need for staff or even the remote operations of major solutions running on NonStop, examples of these offerings can be found around the planet.

 

Summary of ABCs

Availability, Bandwidth, Costs, Database and Eco-system together provide a compelling case for the TCO and customer satisfaction when choosing to standardize on IT technology.

When it comes to the true sums supporting TCO, no matter which model you chose, you cannot ignore the soft costs that include avoiding the cost of downtime, or the fragility arising from complexity and yes, when it comes to HPE’s own IT, it should come as no surprise to businesses everywhere that today HPE IT also chooses to run HPE business on an affordable NonStop in-house using the NonStop database!

Points of failure, inability to scale, complexity, focus on attracting and then retaining highly skilled operations staff, layering software upon software for greater operational insight and much more are all real world considerations when tabulating TCO for any platform.  Unfortunately, oftentimes it is just one consideration that casts its long shadow over all other considerations and that is the initial purchase price. But in the world of hybrid IT that has become ever-present in all data centers, NonStop offers value for business critical applications that is not lost on IDC:

“…  the homogeneous datacenter of the past, in which they stood out, is changing rapidly, with purpose-built platforms for various workloads now becoming the norm. As a result, AL4 systems are no longer considered the “sore thumb” in an otherwise homogeneous datacenter but are rather seen as one of those purpose-built platforms — the one where the most valuable data is securely processed with extremely high availability (HA) as part of an organization’s on-premises/cloud infrastructure. These diverse platforms built for various workloads increasingly share a common, open source plane for application mobility and data sharing.”

In the past, it has not always been an apples to apples comparison. NonStop systems come with all the software layers provided on the system, so the total system cost needs to be compared, not just the price of a Linux server with a barebones Linux system with only the OS running on it.  On a Linux system, the OS layer, Security layer, Database, Manageability tools, Replication/DR tools, application etc. all have to be purchased separately. On NonStop everything needed to run the application can be included to create the customer’s environment for the system out of the box. TCO has never been an exact science and all too often the models can be weighted to favor one metric over another. Having noted this, it may come as a surprise to many just how affordable NonStop Systems have become lately.

Author

  • Richard Buckle

    Richard Buckle is the founder and CEO of Pyalla Technologies, LLC. He has enjoyed a long association with the Information Technology (IT) industry as a user, vendor and more recently as a thought leader, industry commentator, influencer, columnist and blogger. Well-known to the user communities of HP and IBM, Richard served on the board of the HP user group, ITUG (2000-2006), as its Chairman (2004-2005), and as the Director of Marketing on the board of the IBM user group, SHARE, (2007-2008).

Be the first to comment

Leave a Reply

Your email address will not be published.


*