Wednesday, May 6, 2020

Scaling out: NonStop, the “secret ingredient” within HPE product portfolio

An uptick in blog posts and even video clips helps ensure the story of NonStop continues to resonate with IT professionals even in times severely impacted by global events.

There has been a constant reminder in our neighborhood that we each are looking at the world differently. The global pandemic has forced us all into thinking about how big our social network should be – do we have enough friends? Do we need to reach out to even more colleagues and acquaintances? Is there a minimum set that ensures our happiness? A maximum? Can our emotions scale accordingly, such that our friendship is all embracing without breaking down into an inner circle and a perimeter?

This weekend we saw an impromptu performance taking place on a neighbor’s driveway that attracted a number of passersby. A simple guitar was all that was involved and yet those listening to his performance considered it good enough to stop whatever else they were doing. Our next door neighbors arranged a number of deck chairs on their backyard lawn, the better to maintain social distancing, for an opportunity to roast s’mores.

With very few breaks we have watched golfers walk by, isolated as they are in individual golf carts, all the while practicing social distancing as they take turns to sink those final putts.  Colorado weather changes in the blink of an eye; it was only a week or so ago that the course was covered in snow.

The world, however, isn’t the same as it was as we still have cars sitting in the garage on battery tenders as we wonder when next we can plan on driving to a NonStop Regional User Group (RUG) event. The latest update on planned events for 2020 that we received from Kathy Wood was depressing as we scrolled down through the list of cancelled and postponed meetings.

What hasn’t changed in these times, however, is the need for our infrastructure to remain reliable. It would seem that with each government initiative that is announced, the supporting infrastructure runs into problems. In Australia and the US, program beneficial to the community seem to have run fowl to the collective inabilities of applications to scale up to handle the rush of responses each program created. Higher education bodies have struggled to support student populations working from home.

All the while financial institutions we have put our faith in have likewise struggled to provide the reliability we expect at a time when accessing the dwindling amount of cash we might have becomes a priority. In her May 1 post to the HPE Community blog, Unplanned downtime and outages can happen. But not with HPE NonStop Karen Copeland, Manager, Worldwide HPE NonStop Product Management, references recent outages at well-known fintech, Robinhood.

Popular investing app Robinhood went down in early March,” said Copeland. “According to CNBC, the commission-free trading platform experienced a massive outage on March 2 that resulted in clients unable to access the market during a huge rally. The service faced downtime again on March 9 before managing to restore its systems. CNBC noted the firm blamed the outage on an ‘unprecedented load’ on its infrastructure caused by volatile market conditions, record trading volume, and a piling in of new users.”

Copeland also referenced another source, Cointelegraph, and quoted Pankaj Balaji, Delta Exchange CEO, who said, “The technology needs to be built in a fashion that it is able to distribute the load and handle such spikes. These frequent outages expose the fact that there are issues with Robinhood's technology and its ability to handle volume spikes in such high volatility environment.” For the NonStop community, the collective sucking noise as we gasp at such statements should be heard worldwide. Surely not – unable, you say, “to distribute the load and handle such spikes?”

If you have trouble accessing this post via the hyperlink above you can cut and paste this url into your browser –

Availability has always been the strong point favoring selection and ultimate deployment of NonStop systems. As a very modern interpretation of a fault tolerant architecture that has proved reliable for decades, it still surprises many that so little is discussed about the suitability of NonStop to support applications in a way that no single point of failure will bring them to a halt.

While other vendors promote the level of redundancy built into their systems, and of how such redundancy extends to duplicate power sources being supported, fault tolerance is so much more. What HPE has with NonStop is an integrated system from the metal to the operating system to the middleware stack – it’s all integrated in a manner that application failings don’t bring the system down.

How often do we have to remind ourselves that it is take-over and not so much fail-over? NonStop simply steps in and intervenes whenever situations arise that NonStop deems likely to end in a failure of one type or another. However, availability is only one part of the story. As a massively parallel processor with a shared-nothing architecture with NonStop, scalability no longer is an issue.   

In an upcoming article to be published in the May, 2020, issue of NonStop Insider, HPE NonStop Sales Specialist, Steve Kubick, promotes his most recent video posted to LinkedIn on Integrity and Inheritance. “When it comes to availability we measure it as being five nines (of availability) – three seconds of downtime a year. When it comes to maintenance or planned downtime, NonStop does that on the fly – all those things you need to do to maintain a system (can be) done online,” says Kubick. 

Just as importantly, Kubick continues with the observation that, “Without an outage: NonStop delivers AL4 uptime. And it does this while still providing a level of scalability that delivers 98% of each added processor resource (up to 4080 processors) – all in a single system image. So imagine what you could do with NonStop.” When others are struggling to scale-out to handle the unexpected volumes that come with business spikes, as we are seeing happen more often of late, NonStop takes such spikes in its stride.

For those relatively new to NonStop, you may have not realized that in the stock market crash of 1987, Black Monday, when the market dropped 20% in a single day, processing of orders were being performed on a NonStop system. Through planning, the exchange had additional processors on hand and then-Tandem engineers were able to add these processors online and without any necessity for an outage of any kind. The exchange didn’t stop trading that day because of a lack of scale but rather, its famous stock ticker couldn’t keep up.

No comments:

Looks can be deceiving! HPE NonStop; when being the best still matters!

For the NonStop community, we know what looks good may not only be deceptive but borderline dangerous; mission critical applications are bes...