In an interview published in the January 1, 2016, issue of Fortune Magazine, discussing questions about the importance of Big Data when it comes to saving lives, the doctor being interviewed made a couple of significant observations. “When you have an entire medical record, there’s context around the data,” said Dr. David B. Angus, MD (and a best-selling author). “We may never have made that association based on biology, but that association … may tell us something important about how to treat such patients.”
However, it was another observation later in the interview that caught my attention. “With enough data,” said Dr. Angus, “error goes away.” Startling and yet, pretty obvious to anyone who has been working with databases of any kind. Paraphrasing Archimedes who once said, “Give me a lever long enough and a fulcrum on which to place it, and I shall move the world” today’s data scientists may be forgiven if they add, give us all the data and we can answer every question!
For the NonStop community, operating without limits has always been a positive attribute of NonStop itself – scale up and scale out provide almost limitless processing power. No user that I know of has ever sought the upper limits of NonStop although at the November, 2015, NonStop Technical Boot Camp, OmniPayments, Inc. CEO, Yash Kapadia, did reference one account in Asia where his OmniDirector message-bus based system is fully tested on a production system with 3,000 CPUs, where triple active-active-active replication is deployed.
The sheer power of 3,000 CPUs is somewhat mind-blowing and to put this into context the largest payments solutions provider supporting NonStop today has about 300 users who in total don’t have 3,000 CPUs between them. This may not be the fulcrum Archimedes had in mind as for many data scientists this amount of compute power is considerably bigger than anything Archimedes could have dreamt about – giving them all the data may not be as far-fetched as Dr. Angus may have thought. On the other hand, it’s also just the beginning and for the NonStop community that trusts in the direction Hewlett Packard Enterprise (HPE) is taking, the tap is about to be opened up more powerfully than at any time in the history of Tandem Computers / NonStop systems.
When it comes to the capabilities of Striim, and what it can determine when directed to “watch data streaming by,” can move the earth. Well, at the very least, the opinions of those capable of shaking up widely held opinions of those capable of making momentous decisions. For the NonStop community, simply gaining valuable insight into the operations of their data center in a manner that allows them to better defend against unauthorized access as well as more proactively support negotiated SLAs, no matter how aggressively presented.
To date, data center managers have struggled over graphs and charts that for the most part rely on assessments and projections that often are less than scientific (utilizing the data provided by the application / middleware under review with no independent or secondary contextual data) and it is the likes of products like Striim that can be configured to see all. Yes, all the data can be seen and the potential for error lessened, if not removed completely.
I am in regular conversation with the team at Striim and the opportunity that they have with the NonStop community is only just ramping up. The noise surrounding the old argument about “what has Big Data got to do with Transaction Processing?” is gradually receding – almost every middleware provider operating in the NonStop marketplace is waking up to the value that seeing all the data provides and are addressing accordingly (after a fashion). Can Striim add value to what every NonStop vendor provides today? Of course …
First NonStop users are just starting to deploy Striim in production settings. The first truly operational implementation is now only weeks away and I hope to be able to cover it in more detail shortly – even if it’s only the result of information provided at a user event which I am keeping my fingers crossed will happen before the end of this current quarter.
Most NonStop users will fall well short of having a need for NonStop systems that include 3,000 CPUs (although, the latest family of NonStop systems, the just released NonStop X, supports 4,000+ CPUs in a single configuration so there’s markets already being identified where such processing power is relevant), but knowing that such configurations can be supported allows these users to better trust what is coming from HPE. And trust is an important element – trusting the data is important but having trust in the primary vendor is every bit as important.
Witnessing just how deep an investment HPE has made in NonStop in bringing it to the Intel x86 architecture provides a huge dose of confidence for all of the NonStop community. Now it remains to be seen just how many NonStop users step up to seeing all the data – it can be done and the benefits are only just beginning to be realized by those sharing similar beliefs to Archimedes. Yes, show me the data!