Friday, February 14, 2014

Yet three more wishes!

At any major event, there is no more anticipated presentation than the vendor’s roadmap presentations … but wait, there’s always more! Looking ahead and framing as my wishes is just one approach that I take to speculate, which I can do without suffering the consequences vendors would face should they speculate!

It’s been six years since I first blogged about my wishes for NonStop – in the posts of February 12, 2008, and February 13, 2011. Now that it is February 14, 2014, it seems only appropriate for me to look at how far the NonStop platform has come, in terms of modernization as well as commoditization, and to express opinions on a likely future of the NonStop servers.

Whether it’s coincidence or not, the previous posts have featured photographs of cars that no longer can be found in the garage. Does it truly reflect my ability to predict the future or not, they didn’t make the cut! The cars pictured above tell a story of their own and I am keeping my fingers crossed that they will remain in the garage a little longer. They are both Chrysler SRT vehicles – the V10 being the Viper SRT while the V8 is the Jeep SRT. That these cars even survived at Chrysler, enjoying better days, is close to a miracle, and is a good lead-in to this post on NonStop.

In an article in the Harvard Business Review (HBR) blog, posted on December 24, 2013, Big Data and the Role of Intuition writer, Tom Davenport, responds to the question of “Whether intuition has a role in the analytics and data-driven organization.” Davenport begins his response with “The difference with analytics, of course, is that you don’t stop with the intuition — you test the hypothesis to learn whether your intuition is correct.” Furthermore, he then notes how often “The choice of a target domain is typically based on the gut feelings of executives.” Setting the stage for his final comments, Davenport then remarks, “In short, intuition’s role may be more limited in a highly analytical company, but it’s hardly extinct.”

To read of intuition, hypothesis, gut feelings, and then know that this method of figuring out what is actually taking place (in a market) is still valid (and not extinct), puts my mind at ease. Speculating about what is to come for NonStop servers therefore can be legitimately pursued based on my own intuition. As well as my gut feelings! With what has transpired these past three years, I believe anyone prepared to spend time reading everything that HP publishes and then asks pertinent questions of well-placed executives, should come to the same conclusions that I will make in this post.

In the post of February 12, 2008, "My Wish" for NS Blades my three wishes came down to firstly, wanting HP to deliver a shared infrastructure blades chassis. I then pushed a little further and with my second wish, wanted HP to provide a hypervisor whereby NonStop could become a guest operating system (OS). This essentially was my first take on the proposition that at some point, virtualization would become a factor in future NonStop configurations. With my third, and most ambitious wish, I wanted HP to provide an API whereby programmatically NonStop could be loaded on standard blades on demand. What we refer to today as provisioning was a capability I thought businesses everywhere would welcome and look to vendors to provide the monitoring tools whereby the contents of incoming transactions would determine just how many NonStop servers would be present within any given shared infrastructure blades chassis.

In the post of February 13, 2011, Three years on, and three more wishes! my next three wishes built on the earlier wishes (and yes, the shared infrastructure blades chassis came about), but added a couple of twists, or wrinkles, depending upon your perspective. This time, my first wish centered on NonStop becoming a purely software offering on the understanding that HP would introduce standard blade packages – going so far as to suggest that mezzanine, or daughter, cards would no longer be required. With my second wish, I proposed that future HP server offerings targeting the data center would all include NonStop servers as part of the package. Just as today we have SSDs with enough capacity to house important, frequently used data and files, so too would there be (shall we say) SSNS with enough capacity to ensure critical components would always be available. Finally, with my third wish, I drew a deep breath and expressed my desire for HP to pursue greater cooperation with solutions and middleware vendors as the built for the future – in other words, HP shouldn’t contemplate going it alone.

Looking back, much of what I wished for has come to fruition. There is now a shared blade infrastructure – the C7000 chassis providing the underpinnings for this. Throw in the soon-to-come InfiniBand support along with the plans to support Intel x86 architecture and much of what else I wished for is in sight. Equally as importantly, HP is working with partners – comForte, ETI, etc. come to mind – and that’s an encouraging sign. But now what? Virtualization seems to be as farfetched an idea as it was when I first began writing about it in 2008, even as NonStop packaged with everything seems to be getting closer with HP’s strong push into hybrid computing. Will we ever see NonStop inside a homogeneous array of servers stretching far into the distance, where physically it will be unrecognizable?

Expressing as I am yet three more wishes, here is what I expect to see rolling out over the next three years. First, how about repackaging TS/MP (Pathway) as VM/MP? I would really like to see some movement on this front, as today we have all accepted a form of process virtualization as we deal with Java’s Virtual Machine and .NET’s Common Language Runtime – what TS/MP provides today is independence from any awareness of where a Pathway ServerClass instance runs, and with the latest releases, this can include running on other NonStop servers. It is all there today available to all parties running NonStop, so what really is the wish? I really want to see HP become more proactive about the virtualization characteristics that already are available with NonStop and leverage as a starting point to even more adventurous options to follow. After all, surely NonStop can span multiple VMs, as today it spans multiple real machines!

Secondly, I want HP to build on its hybrid computing plans to embrace more fully NonStop – for the moment, it’s holding it’s punches somewhat and enticing a message about NonStop’s role in hybrid computing is even more difficult than getting HP executives to concur with the idea that NonStop has a role to play in clouds! Hybrids and Clouds are joined at the hip and whether outside the cloud, as a source of transactions, partway in the cloud as a gateway (directing transactions to the appropriate cloud based upon potential SLA commitments), or inside as a guardian overseeing security and fending off unwanted intrusions, NonStop has a big role to be played going forward.

Finally, my third wish? As I watch the UNIX business tank, the OpenVMS roadmaps come to an end and as Linux and Windows garnish more headlines (my observations, of course), then my final wish is for greater pride in NonStop emanating from all HP executives. Having plans now in place for NonStop to support the x86 architecture (on universal blades) is a huge step in the right direction but as young Oliver said, “please Sir, I want some more?” The more I get involved with Project Moonshot, for instance, the more I am becoming convinced that the potential to have a couple of Moonshot cartridges supporting NonStop with engagement in what’s being asked of the other Moonshot cartridges hold promise for much wider acceptance of NonStop.

These wishes may not come as a complete surprise to some readers, although the sequence may be different. In the post of July 19, 2013, Are our wishes still important? I talked about wishes in general, but now I am more confident that I am on the right track. While it remains very important for HP NonStop to present it’s roadmap for NonStop and to keep the NonStop community informed, roadmaps today aren’t what they used to be and don’t always convey the full story. Likewise, presenting a vision for NonStop is rarely done as public companies like HP are bound by regulations about making statements that are not based on something tangible. Discussing wishes, no matter who makes those wishes, is a healthy thing to do and I am hopeful what I cover here today leads to further discussions and possibly, even products. Yes, wishes remain very important!

For decades, we have lived with the Tandem fundamentals – availability, scalability and data integrity. With all that is happening in the world today, is it time to revisit these and add securability? I only throw this in as part of my final observations for surely, if this truly is an attribute helping to define the future for NonStop, then mixing NonStop into hybrid blades, into hybrid clouds and potentially into new technologies like Moonshot, assures a future for NonStop well beyond the scope of what may be on our minds these days. To paraphrase a popular television show “Living in Hawaii is not expensive – you just have to want it!”

Chrysler treats SRT as its halo brand and its products, representative of where the imagination of Chrysler engineers can take the company. It shouldn’t come as a surprise to readers that I am attracted to vehicles like this. Hopefully, as you consider my impressions about the future of NonStop your gut-feelings will kick in, as will your intuition, and experience similar insights. Pride in NonStop should be easy to accomplish. Participating in Clouds and Hybrids, a no-brainer! Word-smithing a credible story around NonStop in a virtual world? Yes, equally, a piece of cake! HP just has to want it!

3 comments:

Gerhard Schwartz said...

Hello Richard,

quite frequently you keep the best content for the end of your blog, also here in this case:

"For decades, we have lived with the Tandem fundamentals – availability, scalability and data integrity. With all that is happening in the world today, is it time to revisit these and add securability?"

I'd fully agree, NonStop is extremely good in terms of platform security, only very few other platforms like the old IBM mainframe and now retiring OpenVMS do come close.

With malware issues all over the place and resulting losses in the billions of dollars range (see http://finance.yahoo.com/news/global-credit-debit-prepaid-card-173000357.html), it is important to promote such a highly secure platform - which ultimately benefits everyone, except the hackers.

But this slightly conflicts with your other wish of seeing NonStop becoming part of component virtualization a la VMware et al. You can't expect very high IT security there, eg. as of today the NIST vulnerability database lists 654 known WMware vulnerabilities - significantly more than are listed for AIX or HP-UX.

When running on top of such virtualization software, NonStop OS would have no chance to see what goes on underneath and could become vulnerable.

Standardization is good, but too much standardization is counterproductive in the value business. To be better, you will always have to be a bit different.

So yes, I fully believe we should keep NonStop integrated in HP's Converged Infrastructure, but also I'd believe that NonStop should not be separated from the underlying hardware.

Anonymous said...

Well said, I agree, lets hope your dreams will come true. if not our last 10% installed base will be gone.
wishing along with you.
regards,toril

Richard Buckle said...

This subject - clouds and then virtualization - came up last week during my presentation to SunTUG.

I see the emergence of hubs or gateways (or whatever) emerging providing both a sentinel and a traffic cop role - protecting and serving, so as to speak. And this will apply to how we get to cloud(s).

I don't see this conflicting with NonStop at some point treating whole virtualized environments as if they were real, and for NonStop to be astride multiple virtualized environments switching from on to another when it suspects one to be failing ...

Whether NonStop ever sits directly atop a metal-specific virtualization, I am still not sure about this for the very reasons you raise. Does this help clarify?