Tuesday, August 25, 2020

Time to get my head out of the clouds …

We have now heard that the easy stuff has moved to the cloud but that leaves 70 percent of apps and data where they have always been. What does that mean for NonStop solutions?


I was sitting inside the house for the past couple of days, avoiding breathing in the outside air, as the sky has been a source of constant smoky haze. Living in Colorado, we often refer to the end of summer as the beginning of the bushfire season and it hasn’t disappointed us this year. It’s awful and for anyone with allergy problems it is a time to hit the antihistamines hard. On the other hand, our friends in California are facing an even more problematic circumstance – should we stay or should we go. To think that the governor is calling for Aussie firefighters telling all and sundry that they are the best in the world!

Put me in the category of agitators extolling the virtue of pulling out all the eucalyptus trees growing in California. Perhaps the Aussie firefighters are that good because they face horrific eucalyptus fires each and every year and know the value of ensuring the forest floors are kept free of debris. Oh well, with smoke in the air and the stench of burning timbers proving difficult to ignore, it’s hard to think about the sky and the clouds traversing the heavens above us. To be truthful, Margo and I haven’t seen any clouds for the past couple of days as the smoke haze hangs low.

However, for anyone who checks social media feeds, you must have seen the many references to clouds, cloud computing and hybrid IT. At a time when the discussions among IT professionals continue to address Digital Transformation (DX), it seems as though clouds and DX have become almost inseparable. Though essentially different topics the intersection of clouds and DX is a reminder that when it comes to our data centers and the solutions theses data centers support, any moves to pursue DX has a very visible line of clouds hovering overhead.

Like everyone else who attended this year’s HPE Discover Virtual Experience I was surprised to hear HPE CEO Antonio Neri reference recent data from Gartner:

Today public clouds are extending their platforms into the data center in what [market researcher] Gartner calls distributed cloud. The problem is these solutions were architected to be very centralized and rely on connectivity to their control plane. They still don’t address the over 70 percent of your apps and data.

This is not the first time I have referenced this remark of Neri, as reported by Steven Burke in a CRN update, but it is worth taking a second look. With as much discussion as there currently is about clouds, should we be surprised to hear that only 30 percent of our apps and data have made it to the clouds? Should we be a little concerned over the fact that it’s not going all that well for those enterprises banking on living in a cloud-dominated operational IT environment? Clearly, security always steps into the discussion but what is really hindering the move to clouds, distributed or otherwise?

A successful advertising campaign features a character called Captain Obvious. As I dug into this movement to clouds a little deeper I came across a posting to LinkedIn by HPE influencer and blogger, Keith Townsend. Otherwise known in social media circles as the CTO Advisor, he promoted a short video clip by podcaster and economist Corey Quinn. According to his LinkedIn profile, Quinn is “the Cloud Economist at The Duckbill Group (where) Corey specializes in helping companies address horrifying AWS bills.” In his preface to the video, Townsend posted:

Cloud migration is hard. Every option to do it using Corey's language "Sucks." So, we end up with a hybrid infrastructure. Which is where we should have started by design.

The key words here aren’t so much “hard” or even “sucks,” but rather “start with design.” No enterprise looks forward to migrations. Neither does it look fondly on re-architecting existing solutions. In the 1990s, so many major corporations embarked on complete overhaul of their systems where timeframes were reported as being three to five years and yes, much longer than the typical duration of any CIO at these enterprises at the time. No surprises here to read that midway through the re-architecting and re-implementing, the enterprises’ IT strategy changed dramatically.

Furthermore, I haven’t heard of any system or platform migration where the financials were fully understood ahead of time. As we so often like to remind ourselves; if it seems too good to be true, it likely is! To date, very few enterprises have publicly acknowledged their cost projections as it’s still all being pursued on the basis of a “feel good” philosophy. Budgets, fully described and appropriately annotated, are a rarity and I would challenge any enterprise to fully disclose to agencies, like the SEC, the likely impact on shareholder value!  Ouch; we are paying for this?

But moving to clouds is hard. And as Neri reminded his audience, what was mostly moved to the cloud had been the easy-to-move applications. In response to Townsend and Quinn, another cloud leader, Richard Bown, had this to say: 

I don't see much magic here. Cloud itself is not an end state. There is no end state. All we are doing is applying previous (project) learnings to a new array of diverse commodity products. Make yourself technically aware of the offerings and apply sound approaches to migrations. Any effort you spend in relearning or even rebuilding your systems will pay support and running cost dividends.

Unfortunately, as true as this might be, what might concern the NonStop community in particular is the reference to the pursuit of “relearning and even rebuilding!” Whenever something new floats up from the technology horizon, our first instincts is to contemplate a complete re-do. But as those massive technology undertakings of the 1990s reinforced, it’s always better to pursue baby steps than it is to attempt wholesale re-engineering. Big projects often fail not because of the lack of skills and availability of expertise but because something even newer comes along!

Call me a pragmatist and someone who is adverse to any rip-and-replace mindset. And yet, when thinking of any migration to a cloud environment, the pundits are now arguing strongly to simply migrate without changing a thing. It’s as though rip-and-replace has been superseded with lift-and-shift. Townsend’s posting on LinkedIn attracted another interesting comment that shone additional light on this mindset:

For existing applications lift and shift is where it starts but it must not end there. Having done several, I mean “several” big and small cloud migrations after 100+ datacenter audits, I can attest the fact with much confidence that lift and shift is where you should start when it comes to moving existing workloads to cloud.

Never try to do any type of transformation during the move and doing it prior to move is a moot point anyways. So workload transformation should occur only after migration to cloud … people make shortcuts and those shortcuts can cause performance issues as environments stay hybrid for quite a long time (perhaps forever) for most enterprises.

This is more or less consistent with my own observations through the decades. Firstly it was the big push in favor of distributed computing not forgetting our brief dabbling in time-sharing; whatever happened to service bureaus? Then followed client/server, not forgetting either the terrible turmoil that came first with the rush to move to an object-oriented world; anyone still relying on Object Request Brokers (ORBs)? All the while we were left to deal with database offerings that kept us on the back foot; anyone still running S2000 or Total or even IDMS? And now, it is cloud computing; are you preparing for what will inevitably follow?

Perhaps it’s one of the more recent comments to Townsend’s post that may be of greater interest to us all. Jason Benedicic who bills himself as a Cloud / DevOps / Automation Consultant suggested that there is an underlying reason why we still have 70 percent of our apps and data still running on-prem, on traditional systems: 

It really is harder than it looks, in so many ways. Once you get into the weeds of your own systems and start trying to break it down into manageable pieces to migrate or work on you almost certainly find something unexpected.

Building fresh in the cloud is relatively easy, and as was said lift and shift, while hated and expensive is relatively easy. Anything else is complicated and will either end up being far more costly than you anticipated or take years beyond what you’ve planned for in project timescales.

The NonStop community is widely read and stays abreast of change. In times where DX, Virtualization, Containers and even Hybrid IT have become talking points for many NonStop users, perhaps the key message here is to proceed with caution. Mission critical applications are called mission critical for a reason; the enterprise depends upon them to stay in business. For some NonStop users the opportunity to lift and shift some mission critical applications from traditional NonStop systems to virtual NonStop deployed within a private cloud may represent the first and indeed manageable next step to take.

Not for the purpose of saving money, but rather, for more intangible reasons like already having the hardware in place and the skillsets onboard to manage virtual machines. For these enterprises then the lift and shift philosophy makes sense. What would be tragic to read in the future is of those enterprises even with the presence of NonStop attempt wholesale rip-and-replace on a scale that simply isn’t manageable. For yes, it is a lot harder than it looks!

As I look to the distant horizon clouds may be hard to see as today smoke haze covers the sky. But perhaps what is worth considering is that clouds are transient at best and as plans are drawn up, NonStop users do need to be cognizant that, in time, these clouds moving into sight may indeed be fleeting as they too will pass. NonStop continues to demonstrate a resilience that oftentimes defies common knowledge but then again, for the NonStop community, this isn’t hard to come to terms with; not at all!


Monday, August 17, 2020

You cannot teach anyone to go fast! You can teach developers to NonStop …

With the Indy 500 looming, leave it to legend Mario Andretti to talk about an up and comer we can relate to in the NonStop marketplace …

It seems only fitting that at this time of year we include a reference to the upcoming Indy 500 race. Marketed as the greatest racing spectacle on earth, that despite contrary arguments from my colleagues in Europe, has challenged even the best of Formula One drivers through the ages. It looks so simple but even as someone who has pulled on a helmet, strapped himself into a “race-able car,” driven at 140 mph onto the banking of the Auto Club Speedway (Fontana, CA), I cannot comprehend what it would take to enter turn one of the Indianapolis Speedway at 240 mph as was the case this past weekend during qualifying.

Every country has its racing heroes and there is none bigger in America than Mario Andretti. It was almost two years ago that I had the opportunity to sit in the “second seat” behind Mario in a Honda-powered Indy car for an inaugural two-lap “session” at Sonoma just as qualifying for that Indy event in 2018 wound down. This was the subject of a post to Margo and my social blog that I wrote back on September 20, 2018, That “flying fickle finger of fate” together with a little serendipity and a whole lot of good fortune! 

When the magazine Robb Report turned to Mario Andretti and asked the question, “Who’s the next you?” he responded not with his grandson Marco Andretti, but rather, “Up-and-comer Colton Herta.” For many years Margo and I have followed the career of Colton as he is the grandson of our good friends in Simi Valley, CA, the Kennys. Colton is the grandson of Brian and Jan Kenny even as Bryan Herta is their son-in-law. So we consider them as part of our extended family! According to Mario:

“I see myself in him for several reasons. Number one, in his very first season at the top level of IndyCars, he was the revelation as a rookie. He won two major races, including the last one of the season.”

I have often thought that it would be a good exercise to ask the NonStop community to name their next you. Recently, I published a document on the Total Cost of Ownership (TCO) of NonStop and I have to say thanks to a lot of folks who contributed to this research note which is now in the public domain. First published in the August 2020 issue of NonStop Insider, you can read it in full by following this hyperlink, The ABC’s of an Affordable Business Critical Compute

At a time when there are many discussions circulating about the next us – the “up-and–comer” community focused on NonStop and continuing to build and support new applications on NonStop – it was from a younger group of executives that I heard the following:  

UK-based TCM Managing Director, Daniel Craig, “There’s no longer a pressing need to learn NonStop, not when the option is there to 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.”

What is meant these days by “learning NonStop” and what are “the benefits of the NonStop platform?” And how often do we avail ourselves of those open to “sharing our knowledge / expertise?” It is often alluded to in our industry that you cannot teach NonStop. To program NonStop at its lowest level, down at the message system level, you have to approach it almost as a religious experience and if you haven’t been accepted into the fold then well, nobody can expect you to program NonStop, right? 

For the NonStop community at this time, it is Mario’s next response to Robb Report that many of us can identify with: 

“Here’s the thing: You cannot teach anyone to go fast. You either have that inside of your belly or you don’t. The only thing you can teach is to minimize mistakes. There are some good drivers, some great drivers, and then you have the racers.”

You cannot teach anyone NonStop … the only thing you can teach is to minimize mistakes! Ouch; but is this true? Is there still “passion for NonStop” present in our community? Are we sure we cannot introduce NonStop to the “up-and-comer” generation in a manner that ensures the underlying fault tolerance is leveraged by solutions that truly don’t fail? Just as importantly, do we even program NonStop anymore?

For many years now we have watched frameworks appear that have made the task a lot simpler. Pathway (now TS/MP) was certainly a great starting point when it came to developing new solutions for processing transactions. Then we saw support for Java Virtual Machine (JVM) added as a Pathway application. As we now recognize the JVM for what it really is, a container, it opened the doors to solutions being written in Java and today, with a couple more tweaks coming, Java rocks on NonStop!

But wait, there’s more! If TS/MP and JVM improved the flexibility at the API, micro-services level, then much deeper down the software stack, the NonStop team revolutionized NonStop by giving it the ability to run virtual. This is ground that has been covered for some time, but when any discussion about Digital Transformation (DX) arises, then there’s no reason at all to discount NonStop. Yes, run NonStop on the hardware you have. Run NonStop as part of your private cloud.

We don’t program to the metal or to the message system but rather, we leverage the work done by a lot of people to make it easy to bring applications to NonStop. Yes, we are NonStop developers and yes, we really get it, thank you Mario. If you weren’t sure of this, just look at the fun times that are being had by those in the Under 40 SIG and to the level of enthusiasm for NonStop they exhibit whenever they meet. If you want more good news as a NonStop developer then you just have to read the blog post of March 9, 2020, now on the HPE web site, Modernizing the development world of NonStop applications where the topic is DevOps and covers the many tools to ease the development cycle that are supported by NonStop: 

What may also surprise CIOs is how DevOps tools like Git and Jenkins can be used to develop applications even as NonStop developers directly interface with products like Ansible. What may surprise IT management even more is that NonStop engineers actively use GitHUB, Jenkins, Ansible and other open source tools for DevOps to deliver many new offerings for the NonStop platform. 

And there is more to come. A new Ansible-based NonStop Manageability Framework (NSMF) is currently being created to help customers configure and deploy NonStop as well as manage HPE supplied software products on NonStop. 

There is no discounting the challenges the NonStop community is facing in these times where keeping the conversation going for NonStop seems difficult at times. But here’s the thing. Are we looking at the problem the wrong way around? Surely, we are no longer wanting NonStop programmers but we do want programmers who want NonStop. In other words, let them bring their favorite development environments with them and view NonStop as just another target server environment – one in tune with the requirements of running real time mission critical solutions.

NonStop programmers may be thin on the ground even as they are surely thinning on top. And graying! But that’s not the real problem today. It is so easy to get the next generation fired up about NonStop and what it can provide leaving us only to muse on how best to communicate this message. Can you hear me now? Good programmers are a breed apart where all we need to teach them is how to minimize mistakes. And yes, they do get it.

When it comes to the NonStop community, the companies willing to support development and deployment and who are willing to share their knowledge are out there, at hand and ready to help you. It’s no accident surely that this sector of the NonStop marketplace is growing. It’s also no accident that it’s leaders like TCM’s Daniel Craig who are actively engaged in the Under 40 SIG. 

Those original thirteen coders of the original NonStop software stack were brilliant young talent with leaders prepared to support them all the way. Not only support them but with the ability indeed guile to push them to where they achieved amazing outcomes. It was through their efforts that so much of the NonStop software stack is approachable at levels that make it a lot easier to program. 

With little reservation, those who develop today’s games are the same developers who can roll-out your next killer app on NonStop. Or as one NonStop enthusiast noted, "the current generation of 'Kernel Kiddies' empowered by management could continue to deliver" results similar to that initial group of thirteen. With NonStop part of HPC and with HP CEO Antonio Neri's recognition that "we have a transnational business which is very large," perhaps such empowerment is highly likely. 

Mario’s closing words to the Robb Report still resonate with me as they should with us all: 

“The racer is the one who really gets it. And (Colton) is a racer.

 “For me to mention that I see myself in him? It makes me look good.”

Shouldn’t we all be encouraging / mentoring the up and comer developers? After all, shouldn’t we all like Mario be looking to make ourselves look good, too?

Monday, August 10, 2020

Fields of grain; with a side of mountains!

A drive alongside of Colorado’s front ranges had me thinking more of hybrid IT and of possibilities for greater participation of NonStop in what is yet to come …

With some ability to get around, but still with the need to wear masks entering any state or commercial building, we have taken plenty of trips of late between Ft Collins and Boulder. However, it was something Margo and I talked about, as we drove to Boulder, that captured my imagination and for many reasons – the cornfields were high and behind them were the Rockies. Look, “cornfields; with a side of mountains!” Who wouldn’t want to take in this view every chance they had? 

Given my imagination all I could think of were "clouds; with a side of NonStop!" Just as the scenery would have been incomplete without the mountains to add depth and contrast, so too it is when we cover the topic of clouds without talking about NonStop! No hybrid IT deployment can realistically be considered complete without the presence of NonStop. Too strong? Mission Critical solutions are an integral part of hybrid IT and any argument about what system should be front-ending the arrival of mission critical transactions can’t be complete without a reference to NonStop. Right? 

And when it comes to our view of what the future might be like and what industries will be thriving as we crawl out from underneath this scourge that is COVID-19, there will be this contrast between what we see ahead of us and what might be helping out, on the sidelines as it were. Conversations about NonStop have never been about a homogenous NonStop presence, although there have been exceptions in the past, but rather, it’s about the business benefits that come from having NonStop supporting mission critical applications or even as the 24 x 7 database server. 

However, there really isn’t any way to sidestep concerns many of us share with regards to whether the mantra “good enough” will suffice for most enterprises. Nobody’s phones stay connected all the time and in today’s working environment where our workplace is carved from available space in our home, laptop / desktop environments aren’t all that reliable. Just this morning as I sat down I noticed that yet another Microsoft update had happened while I slept. Yet, even with all of this in mind, there are markets where NonStop excels and as HPE ramps up its promotion of its growing transactional business, there is plenty of room for growth for NonStop. 

In this year’s HPE Discover Virtual Experience, HPE CEO Antonio Neri touched on this topic perhaps to the surprise of some. When very large vendors like HPE talk about new business directions, oftentimes tech communities see this as an all or nothing break from past practices. Given how much emphasis HPE placed on Edge to Core delivered as a service and backed this up with considerable emphasis on the pay-as-you-go model that HPE has branded GreenLake, you would expect that the full weight of HPE marketing has been energized to support this brand to the exclusion of all else.

However, not so fast! On reflection of Neri’s boldest statements made during his keynote address, CRN reporter Steven Burke was quick to note that HPE was more in touch with reality than perhaps Neri was being given credit for honing in on one of his observations -   
 

“It is not just about GreenLake because we have a transactional business which is very large and we are pivoting to as-a-service, which is the long-term future. We need to be able to do both and be able to give partners the flexibility to come along.” 

We need to do both and be able to give partners the flexibility to come along to which we can easily add the need to give enterprise customers a choice as well. However, herein lays the conundrum for the NonStop community. The days of operating a proprietary solution in isolation, silo-ed to some extent, are over. The reason for this is that as HPE sees things, we are entering the age of insight and insights can only be derived from data; fresh data, as it’s created and NonStop as we all know, is where much of this fresh data is created. So integration with where data is being assembled and where analytics and AI live is important for the enterprise.

Connectivity with the outside world has always been an important attribute of NonStop in the past but today, much more is needed. It’s not just about connecting and moving data so much as it is about tapping into applications in a way that they too may be on the move. When there is as much talk about containers and the orchestration of containers as there is of late, attention needs to be paid to the implications for NonStop. I have been proponent of virtual machines in support of the transactional business NonStop has supported preferring to discuss hypervisors than I am discussing the merits of containers and yet, I cannot ignore this movement to a world of containers.
 

This would be all well and good if NonStop was nothing more than a Linux – like distribution, but then again, there is a reason why NonStop isn’t just Linux dressed-up for the occasion. Simply put, at its most basic level, mission critical applications running on NonStop and subject to all sorts of mandates and oversights wouldn’t be valued as highly as they are today if NonStop was subject to patches, fixes and reboots every couple of days. However, the argument persists. In a data center that is increasingly hybrid and where private clouds are appearing, participation is now trumping simple connectivity.

One of the points emphasized by Neri during his keynote was the importance of bringing the cloud experience into the data center. Recognizing that only 30% of apps and data have completed the journey to public clouds, there is value in having the other 70% residing in the data center take on the appearance of being cloud-like. Of course, GreenLake is a part of HPE’s answer along with Ezmeral, but there is more. According to Neri as reported by CRN’s Burke: 

“On top of that, HPE is bringing the full GreenLake Cloud experience to HPE.com, where customers can price and try the various services - from disaster recovery/backup to private cloud and container as a service.” 

Followed by one more bold statement: 

“With HPE GreenLake and our container services based on Kubernetes, you are not locked in. You can move those workloads and data to any public cloud if that makes sense for you.”   

Again, it’s worth repeating that the context here is GreenLake and already, we have noted that transactional business is different even as HPE needs to do both. However, it’s clear to all analysts following HPE and NonStop that at some point, NonStop would benefit greatly from playing nicely with the world of containers and with NonStop applications delivered as a service. But does this imply that at some point, there is need for NonStop to support Kubernetes? 

Maybe? Maybe not? This was the topic of a recent exchange on the LinkedIn group, Tandem User Group. If you didn’t catch it you may want to check it out and if as yet you aren’t a member of this group, then perhaps it’s time to join. We talk about NonStop running as a virtual machine but within the comments that followed, NonStop architect Keith Moore asked the very pertinent question:

“What if we could have an application or data instance of NonStop under Kubernetes? Is that of interest? I (we) need to know if this is a customer need. Please do-tell, folks.”            

A strong point of the NonStop team is that they prefer to pursue development projects after input from customers. It’s still too early to know for sure whether support for Kubernetes in one way or the other makes sense but given the emphasis HPE has placed on Kubernetes, it’s something the NonStop community needs to discuss. It might lead nowhere but again, it’s probably not a good idea to ignore the topic. Kubernetes at its simplest is an open source container orchestration enabler with an API “that controls how and where those containers will run.” Kubernetes recognizes that (indeed is in response to) today’s applications may “grow to span multiple containers deployed across multiple servers.” NonStop included, in time.                         

Our drive back along Colorado’s front ranges had us passing through many more fields as we continued to glance sideways at the mountains. These fields of grain come and go but those mountains, well, they have been with us for eons. So it is with NonStop whose presence within IT might be measurable in eons. We have seen trends and fads develop, flourish and eventually fade but transactional business continues to demand NonStop. On the other hand, as part of the HPE community can we truly ignore the potential that would come from clouds, with a side of NonStop?  


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...