“I’ve got a feeling and I just can’t let it go!” This is a line from a haunting song by Los Angeles indie electro-soul band, Caught a Ghost. Not to be confused with the song from Frozen, it became the theme song of Amazon Prime series “Bosch” that was based on numerous novels by Michael Connelly. For those familiar with the novels or those who have viewed the Prime series, you will know that the main character Bosch enjoys jazz and the choice of this song for the series title may have been a surprise and yet, it absolutely captures the indomitable spirit of Detective Bosch.
This is the same feeling I get whenever I hear news about the latest cloud offering or about the enterprise that has elected to migrate everything to the cloud. And this feeling is amplified whenever I read of news about yet another outage affecting millions of users. It is almost as if IT professionals have forgotten the first law of IT: Change. Call it cause and effect. Call it the swinging pendulum. Maybe think of it in terms of Newton’s third law: For every action, there is an equal and opposite reaction. It’s not surprising then that there are those who are beginning to question the validity of migration to clouds.
No matter what you might think of enterprises embracing clouds that IT constant of change suggests that something better will always come along and for a period, represents a return to stability. Just as importantly a return to reliability. As for those of us who watch these movements, then perhaps we are among those that understand the lines in that song Can’t let it go:
It isn’t easy to ignore the current conflict within IT that has us wrestling with staying traditional or going virtual. Of staying centralized or going distributed. Of CapEx or OpEx. And yes, of doing nothing or doing both. Welcome, as vendors like to say, to the hybrid world of IT where it is OK to be supporting both teams. However there is legitimacy in clouds that we as the NonStop community need to acknowledge.
According to AWS, “Cloud computing is the on-demand delivery of IT resources over the Internet with pay-as-you-go pricing.” And yet, as Computerworld in an HPE sponsored opinion column of May 31, 2019, noted,
“The business and IT needs of any organization are defined by many factors according to IDC, such as the organization’s specialization and business model, mix of workloads, financial health, maturity, workforce and customer base.”
Furthermore there are positive outcomes that come with the cloud experience. Simply stated, Computerworld says:
“The overall cloud experience has had far-reaching benefits on all groups within the enterprise.
“Additionally, the pay-as-you-go model used with cloud-based IT lets the enterprise better balance capital and expenses with corporate objectives.
“Even though the way applications are built and consumed is changing, IT teams can’t easily abandon existing business-critical workloads. After all, the smooth operations of the business depend on the reliable and predictable delivery of business-critical workloads.”
For the NonStop user, you just have to love this last sentence. After all Computerworld reports – and yes, it is a HPE paid-for infomercial after all – “business depends on the reliable and predictable delivery of business-critical workloads.” It is true that this article was written back in 2019 when little was known about GreenLake (and bringing the cloud experience to the enterprise) or the support of GreenLake by NonStop, it does highlight a potential opportunity for NonStop.
And yet there is, however, vulnerability with an “all-in on clouds” approach. Call it fragility or perhaps better still, a lessening of reliability to where outages are occurring with such frequency that enterprises are beginning to rethink their options. A number of British banks for instance have backed away from going all in with clouds. Recall the recent news coming out of Barclays and yes, there are other financial institutions following suit.
In my last post, What’s not broken and just keeps on running? NonStop delivers! I wrote of how “Two opportunities come to mind that in the coming months I will be exploring in more detail. And they have to do with how we think about NonStop going forward and whether our own ideas about the role of NonStop may indeed be outdated. Might there be the potential to have NonStop play a guardian role – no pun intended? Should there be a NonStop essentially overseeing the hybrid multi-cloud environment common today among enterprises so that exposure to any one cloud can be marginalized to where outages have no impact on the running of mission critical applications?
When we think about the value that NonStop brings to any discussion on hybrid IT and cloud computing given the above then perhaps it is best explained albeit with a goodly amount of humor by HPE Distinguished Technologist, Keith Moore. “HPE NonStop is a mature peer-group soft system of super-paranoid, autocratic, self-centered kernels working together but with communication rules for up to 4080 enemies, all working independently and together, watching themselves and all other kernels with universal distrust.”
When IBM first unveiled Parallel Sysplex for the mainframe – an overt attempt to upgrade the resilience of the mainframe when delivered as a cluster – they introduced the “coupling facility.” Initially a dedicated mainframe but later updated to run virtually. The coupling facility runs no application software and has no I/O devices – it is purely in place to oversee the key elements needed by mainframe applications to ensure their continued operation. Often referred to as “clocks, locks and lists” you couldn’t run a mainframe cluster with Parallel Sysplex without the coupling facility.
Fast forward to today; clouds and the possibility for NonStop playing a similar role. When an enterprise pursues a hybrid cloud deployments that include traditional systems, on-prem private clouds together with access to a number of public clouds, it’s almost impossible to guarantee levels of availability we set down as mandatory for our business-critical workloads. But how would this work? What would it entail?
This has been the subject of a number of conversations that are at best introductory in nature even as they continue to be ongoing. When it comes to how they might work the biggest hurdle is in building out the type of resilient and redundant high-speed interconnect fabrics. Forcing a take-over of a cloud determined to be misbehaving and offloading to another resource, be that another public cloud, or something on-prem demands some level of application awareness. Maybe not check-pointing but perhaps something akin to Pathmon with the resilience it provides may be all that needs to reside on our NonStop “guardian.” And then there is the modernization of applications taking place and the way we build new apps may indeed prove beneficial when it comes to easing into this approach to cloud resilience.
“There should be a balance within IT where successful innovations to the business migrate to the traditional side where efficiencies, standards and cost reductions are applied to that innovation. All truly successful innovations should become foundational IT systems. I’ve never seen that as a goal within agile development but a successful IT project should deliver business benefits for many years,” said HPE Master Technologist Justin Simonds. “Doing Agile, fast-fail new business applications in the cloud makes sense since they are quickly constructed and just as quickly deconstructed (if they fail). However I have yet to see a strategy for migration from the cloud, when a new application/service is a clear benefit to the business. When something runs 24x7x365 it is very expensive in the cloud. Applications that become critical to the business are most effectively, efficiently and inexpensively run in-house.”
All things are possible and I don’t think that the presence of a high-speed duplicated fabric will present an insurmountable problem – aren’t today’s WAN offerings already superior to what was present on early models of Tandem Computers? Isn’t Pathmon sort of already doing this (above) in a Virtual Machine? Haven’t NonStop vendors already taken some baby-steps towards achieving this, whether their approach was simply for disaster recovery or better load balancing? And the monitoring vendors would be well-positioned to provide real time analysis of where your apps and data were running for all those with regulatory issues needing to be addressed?
If you consider this a little too futuristic the only ingredient missing as a real world customer requirement and as it has always happened in the past, unless we take our introductory level conversations public, how will they know that something like this is possible? It certainly lends itself to a baby-steps approach, one app at a time. Maybe you do keep the database on NonStop after all and simply execute the apps anywhere you like under the watchful eye of your NonStop guardian?
As I began this post “I’ve got a feeling and I just can’t let it go!”