Clouds
are coming and for NonStop it’s inevitable. One likely cause? The need to tap
data from IoT – but be careful, clouds can burst and rain on our parade.
As I walked into the kitchen the other morning to pour
the first cup of coffee of the day I happened to look out the window. The view
of the mountains to the west caught me by surprise with a vista of a bright orange
glow edging the clouds – it may not look all that impressive, but now I know
why the Denver Bronco football team wears blue and orange. I am rarely the
first to wake up in our household and it always takes me a while to get going
but on this occasion, as I poured the coffee into my mug, I took just a little
longer to take it all in – breathtaking, indeed. I often wonder how many times
circumstances catch us out. Surprise us, perhaps, and even on occasion, simply
take our breath away. Being IT professionals it takes a lot to really surprise us but on those occasions when we are surprised it’s rarely because something went wonderfully well. However, when it comes to clouds and IT, it’s not the hues that matter as much as the ahas! With the advent of cloud computing it’s almost as if we are quite prepared to ignore all the experiences we have accumulated over decades – let’s just leave it to the cloud provider to ensure we have access whenever we need it. The cloud is all rather nebulous anyway and could be anywhere – surely, it’s all virtualized to where it will find a path to a working server / comms / storage component no matter what. And to that I can add, that swampland I bought in Florida is begging to look very promising indeed. Do you have a couple of bucks and I will sell it to you, right?
Coffee and clouds or probably more specifically, clouds
in my coffee. As Carly Simon used to sing all those year ago:
I had some dreams, they were clouds in my coffee …
I had some dreams, they were clouds in my coffee …
Yet, here we go again; dreaming! It takes very little to upset it all and
suddenly it’s simply vapor. Nothing. Clouds in our coffee! What prompts these
musings is my acknowledgment that without taking the necessary precautions,
clouds can be as temperamental as any system and the level of availability as
fleeting as any commodity server. And this brings me back to what transpired at
Amazon last week. Their much-vaunted cloud offering, Amazon Web Services (AWS),
went AWOL. Just like that, access to files and databases – anything that was
pushed to storage – went dark. Literally hundreds of thousands of companies
large and small no longer had working, operational, systems in support of their
business.
If you missed reading about the AWS outage, it may be worth the time to check out the story in USA Today where the outage is addressed in mostly non-technical manner. In the February 28, 2017, article, Massive Amazon cloud service outage disrupts sites reporter, Elizabeth Weise, opens with the observation that the outage ‘didn't quite break the Internet, but a 4-hour outage at Amazon's AWS cloud computing division caused headaches for hundreds of thousands of websites across the United States.” Furthermore Weise quotes Dave Bartoletti, a cloud analyst with Forrester, who said, "This is a pretty big outage. AWS had not had a lot of outages and when they happen, they're famous. People still talk about the one in September of 2015 that lasted five hours.”
A few days ago I included the above as part of a weekly email that I provide my clients. I write this email to keep my clients alerted to events that perhaps otherwise they may have missed, but in this case, I don’t believe anyone missed such a newsworthy occurrence at Amazon. However, it was what I also included in that weekly client email that perhaps triggers more memories among the NonStop community. I jumped to the end of the USA Today article to something that the NonStop community could really take to heart.
“Companies have been steadily moving storage to the cloud because it is cheaper, easily accessible and more resilient. But the downside is that when there are problems, there's a cascade effect. It's possible to contract with multiple companies to avoid potential problems, but that strategy is pricey, so many companies make peace with the knowledge that on rare occasions they're going to have a very bad day. ‘Only the most paranoid, and very large companies, distribute their files across not just AWS but also Microsoft and Google, and replicate them geographically across regions - but that's very, very expensive, Gartner's (cloud analyst, Lydia) Leong said.”
Only the most paranoid? Is that us? USA Today writes that it is only the most paranoid and yes, those very large companies (enterprises) we know well that distribute files across multiple clouds (and replicate) to make sure an individual cloud outage doesn’t impact their business, are the exception. Paranoid? Really? And yet, this has been the story of NonStop for decades – loosely coupled, shared nothing, massively parallel processing. Again, are we now to be considered paranoid because we bought a system that survives single points of failure? Rather humorous when you consider what is being implied, don’t you think? How many times of late have we proffered the advantages inherent with NonStop to senior management and business unit managers only to be looked upon as having two heads but then again, two heads? That’s another story, obviously. On the other hand, maybe you say paranoid like it’s a bad thing!
The innuendo from such discussions, when they occur, is that we are caught up in a legacy model that no longer applies to modern IT. Modern data centers, complete with redundant everything, no longer take outages or so we are told. They are reliable and there’s simply nothing to worry about – any additional costs to do something better is simply not worth the money. Well then, tell that to Amazon. Tell that to any of the airlines too who have seen cascading failures of their reservation systems. Tell that to some of the biggest stock exchanges as well.
However, not lost on anyone is the inevitability of cloud computing and of the inroads it’s already making into enterprise data centers. Staring out at the morning clouds, the aroma of fresh coffee unmistakable, I began to wonder; it may not be the applications we already run but brand new applications that we implement to leverage clouds. And for the NonStop community this may very well be the result of looking at Industrial IoT (IIoT). That’s not to say transaction processing is going away any time or that the ability to process transactions in real time is going to lessen in importance but rather a simple acknowledgment that more may be happening at the end points than simply the transaction.
If you missed reading about the AWS outage, it may be worth the time to check out the story in USA Today where the outage is addressed in mostly non-technical manner. In the February 28, 2017, article, Massive Amazon cloud service outage disrupts sites reporter, Elizabeth Weise, opens with the observation that the outage ‘didn't quite break the Internet, but a 4-hour outage at Amazon's AWS cloud computing division caused headaches for hundreds of thousands of websites across the United States.” Furthermore Weise quotes Dave Bartoletti, a cloud analyst with Forrester, who said, "This is a pretty big outage. AWS had not had a lot of outages and when they happen, they're famous. People still talk about the one in September of 2015 that lasted five hours.”
A few days ago I included the above as part of a weekly email that I provide my clients. I write this email to keep my clients alerted to events that perhaps otherwise they may have missed, but in this case, I don’t believe anyone missed such a newsworthy occurrence at Amazon. However, it was what I also included in that weekly client email that perhaps triggers more memories among the NonStop community. I jumped to the end of the USA Today article to something that the NonStop community could really take to heart.
“Companies have been steadily moving storage to the cloud because it is cheaper, easily accessible and more resilient. But the downside is that when there are problems, there's a cascade effect. It's possible to contract with multiple companies to avoid potential problems, but that strategy is pricey, so many companies make peace with the knowledge that on rare occasions they're going to have a very bad day. ‘Only the most paranoid, and very large companies, distribute their files across not just AWS but also Microsoft and Google, and replicate them geographically across regions - but that's very, very expensive, Gartner's (cloud analyst, Lydia) Leong said.”
Only the most paranoid? Is that us? USA Today writes that it is only the most paranoid and yes, those very large companies (enterprises) we know well that distribute files across multiple clouds (and replicate) to make sure an individual cloud outage doesn’t impact their business, are the exception. Paranoid? Really? And yet, this has been the story of NonStop for decades – loosely coupled, shared nothing, massively parallel processing. Again, are we now to be considered paranoid because we bought a system that survives single points of failure? Rather humorous when you consider what is being implied, don’t you think? How many times of late have we proffered the advantages inherent with NonStop to senior management and business unit managers only to be looked upon as having two heads but then again, two heads? That’s another story, obviously. On the other hand, maybe you say paranoid like it’s a bad thing!
The innuendo from such discussions, when they occur, is that we are caught up in a legacy model that no longer applies to modern IT. Modern data centers, complete with redundant everything, no longer take outages or so we are told. They are reliable and there’s simply nothing to worry about – any additional costs to do something better is simply not worth the money. Well then, tell that to Amazon. Tell that to any of the airlines too who have seen cascading failures of their reservation systems. Tell that to some of the biggest stock exchanges as well.
However, not lost on anyone is the inevitability of cloud computing and of the inroads it’s already making into enterprise data centers. Staring out at the morning clouds, the aroma of fresh coffee unmistakable, I began to wonder; it may not be the applications we already run but brand new applications that we implement to leverage clouds. And for the NonStop community this may very well be the result of looking at Industrial IoT (IIoT). That’s not to say transaction processing is going away any time or that the ability to process transactions in real time is going to lessen in importance but rather a simple acknowledgment that more may be happening at the end points than simply the transaction.
Dr. Tim Chou, the keynote
speaker at last year’s NonStop Technical Boot Camp, just posted to the Avnet
web site, Is
IoT the killer app for cloud computing? According to Tim, “the major wave
to using compute and storage cloud computing will not be driven by
migrating existing applications. Instead, as with the move to client/server,
the driver will be building completely new applications. So what are these
applications? And what will be the killer app for cloud computing? Perhaps
IoT.” This is something we often overlook in our discussions about NonStop and
clouds – while some of us are busy trying to shoehorn existing applications
into clouds perhaps we are missing the opportunity to come up with entirely new
applications.
Take the humble ATM for instance. With each new generation of the ATM, more sensors are being added. This trend will likely only continue to where ATMs become agents for tracking largescale demographic and other social-related data. One connection will be involved in an ATM transaction while other connections will be transporting multiple data streams. New applications may simply start out by correlating cash withdrawal activity to the weather, what’s showing at the cinema and which direction passers-by are headed. There’s no limit and it’s this hybrid world of what Tim calls the Internet of People juxtaposed with the Internet of Things where the action may lay in the coming years. And NonStop exhibits no signs that it will back away from support of either. So yes, start dreaming big!
So, what did happen at Amazon? What was responsible for creating a four hour outage? It has now been revealed that members of the AWS S3 storage team “were debugging the billing system and someone entered an extra command. After accidentally taking the servers offline, the various systems had to do ‘a full restart.’” Yet one more instance of “Operator Error!” Recalling those lines by Carly Simon, “I had some dreams, they were clouds in my coffee,” maybe there are many of us within the NonStop community simply dreaming after all, but then again, having systems that don’t fail running in support of People and Things makes sense and it’s becoming a common practice to accept outages then surely, new business opportunities abound for those NonStop vendors and users who dare to disagree. And yes, perhaps IoT will bring out the best in the new NonStop after all!
Take the humble ATM for instance. With each new generation of the ATM, more sensors are being added. This trend will likely only continue to where ATMs become agents for tracking largescale demographic and other social-related data. One connection will be involved in an ATM transaction while other connections will be transporting multiple data streams. New applications may simply start out by correlating cash withdrawal activity to the weather, what’s showing at the cinema and which direction passers-by are headed. There’s no limit and it’s this hybrid world of what Tim calls the Internet of People juxtaposed with the Internet of Things where the action may lay in the coming years. And NonStop exhibits no signs that it will back away from support of either. So yes, start dreaming big!
So, what did happen at Amazon? What was responsible for creating a four hour outage? It has now been revealed that members of the AWS S3 storage team “were debugging the billing system and someone entered an extra command. After accidentally taking the servers offline, the various systems had to do ‘a full restart.’” Yet one more instance of “Operator Error!” Recalling those lines by Carly Simon, “I had some dreams, they were clouds in my coffee,” maybe there are many of us within the NonStop community simply dreaming after all, but then again, having systems that don’t fail running in support of People and Things makes sense and it’s becoming a common practice to accept outages then surely, new business opportunities abound for those NonStop vendors and users who dare to disagree. And yes, perhaps IoT will bring out the best in the new NonStop after all!
Comments