Head in the Clouds: SaaS, PaaS, and Cloud Strategy

July 8, 2015  7:13 PM

Computer glitch? Admit it, there’s no such thing

Joel Shore Joel Shore Profile: Joel Shore
Application development, Error handling, Quality assurance, Software QA

Update on July 9: Reuters is reporting that yesterday’s New York Stock Exchange computer outage was caused by a software update. According to Reuters, a spokeswoman for the NYSE said the root cause was a “configuration issue.”

Original blog post on July 8: Today was a rough one for enterprise IT. The New York Stock Exchange came to a screeching halt. United Airlines was directed by the FAA to halt all flights systemwide. And the Wall Street Journal’s website was down for much of the day. Suffice it to say that Wednesday, July 8, 2015 will not soon be forgotten.

United’s woes were blamed on a bad router. We’ll save the discussion of hardware and communications redundancy for another day. As of this writing, no reason for the Journal’s outage had been made public.

And then there is the NYSE. Turn on any news program or read about it online and you’ll learn what happened: a computer glitch. I can’t think of anything more annoying. Or more wrong. Let’s be very clear about this one thing: There is no such thing as a computer glitch.

Systems go down for only three reasons. First is hardware failure. We all understand that. Routers fail, servers fail, a drive crashes, the power goes out, squirrels chew through power or communications lines, lightning strikes, or something else. Redundancy should reduce the vast majority of incidents to a momentary blip during a cutover.

Second is sabotage, hacking, denial-of-service attacks, or some other deliberate action. No further explanation needed.

Third is that glitch. The problem is that glitches don’t exist. They never have. And the general news media doesn’t understand that. Computers or devices are very stupid, they can do only what they are instructed to do. You and I have a word for those instructions — software. If the software is poorly written, fails to test for all conditions, has security holes, contains faulty logic, or was installed or configured incorrectly, the program and whatever hardware it’s running on may behave in an unexpected manner or produce undesirable results. I don’t want to say “wrong” or “erroneous” results, because the software is functioning exactly as written.

It doesn’t have to be servers or networks. My DLSR cameras have been known occasionally to perform in a way that defies explanation. But, then along comes a firmware update, and all is well. Until the next time.

What’s your take on glitches? Share your opinion; we’d like to hear from you.


July 8, 2015  9:16 AM

Choose integration tools with care

Joel Shore Joel Shore Profile: Joel Shore
Application integration, Cloud integration, Data integration, Development tools

No two IT infrastructures are alike. We all know it’s true. Tiny differences in applications, configuration, or patch status pretty much guarantee that no two “identical” servers can ever be exactly alike. Similarly, no two integration projects are alike. Choosing the right tools for integration projects varies widely, even for different corporations that run the same software. Choose carefully, says Forrester principal analyst Henry Peyret.

“The integration landscape is changing under cloud, mobile, and IoT,” he says. That means new and continually changing requirements inevitably lead to integration scenario complexities. What was true yesterday might be just a little bit different today. You already know that if you’re rolling out weekly updates to a mobile app.

“This complexifying integration landscape is challenging the existing integration strategic investments like ESB (enterprise service bus), EAI (enterprice application integration), and ETL (extract, transform, and load).” Even though “complexifying” wasn’t in my spell checker, what Peyret says is crystal clear.

It’s not just cloud, mobile, and IoT. It’s about which specific applications you’re trying to integrate, whether they reside on-premises or in the cloud, and which data sets they need to access and share. These are all complicating — or complexifying — factors. Add security and access considerations to the mix, along with data sovereignty requirements, governmental mandates, and corporate governance rules, and you can get in the middle of a serious challenge very quickly.

So what does this mean for integration tools makers? A lot. It means the tools have to adapt nearly as fast as your company does to ever-changing market conditions. Says Peyret, “The tooling market is facing multiple convergence between cloud and on-premises (creating hybrid integration), convergence of data, and event-based interfaces, such as batch and event in a single integration solution.”

The first consequence is that there are multiple new offerings available, but not all are adapted to support what Forrester characterizes as dynamic Integration. What do you need to do? “Companies should plan for tactical choices today which have the potential to become strategic investments, but they should not expect to get one single solution for all their integration scenarios,” says Peyret.

And there it is. No one tool is likely to meet all of your application- and data-integration requirements.

Have you found this to be true? Which tools did you choose and what consideration drove you to that decision? Share your thoughts; we’d like to hear from you.

June 26, 2015  12:20 PM

Red Hat’s new OpenShift Dedicated PaaS enters beta phase

Joel Shore Joel Shore Profile: Joel Shore
OpenShift, PaaS

At the Red Hat Summit, in addition to announcing that its OpenShift Enterprise 3 private PaaS is now in wide release, Red Hat took the wraps off of a beta version of OpenShift Dedicated, a totally new public cloud service based on the OpenShift 3 platform.

There’s no announced target date for release, but Red Hat is currently accepting online applications to participate in the tech preview.

The Red Hat team said that Dedicated uses the same code base as OpenShift Enterprise 3. The main difference is that OpenShift Dedicated is hosted on the public cloud and managed by Red Hat. In comparison, OpenShift Enterprise 3 subscriptions entitles businesses to host and manage the software in the infrastructure of their choice.

This service builds on the OpenShift Online public PaaS platform, but adds the ability for businesses to  build, launch, and host applications in the public cloud by offering enterprises a dedicated instance of OpenShift that’s managed by the OpenShift operations team. As the Red Hat folks put it, OpenShift Dedicated “brings the power and flexibility of OpenShift 3 to the managed public cloud.” It’s hosted on AWS.

June 26, 2015  12:01 PM

Red Hat’s OpenShift Enterprise 3 PaaS gets a formal launch

Joel Shore Joel Shore Profile: Joel Shore
OpenShift, PaaS, Red Hat, Web Application devlopment

Stand in any hallway during the Red Hat Summit in Boston this week, and you were likely to hear the c-word. Containers. And the d-word, too. Docker. Those two words seemed to get people more revved up than the energy drinks I saw being consumed everywhere, too. (The p-word, Python, wasn’t that far behind.) None of this is lost on Red Hat, of course. The company responded with its big announcement of general availability for OpenShift Enterprise 3, its enterprise-ready Web-scale container private PaaS. It’s based on Docker format Linux containers, Kubernetes orchestration, and Red Hat Enterprise Linux 7, providing full support from the operating system to application runtimes. (Version 7.2 of Enterprise Linux is not too far away.)

Of the three flavors Red Hat offers for OpenShift, this is the one that changes the least and is likely the logical choice for corporate cloud app development. OpenShift Origin, the free community PaaS is where changes and updates first get posted, and that can be dozens over the course of a month. Though it’s a great way to get started, it’s ultimately not where businesses will want to be. In the middle is OpenShift Online, operated and supported by Red Hat in the public cloud.

So what is Red Hat doing in OpenShift Enterprise 3? It delivers a container-based application platform based on Docker and powered by Red Hat Enterprise Linux. The idea is to provide a secure, efficient, and portable way to develop, deploy and run application services. OpenShift users also get access to a pretty big ecosystem of vetted, secure packaged application components, thanks to Red Hat’s Container Certification Program.

OpenShift Enterprise 3 includes the Kubernetes the open source, container orchestration and management engine developed with Google. What’s interesting is that with Red Hat as a contributor to both the Docker and Kubernetes open source projects, it’s eating its own dog food.

After sitting through more than a half-dozen one-hour sessions over two days, OpenShift was right at the top of those hallway conversations.

Shawn Zamacheck, a research developer at the Wharton School in Philadelphia is a fan, and a user. He uses OpenShift for creating research surveys. He looked at competing PaaS offerings, including Heroku. AWS didn’t have anything suitable a couple of years ago when he was looking around. He settled on Red Hat mainly because it was stable and supported a wide swatch of technologies.


June 26, 2015  8:49 AM

Dell, Red Hat team for new PaaS offering

Joel Shore Joel Shore Profile: Joel Shore
Cloud Services, Dell, PaaS, Red Hat

At the Red Summit this week in Boston, I had a chance to do an extended sit-down interview with Jim Ganthier, Dell’s vice president and general manager of engineered solutions and cloud. For a guy who works for what we all picture as primarily a hardware company, he had very little to say about hardware. Very little. These days, it’s about cloud, it’s about services, it’s about open source, and it’s definitely about partnerships with Red Hat. In the spotlight on this morning was Red Hat’s OpenShift PaaS.

For enterprise customers seeking a services-based cloud solution, Dell’s services organization has been working on a new solution for public cloud. It’s built on a Dell-branded OpenShift platform and aims to deliver deployment, migration and management of the operating system and applications via an OpenShift Online platform. Red Hat is not just a bystander to this; the company will manage this all-new PaaS environment. Ganthier told me this is the first Dell-branded PaaS platform designed specifically for Red Hat.

In an interesting side move, Dell Services is also joining the Red Hat Global System Integrator Program. That’s a program designed, to use Dell’s wording, “for partners that demonstrate leadership, unique capabilities and commercial relationships with global enterprise customers.”

It turns out that Red Hat and Dell go back a long way together: Dell was the first computer maker to pre-install Red Hat Linux.

Ganthier had plenty to say on other topics, too. He’s very hot on the idea of “cloud repatriation.” We’ll examine that in another blog post.


June 22, 2015  10:32 AM

HP keeps the tools coming; this time for continuous testing

Joel Shore Joel Shore Profile: Joel Shore
Automated Testing, Continuous delivery, Continuous development, Quality assurance

It’s becoming apparent that 2015 is a year in which we’re being showered with cloud and mobile development tools from all sides. It seems new tools are being announced weekly. One of the companies that has become quite aggressive is HP. That’s a really good thing to see. I hadn’t been sure what HP wanted to be, but the company is clearly going back to its roots as a supporter of developers and development technology.

Following other tools announcements earlier this year, the company in June launched HP LeanFT,  a new functional test automation solution. It allows software developers and testers to leverage continuous testing and continuous delivery. This means you can rapidly build, test, and deliver secure applications. This is exactly the sort of thing that is increasingly important as organizations strive to achieve faster time to market without sacrificing quality. While “fail fast and fix” does have its devotees and may be fine for non-critical consumer-oriented entertainment apps, it’s not going to embraced by developers of corporate and commercial apps that have no room for error.

Someone who had hands-on with LeanFT during its development is Jonathon Wright, director of testing and quality assurance at Hitachi. “With Agile and DevOps, there is a strong focus on continuous testing and continuous delivery. This means there is more emphasis on testing much earlier in the software development lifecycle,” he said.  He’s certainly right about the testing aspect. “HP LeanFT is built for continuous integration and continuous testing to help software developers and testers rapidly build, test, and deliver secure, high-quality applications.”

Well, sure, there’s a little marketing-speak in that last sentence, but Wright is corroborating the theme I’m hearing more often: continuous testing and continuous delivery rules the day. HP designed this to fit into existing ecosystems (such as Microsoft TFS, GIT, and Subversion) and frameworks that support test driven and behavior driven development.

To simplify testing, LeanFT provides project templates for standard unit-testing frameworks, including NUnit, MSTest and Junit. The idea is to improve collaboration and alignment between software developers and test automation engineers. And let’s face it, we’ve probably all seen situations where that collaboration can use a little boost. What HP is aiming for is a reduction in the time needed to test applications. That, in turn, should help developers to predict and identify defects earlier in the software development lifecycle.

HP LeanFT is currently scheduled for availability in July. HP Unified Functional Testing 12.5 and HP Business Process Testing 12.5 will also be available in July. It will be interesting to see what else HP has up its sleeve.

June 17, 2015  9:27 AM

Is mobile software really this bad?

Joel Shore Joel Shore Profile: Joel Shore
Mobile Application Development, Program Debugging

Every day, I check my iPhone to see which apps have been updated. Think of it as the latest fad in armchair spectator sports. What I see only reinforces my belief that mobile apps may look pretty, but are often scary bad under the hood.

One app’s “What’s New” section reports “corrected issue with removing previous purchase,” “improved reliability,” and “corrected other issues.” Another app reports “we may have exiled some bugs.” From the Google app, we get “bug fixes and performance improvements.” Panera Bread’s app tells me “various bug fixes.” The Starbuck’s app notes “this release contains bug fixes and enhancements throughout the app.” Shazam’s 8.6.1 update tells me “We know, we broke a few things in 8.6. For those of you who’ve been getting crashes on startup, update now and things should be back to normal.” The Walgreens app, version 5.2.2, lists specific bug fixes. Evernote’s June 12 version 7.7.7 update reports “numerous bug fixes and improvements.” Even the app that I use to measure connection speed says on its latest update, “this is primarily a bug fix release.” Was it reporting erroneous speeds?

The Facebook app notes that it delivers an update every two weeks and that “every update … includes improvements for speed and reliability.” You’ve got to wonder how much faster and more stable it can get. Or, if you’re a professional skeptic like me, you wonder how miserably slow and undependable it must have been 26 updates, or a year ago.

Apps that are downright ugly or feature unfriendly, convoluted user navigation are a different matter. We’ll save that for another blog post.

Some apps are indeed problematic enough that they merit being banished from my phone. There are apps that never got past their startup screens without crashing. There are apps there were so slow as to be incredibly annoying. And there is one app that insists I’m someone else. That’s especially disconcerting.

So what’s going on? Are mobile apps as bad as these messages suggest? Well, yes and no, according to people I talked to during the recent Cloud Computing Expo in New York. We all understand that software is never really finished. On my desktop PC, Microsoft’s “Patch Tuesday” provided updates for Windows XP for 13 years, until support finally ended in 2014. The same, apparently, is true for mobile apps, only that updates come more frequently. I’ve seen updates on one day followed by another on the very next day to repair something the first update simply broke.

Is this due to bad app design? Incomplete specs? Poor coding? Inadequate testing? A rush to get apps published? A conscious strategic effort to “fail fast” and fix? Or something else?

If you’re an app designer, developer writing code, or QA tester, you already have pretty strong feelings about this. Let’s hear them.

June 14, 2015  10:35 PM

Incorporating security into an app starts on day one

Joel Shore Joel Shore Profile: Joel Shore
Application development, Application security, cloud encryption, Data Encryption, FIPS, Mobile Application Development, Security compliance

I won a football at the New York Cloud Computing Expo in a session about encryption. Ask a meaty question and the speaker just might toss you a prize. My question was simple: If the only way to protect all the data is to encrypt all the data (except that which is already public), aren’t edge devices like tablets and smartphones going to suffer a potentially significant performance penalty since everything will need to be unencrypted before it can be used and then re-encrypted for transmission?

Well, yes, of course, said Ray Potter, CEO of SafeLogic, as he heaved a small toy inflatable red football emblazoned with the SafeLogic logo my way. Encrypt. Just get over it. Just do it. Don’t encrypt and you’ve got a good chance of becoming front page news. Right, U.S. Government? Right, Sony? And while you’ve got encryption on your mind, understand its three pillars: confidentiality, integrity, and availability.

Confidentiality, the idea that information should not be disclosed to people who shouldn’t have it. This is the idea of encrypting data and making it unreadable to attackers. Integrity, as Potter puts it is “making sure that data isn’t mucked with in transit.” A use case might be financial transfers where you need to make sure that account numbers and amounts are not compromised.

Third is availability, which might not initially appear to be related to security. But, it is. “We see a lot of distributed denial-of-service attacks that take down a server or an entire site,” Potter says. Think about Amazon going down. That’s a major loss of both revenue and goodwill. Yes, availability is security.

Regarding encryption, Potter says it fits everywhere, even on your spiffy new smartwatch. What each business needs to undertake is a risk-management plan, figuring out where data lies in the system, classifying data in terms of its sensitivity or the assets it protects, and then assigning weighted metrics. What’s typically revealed is that encryption, which was once done at the device level, now needs to be everywhere. What it boils down to is that data needs to be encrypted both in transit and at rest. Otherwise, you could be the next Sony Pictures or Anthem Health Insurance.

When it comes to encryption standards, there’s really only one that matters, and that is FIPS 140-2. The Federal Information Processing Standard (FIPS) Publication 140-2 is a cryptographic standard maintained by the Computer Security Division of the U.S. National Institute for Standards and Technology. It encompasses four distinct security levels, but for most commercial (read that as non-governmental) applications, level one is usually sufficient.

Even though FIPS 140-2 is the standard you need to implement when building your applications, usually followed by compliance verification and certification testing for certain environments (governmental and healthcare, for example), there is one gaping hole. The current version, enacted on Nov. 15, 2001, predates mobility — the existence of tablets by a decade and the mainstream advent of mobile phones by many years.

If you’re a business that’s signing up for anything-as-a-service, part of your due diligence is a demand to see the provider’s FIPS compliance certificate. And you also need to know on what exact platforms compliance has been tested — and which platforms haven’t been tested.


June 12, 2015  1:43 PM

Monetization: It may be the real reason we built the cloud

Joel Shore Joel Shore Profile: Joel Shore
Business strategy, recurring revenue

At the Cloud Computing Expo in New York (June 9-11), there was plenty of talk about Docker, APIs, everything-as-a-service, and all manner of things technical. Yet, there was an interesting simultaneous shift in some of the conversation toward leveraging the cloud to conduct business. And that’s a good thing.

After all, you’re not in business to sell shoes, you’re in business to make money. Everyone from the CEO down to the kid in the corner cranking out lines of Python code had better understand that. If there’s one word that dominated these discussions, it’s “monetization.” And montenization is about services, not devices. Your mobile phone company really has only three things it can ever sell you: voice, text, and data. The specific devices are irrelevant. Monetization is about services.

Brendan O’Brien, chief evangelist and co-founder of San Francisco-based Aria Systems, is all about monetization, which he describes as recurring, pay-per-use revenue streams, more so than one-time transactions. Here’s the perfect example: Medical MRI machines cost millions of dollars to buy. That means only large hospitals located in major metropolitan areas can afford them. The problem, of course, is that there is a finite number of these large hospitals. And they’ve already bought their machines. The market is saturated, right? Or is it?

Drill down to the next tier and you’ll find thousands of regional hospitals, 10 for each major city hospital, according to O’Brien. They need that MRI machine, but can’t afford the enormous up-front purchase price. What they can afford is a pay-per-use scheme. And that’s exactly what this major equipment maker is doing. It’s installing machines around the world in these second-tier hospitals with Aria Systems handling the monetization aspect (which includes billing and remittance, but is so much more than that). The hospitals get their gear, the equipment maker has increased its market tenfold, and Aria handles the monetary flow, essentially an annuity. It’s a recurring revenue model.

Other examples of monetization were noted at the conference. On your car’s next set of tires, you may pay per use based on mileage instead of shelling out nearly a thousand dollars as the up-front purchase price. It’s coming. And then there’s that shoe company. What if you could get yourself the best (and likely outrageously expensive) running shoes available if you could pay for them based on a usage model that’s tied to the number of steps you take or miles you run. Don’t ponder the question too long. It’s coming, too.

Note that we’re not talking pretty cloud apps that run on a smartphone or tablet. These are the industrial cloud applications that you can’t see, the ones that run inside an MRI machine or deep within your car. These are the cloud apps of business.

June 4, 2015  9:13 AM

Cloud app functionality is not enough

Joel Shore Joel Shore Profile: Joel Shore
Analytical tools, Application analytics, Applications development, Cloud Applications, Mobile Application Development, User experience

You’ve got a huge portfolio of development tools for building and testing cloud applications. And you’re probably running analytics against that mountain of big data you’ve kept either on-premises or moved to cloud-storage, maybe because you’re not sure which is the better location. You can even see how users navigate through your mobile apps to see where the bottlenecks are and at what point they give up and abandon their shopping cart. But what about measuring the performance of your digital assets? Understanding that is part of delivering a good user experience.

Tools vendor Keynote is giving it a try. Its Keynote Digital Performance Intelligence product sets out to benchmark third-party services and competitive performance against an organization’s digital assets. The idea is to allow IT and business users to collaborate and get a handle on performance proactively instead of reactively. The company says the cloud-based analytics suite provides organizations with real-time insight into the performance of digital assets. That should help them improve customer experiences, which, in turn, makes corporate decision-making easier.

So, what’s really going on? Analyst Raul Castañon-Martinez of 451 Research says Keynote is integrating information from different digital channels. “Everyone talks about this but they are one of the few giving you a holistic view of how you are performing.” Castañon-Martinez says he sees a trend toward realtime analytics and tighter integration with the development function. “There’s a lot of noise in the industry and it’s hard to figure out what companies are doing and delivering. Keynote is not making much noise, but in thought leadership it stands out,” he says.

I like that idea; tighter integration between realtime analytics and development. It’s good for developers to understand how well their app functions in the wild.

Jennifer Tejada, Keynote CEO is pretty blunt about the way things work in the real world. “As organizations look to engage consumers through digital and mobile offerings, they often overlook the performance impact.” That’s probably true, but as developers’ mobile experience deepens, I expect that will lessen. Then she gets to the crux of the matter: “Research has shown that customer tolerance for failure on digital channels is significantly decreasing.”

Tejada is right about that. Attention spans seem to grow shorter with each passing decade. As a developer, what are you doing to build a user experience that makes sense, that’s easily navigable, and that won’t bomb out halfway through a shopping spree, entering sales data, or something else? It’s simply not good enough to build applications that function correctly, it’s necessary to build apps that are enjoyable and engaging to use. Because, if you don’t your competitors will.

Forgot Password

No problem! Submit your e-mail address below. We'll send you an e-mail containing your password.

Your password has been sent to: