Disqus - Latest Comments for rjhintz

Re: Berkeley Mills furniture store is closing Friday after 37 years

Richard Hintz — Thu, 17 Jul 2025 12:05:44 -0000

Sad to read this. Gene used to write children's books, IIRC, and would sit with your kids and read them in the early years.

Re: Serverless COBOL: Rejuvenating legacy code with open source software — Part 1

Richard Hintz — Thu, 09 Sep 2021 13:24:43 -0000

Thanks. Any additional detail with examples similar to real world complexity would be great. I definitely will stay tuned.

Re: Serverless COBOL: Rejuvenating legacy code with open source software — Part 1

Richard Hintz — Thu, 29 Jul 2021 21:37:45 -0000

@Didier Durand (AWS) @

I guess my point is that the hello-world.cob program is far from even the most simplistic COBOL Read and Report, much less a Create, Update, or Delete function.

I'm a big believer in Lambda and microservices, but I don't see how to do the next step of extracting from a simple, not even prod COBOL program that reads from persistent store and writes a report with some simplistic math.

I appreciate the repo as is, and certainly don't expect a CICS replication, but it would be really useful for those of us interested to see the next, more complex iteration.

Also, how one does the transform from JCL execution management to this model would be really helpful. Thanks.

Re: Serverless COBOL: Rejuvenating legacy code with open source software — Part 1

Richard Hintz — Mon, 26 Jul 2021 18:21:25 -0000

I can see the hello.world toy program, but what would be really useful is development for simplistic COBOL + JCL streams that do batch and separately transactional CRUD operations.

JCL execution management replacement in this context isn't clear (to me).

Seeing CICS transactional management simulation would be cherry on top, but first things first.

Re: Memberkit 1.0: Upgrade your analytics and build a data-powered membership program - Source

Richard Hintz — Fri, 22 May 2020 09:00:03 -0000

Consider putting this as a GitHub repo. Easier to track issues and changes.

Re: How to Enroll in Global Entry on Arrival

Richard Hintz — Tue, 12 Jun 2018 23:28:55 -0000

"All that was needed for the interview, aside from already having conditional approval, was my passport."

The title is somewhat misleading. As the article text says, you already have to enrolled previously *and* received conditional approval, which can take time, often weeks.

What's referred to in this piece is "interview on arrival," not "enroll on arrival."

Re: What Oracle is Developing with Its Hyperscale Cloud

Richard Hintz — Sun, 11 Jun 2017 23:34:26 -0000

Interesting comment at around 14:05:
Oracle: Other IaaS providers built semantics into their offerings that require apps to be retooled and rewritten. (paraphrased)

So, Oracle isn't building their own semantics into their IaaS offering?

Re: 4 simple strategies for managing your AWS costs with reserved instances

Richard Hintz — Tue, 20 Dec 2016 13:09:52 -0000

Could you explain a little more, here or perhaps in a separate post, how you use Kubernetes to develop an optimized quantity of RIs?

Re: 4 simple strategies for managing your AWS costs with reserved instances

Richard Hintz — Fri, 16 Dec 2016 19:13:29 -0000

Do you have a guideline for the percentage of RIs you use for workloads? For example, AWS has said that 70% is a good number.

Also, given the rate that AWS is introducing new features, have you developed guidance for acquiring 3 year RIs?

Re: Taming Jenkins at Scale with CloudMunch Turbo

Richard Hintz — Fri, 23 Sep 2016 11:00:05 -0000

Thanks. At this point I'd not be the best early adopter because of other projects not getting attention. I appreciate the opportunity, though.

Re: Taming Jenkins at Scale with CloudMunch Turbo

Richard Hintz — Fri, 23 Sep 2016 10:39:39 -0000

Today? None, they're all dead. ;)

Thanks for an interesting article and being patient with my questions.

Re: Taming Jenkins at Scale with CloudMunch Turbo

Richard Hintz — Thu, 22 Sep 2016 20:58:51 -0000

I'm asking all these questions since HA Jenkins hasn't been exactly easy.

Have you tried killing the master, a la Chaos Monkey, and seeing if Jenkins recovers?

Re: Taming Jenkins at Scale with CloudMunch Turbo

Richard Hintz — Thu, 22 Sep 2016 18:21:00 -0000

Thanks. How do you maintain state for Jenkins functions. Are the data on the separate volumes replicas or some sort of consistent distributed datastore?

If state and data consistency management is just handled by Kubernetes native functions, I need to do some reading, so apologies if that's the case.

Re: Taming Jenkins at Scale with CloudMunch Turbo

Richard Hintz — Thu, 22 Sep 2016 14:27:34 -0000

Is this high availability as scripted? It doesn't appear so. If not, what needs to be done to make it HA?

Also, how are plugins and plugin maintenance handled?

Re: Using DyanmoDb with Serverless

Richard Hintz — Sat, 03 Sep 2016 10:52:15 -0000

Basically, don't do scans. See http://docs.aws.amazon.com/...

Re: Using DyanmoDb with Serverless

Richard Hintz — Fri, 02 Sep 2016 10:22:50 -0000

Just to comment that there're typos in the title and in the text, DyanmoDb. The title typo may interfere with search. No need to retain this comment after correction.

Re: How We (Mostly) Survived The Stormy Apocolypse!

Richard Hintz — Mon, 20 Jun 2016 11:59:10 -0000

Thanks for this additional information. "Settling" for eventual consistency at ≈ 1 minute for 99th percentile is the kind of real world data that's often missing in AWS's descriptions.

Also, useful to know the reasoning behind CNAMEs versus ALIAS. This got me looking at other differences in how CNAMEs and ALIAS records are handled. Suffice to say that there seem to be credible arguments against ALIAS.

Re: How We (Mostly) Survived The Stormy Apocolypse!

Richard Hintz — Fri, 17 Jun 2016 17:49:31 -0000

Interesting writeup. A couple of questions:
- How did you decide on the second region? Client demographics? Cost? Something else?
- What sort of latency between regions do you see in your eventual consistency architecture, if there's a way to tell?
- Do you have a sense if a CDN makes sense for your content? Since you're hosting static content on S3, I'm not sure that a CDN would give you much service resilience benefit, just content latency reduction, though that may be a measurable benefit.

Also, since AWS is providing your DNS services, you might look into replacing CNAMEs with ALIAS records. AWS charges for CNAME queries, but not for ALIAS queries.

Re: Migrating from MySQL (RDS) to Aurora with no downtime

Richard Hintz — Wed, 08 Jun 2016 11:42:08 -0000

"One issue I’ve encountered in ~20% of my migrations, is that the process gets stuck on less than 1% of Preparing data for migration. Sometimes this resolves itself, but my recommendation would be to log an AWS support ticket if the process stalls on this step for longer than an hour. Persist if you get told that nothing is wrong, because the issue seems to require intervention from the Aurora product specialists, something that is beyond the initial call takers capability."

When intervention by AWS staff is required, do they give you a reason or reasons why this happens and anything you might have done to prevent it?

Re: ITIL Eye for the DevOps Folks with Steven Boyd

Richard Hintz — Tue, 02 Feb 2016 16:21:40 -0000

Belatedly commenting that this was a great show. What might be useful now that the subject of ITIL has come up is another show with a contra ITIL person Red Teaming the concept.

There's a reason that ITIL is widely reviled and there were a couple of suggestions why this is so in the show. First, Steve, the guest, is an ITIL expert. How many expert practitioners are around? My guess is that most ITIL implementations are done by less experienced folks and the result is processes hardening over time and institutionalized silos.

It's widely discussed by ITIL haters that everything needs a meeting with minutes. Mordac the IT preventer comes to mind.The CAB, change advisory board, becomes the approval board or the not-on-my-watch board. Technical debt becomes hard to address since the typical legacy monolith has so many interdependencies that the ITIL process gives any substantial change a red light.

Last, I wanted to recognize specifically some great comments by Matt starting around 16:30 about the value of humans vs machines in the process.

Re: Reinventing industries with digital transformation

Richard Hintz — Tue, 02 Feb 2016 15:48:05 -0000

Some of the discussion here isn't clear, at least to me. Some examples:
-"Digital disruption is occurring fast..." What's "digital" disruption? Is digital supposed to be a synonym for technology, perhaps computers? Technology is certainly playing a role, but the basic disruptions are often in process or organizational. If I move a business application from an on premises datacenter to an IaaS provider, that's process and possibly organizational disruption. The "digital" component stays similar, at least.
-"The digital era is here and it is fueling shifts." This was true when computing began enabling other than paper transactional business processes, so how are things different from 50 years ago? Is it really the "digital era" or something else that's newly emerged in the last few years?
-Clients want "an authentic forum for collaboration and strong relationship." Nothing new here. The form might possibly be different, but generally the interactions that a customer gets from a company on, say, Twitter are no less "authentic" than speaking with a call center rep. What's the innovation?
-"Moving decision making to the front line..." This seems controversial. The front line people are still working within policy guidelines provided from above. They aren't making decisions outside these guidelines. What's different?
'"Connecting and securing services and data from different places including core systems of record" Is this a way of saying "getting available information"? What about all the operational data that businesses have in shadow systems because the core systems of record are unresponsive to business needs? These should be addressed in a discussion of innovation and disruption.
-"The ability to assemble them together as services outside the firewall, for mobile and web apps" Is this a way of saying, "Presenting available information to end users while complying with policy?" Firewalls at the edge of the enterprise premises are a 10 year old design. Firewalls now are at the instance edge.
-"Quickly identifying and resolving any performance bottlenecks that could occur." True, but needs scale for time expectations. Is "quickly" seconds, minutes, hours, longer? (I'd say seconds, but most guessing CIOs would probably say hours and be satisfied with days.)
-"Insight from non traditional data – social like twitter, internet of things, wearable devices, m2m is being used in real time business critical processes to create new business moments." What's a "business moment?" Wearable devices are being used to create business moments? Really? Are there some examples? IoT in some applications is just a different way of connecting industrial processes, so not much new there, except for connectivity. I guess my IoT connected thermostat, with sufficient aggregation, could signal the power company something significant for load leveling. Hard to tell what's meant here when Twitter and wearables are lumped with IoT and M2M. They're very different classes of business inputs.
-"Analytics that have cloud speed and elasticity." What does this mean? Cloud by itself isn't a priori speedy. Yes, cloud is elastic, but analytics aren't elastic. Analytics are a business process that consumes resources. The process that handles the analytics may be elastic, but that's different from the analytics.
-"Delivered at the right time, right place, informing business moments" Nothing new here, certainly not disruptive. Presumably business goals have always been to deliver information at the right time and place. Not sure about the 'business moments" part.
-"Agile DevOps capabilities that enable rapid experimentation" "Agile DevOps capabilities" probably means a capable organization. Saying "Agile DevOps" sounds too much like stringing together some of the words in vogue rather than trying to make a clear point.
-"Coordinated lifecycle delivery" I'd love to know what was meant, in detail here.

Why did I take the time to comment on details in this post? I guess because it seems to combine some of the reasonably good ideas that are now in active discussion with a layer of language that obscures most of the possible clarity. An executive reading this might get some buzz phrases to toss around, but would be unlikely to see a recipe for changing their organization to benefit from, much less be a leader in disruptive innovation.

The objective is to inform and provide guidance so that execs can participate in innovation, perhaps even disruptive innovation, or, at least, not get flattened by it. It would be great if future Thoughts on Cloud were reoriented to that.

Re: Test Drive Infrastructure with Arthur Maltson and Michael Goetz

Richard Hintz — Wed, 16 Dec 2015 18:52:47 -0000

Around 12:28 two tools are mentioned, Servicespec http://serverspec.org/ and Leibniz - https://github.com/Atalanta...

Re: Marketers: Fudging the Meaning of Buzzwords Matters (To You!)

Richard Hintz — Fri, 04 Dec 2015 11:54:39 -0000

I thought that the "on premise" flu wasn't actually that widespread. Then I did a search on the term.

Also recently seen in an Intel presentation, "CSP," usually meaning cloud service provider such as AWS, Azure, and GCE, applied to Twitter and Facebook. Nope.

Re: A Brief Look at Immutable Infrastructure and Why it is Such a Quest

Richard Hintz — Wed, 23 Sep 2015 18:01:34 -0000

A couple of comments on this interesting post:
------------------------------
Adron Hall: Immutable simply means something that is created and left unchanged. Immutable things are understood as is, without need to mutate them, and by that have no need to be mutable.
------------------------------

Another way to look at immutable is as “not changeable,” either by policy, as we are discussing here, or in principle, such as a constant, like “2.” For the “immutable infrastructure” type, it’s only by policy that an infrastructure artifact is immutable.

As an example, you could violate policy and change an “immutable” compute instance simply by logging into it and making a configuration change, documented or not. You might do this to correct an error, or adjust a value, or for a zillion other reasons.

If the config was in error, there’s an obvious need for a change. You can change the running instance, as is typically done today, or you can kill it and replace it with a newly generated, corrected instance. In each case, the replacement is presumably without disrupting the higher level service the computing instance supports.

------------------------------
Adron Hall: And thanks to OS-level virtualization, which is spearheading the movement for immutable infrastructure, these images are extremely fast to deploy.
------------------------------

The replacement image may or may not be fast to deploy. It’s common, for instance, to rebuild the replacement image from a minimal, bare system using some tool like Chef or Puppet. It might takes 15 minutes to deploy. There are generally one or more dependencies on remote repositories. If the remote repo is slow to respond or doesn’t respond at all, you might be waiting a long time for the replacement to deploy.

An alternative is to bake a golden image and deploy it. Obviously this can take a lot less time.
------------------------------
Adron Hall: In some scenarios performance on an individual instance is slower than a dedicated box, but it makes efforts to scale-out horizontally dramatically easier, providing another way to build out faster environments, not just from a faster deployment, but actually adding more machines to make processing faster.
------------------------------

I’m not clear about some of this. First, an individual instance or a dedicated box should perform the same if it’s generated as immutable or not. And scale out isn’t especially helped by immutable instances. You can generate a 1,000 immutable or changeable instances in the same time.

------------------------------
Adron Hall: If a patch or update is needed, the server image is updated and a full redeployment of everything is made from an immutable context. No servers in production are patched or manipulated in a mutable way.
------------------------------

The “immutable context” here is just the end result after the recipe for the resource is applied. To be clear, the recipe itself is necessarily mutable, then, after it’s tested and finalized, it can generate the immutable image.

And while production resources are an important category of immutable infra, there’s no need to limit the scope to just production. In fact, you’d often want immutable resources in QA mirroring what’s in production. Development can use the same resources, too. This deals with the “It worked on my laptop” issue. This is suggested later on the piece in the CI/CD section.

Not really discussed here is that there’s an issue for those resources that carry state, such as databases or logging. The entry to the rabbit hole starts by searching on “distributed database” or “distributed logging.”

Re: SNA was Awful. SDN Isn’t Awful (Yet)

Richard Hintz — Wed, 27 May 2015 18:21:43 -0000

This thread is a bit depressing since the SNA/VTAM model doesn't seem that much different than what the industry is now iterating toward with SDN, centralized control, with necessarily increased cooperation between the "coders for system services" and the networking group.

Did the mainframe people have their heads up their butts? I guess everyone's experiences are different, but generally the coders of mainframe system services that I've known worked to order within a fairly narrow mainframe centric scope and weren't especially obtuse, just busy in their own world.

I can guess that many mainframe systems staff thought that the Unix people, the desktop people, the LAN people, the WAN people (sometimes different than LAN), operations, developers, and certainly tech management all had their heads similarly oriented up their butts.

How difficult was SNA/VTAM? Not that bad. The mainframe person bringing up a VTAM "path" would copy the parameters from a working config or the book(s) and, after getting an allegedly tested circuit turned over from networking, would instantiate the path with 37X5 NCP reloads.

It generally worked since innovation in this area was rare. There were bigger issues with 100s of changes in the base OS regularly, using the complex SMP tool.

Did anyone want to be a VTAM expert? Only if you wanted to keep your job as a z/OS systems programmer, so, yes, a few people did. Keep in mind the level of what was called expertise was defined differently since innovation cycles were long.

And SNA/VTAM was an add-on to z/OS, never the main event. Rarely was someone only a SNA/VTAM person, even when Communications Server was introduced and TCP/IP was added to the menu. One did z/OS functions, and maybe VM, first, then communicated to end nodes and partners as needed with Comm Server/VTAM/TCPIP using NCP.

So, all of this is a long winded way to say that centralized config as with SDN will likely have similar issues. I can conceive SDN east west people thinking the SDN north south people working above the SAL have their heads up their butts, too.

And wait until we reach SDN nirvana with app developers using an API call to demand service from available resources. The end to end path seems more complex than VTAM. [Opinions for sale. $.05]