Coté Memo #040: Market cycles in infrastructure software, still more OpenStack action


Hello again, welcome to #040. Today we have 49 subscribers, so we’re +3. What happened?! I’d love to hear what you like, dislike, your feedback, etc.: (If you’re reading this on the web, you should subscribe to get the daily email.)

See past newsletters in the archives, and, as always, see things as they come at and @cote.



  • There was diverse feedback to my research agenda question from yesterday. One person suggested focusing on points of friction that customers are facing, another of you on re-framing DevOps to be “all that stuff needed to become a coding business” (my rephrasing), and yet another in a more commercial sense: what will people pay for? All good input that’s given me plenty to think about.

  • One of the key things to fall out of the various conversations is being more straight forward with the “what I really think” content. Oddly enough, in the analyst world, this is not as easy as you may think. One theory being kicked around in the inbox is to “give away” the facts and to charge for opinion. That is, put “what I really think” behind a paywall. As I said in the thread on the topic, I almost do the opposite nowadays. It’s something to think more on in the endless analyst business model noodling I do. You got opinions? As an example, listen to the last 20 minutes (sort of accidentally recorded) of the most recent Exponent podcast. I think it’s some of the best, most valuable work that podcast has done. I didn’t think of big Apple event at all like they do…and now I do! (There’s also good cultural commentary on Asian luxury buying habits.)

  • I’m sure you’re thinking, “hey, Coté. I know you try to release and podcast every two weeks. What up?” Well, we have a recording in the can. It just needs some editing. SORRY! (I’ll be recording Software Defined Talk this week remotely. We’ll see how that goes!)

Tech & Work World

Quick Hits

Systems Management Cycles: best of breed vs. full suite

On the plane today I started working on a piece considering the question, “should we make a platform out of all our point products?” In that context, I ended up typing up one of my core cycle theories for how the systems management market works. The below is just a first draft, pardon errors and stupidity.

The “infrastructure software” layer in the IT stack is that sinewy and sometimes well marbled layer between the hardware and the application layer: operating systems, cloud platforms, systems and cloud management tools, and all the other software required to keep the applications (or “services” if you were indoctrinated properly in the mid-2000s) running and healthy. These are the concerns of the “sysadmin”: managing the servers at the operating system level, ensuring the network is properly functioning, deploying and configuring the applications (custom written or packaged), managing requests for modifying the software, and then monitoring the performance of all of it so that end-users have the best (or, at least, agreed on) level of performance possible.

This diverse range of responsibilities and the vast array of vendors out available to satisfy each need creates an panoply of tools that a sysadmin can use to do their work. As with most IT, there’s an ongoing cycle of “does everything suite” vs. “best-of-breed” products in the industry. Currently, for example, as the new technologies needed to run and manage cloud penetrate into the enterprise (based on our studies, we reckon that most companies are well through virtualization and now onto cloud), it feels like we’re in a best-of-breed phase of the cycle: no single suite of tools can manage everything perfectly or match the all of the diverse use cases in the market.

In this initial phase of a technology market (as we are in with cloud) startups who seek to move faster and take on more risk than incumbents find holes in the suite approaches and prosper because they offer better performance, often at lower costs.

We’ve seen this happen in systems management, for example, with Splunk going against Big 4 log management tools and APM vendors such as New Relic and AppDynamics out-innovating the slower moving incumbents early in this phase of APM.

A similar cycle occurred with virtualization management with many vendors offering better ways to manage VMware clusters than VMware itself. Several years ago, virtualization market seemed to cluster back up into suite approaches with VMware and others buying up point tools. Atlassian in the developer tools space used disruptive approaches to go-to-market to peddle genuinely innovative developer tools to make trouble for the incumbent application life-cycle vendors.

Successful, best of breed, nimble tools often become the stable but innovation-listing incumbents for a new batch of startups.

As another ask, I’m trying to track down and “fact check” the idea that the Tivoli Framework was very problematic. People always talk about it, but I don’t know if I have any good write-ups on hand. Anyone have some pointers or first hand experience?

Rewrite brought to you by CA


As you may know, I’m interested in CA’s marketing and brand transformation. Like Compuware, they seem to have simplified their face, as it were, a lot. While in JFK for a layover, I noticed CA had a lot of ads up. Above is one of the more interesting ones. It’s very “digital enterprise”/“coding business”/“bi-modal IT”/“third platform” (pick you favorite analyst/pundit phrase). Good for them.

I like the “we’re gonna spend the next 10 years rewriting all the software” hyperbole I start most of my talks with. But of course I’d like it.

Fun & IRL

If pictures of CA advertising in JFK aren’t “fun” for you, I don’t know why you subscribe to this thing!

(No, please don’t leave.)