IT Management Podcast #006 – Oracle IT Management; SML, CML, & Friends; Fog vs. Clouds; SLAs; The Hollywood Model Applied to Ops

Download the episode directly here, or subscribe to the feed for auto-magic downloads in iTunes or other podcatcher.

In this episode, we have special guest William Vambenepe, IT Management blogger and “architect in the application and middleware management part of Oracle’s Enterprise Manager division” as he puts it.

I’m going to try a new tact here for timelessness and write-up the lengthy description later in favor of posting the actual audio quickly. I figure this will be fine as the lengthy text is mostly for archival purposes and those subscribed to the feed will get their episode sooner rather than later without blocking on text they may or may not read. With that said, here is a brief summary:. Here’s the detailed note:

What with William as our guest, we open up with the question, “what’s the deal with Oracle in IT Management?” As William notes, he works on application and middleware management, so that’s the level he likes talking at. I ask how Oracle manages the balance between providing management for its own product lines vs. “everything else.” While there’s a strong focus on providing management for Oracle databases, applications, and middleware, there’s plenty of attention paid to all the non-Oracle stuff out there.

John and William then get into a discussing of the integration problems that exist in IT Management: namely, that there’s a bunch of different stuff laying out there. William points out that the mashup idea and culture doesn’t quite play out in the management space as widely as we’d like.

I then break in with one of my favorite anecdotes about how boring even programmers think IT Management is (hello, Chip Holden!) so we can’t really expect software developers to write their code “IT Management friendly.”

Then, I prompt William to go over one of his recent posts speculating about what Microsoft’s delay to System Center Service Manager means for SML, CML, and CDM. Having worked on the SML spec while at HP, William thinks out-loud about the possibility of introducing yet another big model to represent everything in IT.

Hearing that William worked on SML, we launch into a discussion of the evolution of SML, CML, and it’s origins in Microsoft’s SDM, MoF, and then CIM. As background, I tend to write-up posts on that topic when at Microsoft events like MMS or TechEd, for example, this one from MMS 2007.

We summarize the purpose of the various specs: SML describes a generic way of describing entities and the relationships between them without outlining any specific IT assets. CML, on the other hand, aims to describe specific IT assets. At the moment, there’s an SML spec, while CML currently has a just white-paper.

After raining down a blizzard of TLAs, John brings us back to our sanity by asking about the Service Management Data Base, that layer of integration over all the existing muck instead of just introducing more muck. See episode 004 for our initial discussion on this topic. This of course touches on one of John’s theories of what we need in IT Management, more data warehousing and business intelligence tools.

I then ask John and William one my impossibly vague questions: is there any horizon to all this IT Management tooling, or are we forever going to be 1/4 of the way to fixing it all? As William put it: “I don’t see any finishing line.” More specifically, the underlying nature of IT changes so much and frequently, that the related IT Management tools can’t help but always be churning away at the full solution.

John then tells us about his conversation with Flexiscale this week. In addition to the usual “how do you provide cloud services?” John asked them about the monitoring and management they provided.

At this point, William nicely cracks wise: clouds are great when they’re up in the sky, but if they fall onto the ground, you’ve got fog, and then you get lost in the fog.

From here, we point out a reoccurring idea here: there needs to be more, basic tools for people to manage and peer into the stuff they have running in the cloud. Third party SLA monitoring is sort of the base level, but getting more visibility and tools to track and fix things is the next. Not to mention the need to access logs for auditing purposes and identity provisioning. As an example, how would identity life-cycle management in all this cloud stuff prevent another Societe Generale?

I then ask John and William if they saw the SaaS speculation about Dell following up on the acquisition of MessageOne this week. Larry Dignan has a nice wrap-up of Dell’s SaaS buy-up, and Gordon Haff chimed in with some brief comments as well. As William points out, even if Dell’s initial cloud offers are “low level,” if the plan pays off, they’ll certainly get in Oracle’s and other’s territory of providing business application. As he puts it, they’ve all got the same spreadsheet they’re working off ;>

Pulling back to the broader idea of becoming a SaaS/cloud seller, I note that most people I talk with try to shy away from dealing directly with customers. They’d rather resell software and services through someone else. It’s as if dealing with customers directly is too low-brow. That said, of late, large companies like IBM and Sun have been more open of late to selling clouds directly to customers rather than just being arms suppliers.

After this, we get into some M&A news for the week in IT Management: OpSource bought Irish SaaS billing company LeCayla. I’m always obsessed with SaaS billing because I have several friends working in that area at Austin startup eVapt.

I also quickly go over a briefing I had with Jamcracker who provides a platform for white-labeling SaaS stuff like hosted Exchange, SugarCRM, and other hosted applications.

John mentions that he’s seen a lot of people offering rails hosting in the cloud – see his interview of his with Morph’s David Abramowski. From here, I ask John and William if they encounter a lot these SaaS/cloud people talking about SLAs.

Here, we get into a discussion of classic IT department folks competing with cloud providers. William raises the classic example of mail storage quotas, which illustrates the point well. At a past job of his, the IT department had set a 200 meg email quota, which is of course painful for people with lots of email or who emails around Word documents and PowerPoint presentations. Looking at the seemingly limitless storage GMail (or whoever) provides, these people started asking IT if they could pay the $20 or $50 a year they’d pay hosted emailers to increase their corporate email quota. Of course, this kind of service catalog shopping around doesn’t really work too well for most IT departments.

We wrap-up the main content by discussing the idea that this whole cloud-based IT world might fit better with The Hollywood Model of work. That is, instead of having static teams – “departments” even – workers come together on a project basis to put duct-tape together services in the cloud. How long-term maintenance of this works isn’t too clear, but the initial creation and updating seems alluring when the technology of cloud computing gets into the mix.

Before letting William go, I try to triangulate which buildings are the famous Oracle Redwood Shores buildings. He confirms that you can see them as you’re landing at SFO.

Thanks again to William for guesting, he was great fun and hopefully we can get him back again ;>

Disclaimer: IBM, Sun, BMC, and Microsoft. See the RedMonk client list for clients mentioned.

Technorati Tags: cloud, oracle, s3, sla