After I wrote up the initial activation of Business Intelligence structures for Central Performance History, (“After a Short Hiatus, a Return to Central Performance History“) I waited a few days to see how it looked. Not too surprisingly, while a few data points have been collected and are waiting for deeper analysis, there are parts strewn across the information highway. Let’s see how bad it looks.
The first surprise was the discovery of new MTE Classes and, especially, new Collection Reorganization Schemas (or as someone likes to call them, for pseudo-intellectual coolness, Schemata). Now we’ve got CPH_BI, CPH_BI_HOUR, CPH_BI_MINUTES, and CPH_BI_QUARTER_HOUR. Why do I think that is significant? Because when I searched for several of these terms in SCN, and in Google, nothing came back. It’s a mixture of fear and exhilaration when you stumble onto new key words. Obviously they’ve been around for a couple years or so, I guess, but no online searchable documentation?
CPH_BI CPH_BI_HOUR, etc.
HISTORY of, e.g., CPU
Then, over to OS06N to see if my guess about CPH data being visible as a drill down with other system values is correct. Again, mostly correct. There are values, which on casual glance, would tell you that we’re collecting data without human intervention. Except that a day is missing. That’s ominous.
NO REMOTE CPU HISTORY
Even though it looked like I was getting remote data last week, there’s nothing in the OS06N screen to show for it. In fact, the error message says so: “No Performance Data Available for Category CPU.” Back to the drawing board, as they say.
I noticed a “settings” area on OS06N, so I drilled into that. As I expected, the collection schema that were visible on RZ23N but not doing anything simply don’t show up here. I’m not sure if that means they weren’t activated at all, or were activated but are corrupt in some other way.
The “days” in the far left column doesn’t show the total days of history (I presume), but probably something like a retention period. We might have to wait 6 weeks to find out.
The pop-in rectangle in the bottom left is to show how to get back to this screen.
When I tried to alter or add collection schema through the pull-down tab on that field, nothing happened. Oh, so optimistic, and so wrong.
MORE CPU HISTORY
Drilling into CPU history again, I found not just a missing day, but missing hours. And with no discernible pattern or reason. Was the system too idle for data? Are there masks I’m not aware of? Is it just randomly working? “Argh!”
ST03 Process Chain statistics
Over to ST03, workload history, to see how the automated data loads look. Actually, nothing too surprising here. There are frequent data loads, and a periodic master data load. Each chain looks consistent and complete, as far as I can tell.
ST03 Analysis of Individual Process Chain(s)
Looking more closely at one run, there are a lot of moving parts. Each step is green, although there are yellow lights on the main loads. I’m assuming this is due to no data, but we’ll have to wait until queries are working to find out for sure.
Note that I sorted this view, and the one above, by time (last column on the right). Always good to see things in sequence. Unless they’re running in parallel. Also note that the first “hourly” run starts after 4AM, not midnight. I’m thinking this is because internal times are GMT. I viewed the data at 10:30 AM, and could see timestamps up to 2PM. Time warp.
That’s all for now, folks.