Hi DPF,
Good to hear that HQ has been useful for you, and thanks for the kudos.
The way we store the metric data (it would obviously be unwieldy to
be dealing with all of the data points for larger time ranges) is
that we move and average values into larger granularity tables as
time goes on. So, for example, when you view data for the last 4
days we would look into the table that stores the data at hourly
intervals; but if you view data for the last 4 hours, then we would
look into the table that stores the data at minute intervals. The
way that finer grained data get moved into more coarse data tables is
by storing the average, high, and low values. However, that all gets
displayed in the chart. The blue column displays the average value,
and the overlaying I-beam illustrates the high and low values. Now,
if you are seeing the spike in the very recent past, and the you zoom
out, it could very well be that the value has not been moved to the
more coarse data table yet. Does that make sense?
We have had some discussions around updating the charting with either
flash or other Ajax implementation that would allow you to interact
with the chart more effectively. However, as of yet we do not have a
specific timetable for when that will be implemented. What is your
motivation to see the charts be implemented differently?
Charles
On Nov 7, 2006, at 1:36 PM, DreadPirateFlint wrote:
> Hi there, I've been using Hyperic for a few months now, and its
> great. Its really been helping me get a grip on what happens when
> my site gets heavily loaded, and I've already found a few
> bottlenecks that I might not have otherwise found. Great job,
> everyone.
>
> I have a question about viewing larger date ranges. I have a
> metric (lets way web requests per minute). When I'm fairly zoomed
> in (4 hours view), I see totals that are high- lets say 7,200
> requests per minute. But- when I zoom out, I see that same spike
> only goes as high as 4,329. It would seem that this is by design-
> perhaps to "smooth" out the graph. I'm wondering if there exists
> an option to smooth, but preserve the peaks, so I can get actual
> data on the peaks without having to zoom in to get the actual data
> value of the peak.
>
> Has there been any discussion of using a flash or java applet to do
> the charting?
>
> Thanks!
>
> DPF