Re: PCP design and implementations

Hi, Marko -


> > Well, to visualize the data together, you need the data available
> > together.  That means potentially thousands of pcp archives in a
> > central location, or thousands of concurrent live connections to
> > remote pmcds.  Neither scenario is currently well supported at that
> > scale.  (If nothing else, our internal concurrency story is weak.)

> We don't need to visualize data about more than one server.  Just
> getting a list of servers and visualizing data about one server is
> too slow.

Understood, but in your case you had already centralized many GB of
data from many hosts, and the particular web dashboard configuration
required scanning of (way too much of) it.  If the data resided on its
originating server, or used a custom dashboard, it would not have hit
this degenerate case as hard.


> Getting list of per proc metrics is too slow and not user friendly.

Can you remind me what you're referring to?


> Getting a list of servers means reading these thousands of archives. And
> slowest part is a reading of a last archive for each server (that pmlogger
> is writing to).

Yeah, we got this part analyzed OK under github issues/117 (thanks!),
just need to decide how to correct.


- FChE

-=-=-=-=-=-=-=-=-=-=-=-
pcp mailing list
[email protected]
https://groups.io/g/pcp/messages
-=-=-
Groups.io Links:

You receive all messages sent to this group.

View/Reply Online (#14896): https://groups.io/g/pcp/message/14896
View All Messages In Topic (13): https://groups.io/g/pcp/topic/3287116
Mute This Topic: https://groups.io/mt/3287116?uid=174580
New Topic: https://groups.io/g/pcp/post

Change Your Subscription: https://groups.io/g/pcp/editsub?uid=174580

Group Home: https://groups.io/g/pcp
Contact Group Owner: [email protected]

Terms of Service: https://groups.io/static/tos

Unsubscribe: https://groups.io/g/pcp/leave/354243/563757577/xyzzy
-=-=-=-=-=-=-=-=-=-=-=-