[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[LDM #MBG-654954]: Clustering
- Subject: [LDM #MBG-654954]: Clustering
- Date: Tue, 26 Feb 2013 15:12:48 -0700
Hi Leo,
re:
> We are looking at clustering our LDM environment using Red Hat clustering
> and was wondering if Unidata knows of any customers that do cluster LDM.
There are several sites in the Unidata community that are using a clustered
approach to LDM data relay:
Unidata Program Center (UPC, us)
Texas A&M
Penn State
NOAA/GSD
The first three on this list are using the LVS capability that comes
with Linux (e.g., RedHat). NOAA/GSD is using hardware to provide the
cluster director capability.
re:
> Also, can a single LDM environment handle both Level II and Level II
> radar data being sent over the same dataflow?
Did you mean Level II and Level III data?
The answer is yes depending on what you mean by 'dataflow'. We relay
all of the data available in NOAAPort (which includes NEXRAD Level III
products) and all of the data included in the NEXRAD2 feed being produced
by the NWS and all of the high resolution model data in the CONDUIT
datastream on a variety of machines here in the UPC. We don't try to
request all of that data using a single ~ldm/etc/ldmd.conf REQUEST; we
split the the request into two ldmd.conf REQUESTs. For example:
request NEXRAD2 ".*" lion.caps.ou.edu
request ANY ".*" jackie.unidata.ucar.edu
The first REQUEST asks for all of the Level II data from the OU/CAPS
machine lion.caps.ou.edu. The second REQUEST asks for all of the data
flowing in NOAAPort from the NOAAPort ingest machine jackie.unidata.ucar.edu.
The other thing we typically do is redundantly request all feeds on
machines that act as front ends to our LDM/IDD cluster, idd.unidata.ucar.edu.
The real server backend machines REQUEST all data redundantly from the two
front end machines. The machine running LVS farms out connections to the
real server backends that send the data to the downstreams that requested
such.
If you would like to discuss how we (and Texas A&M and Penn State since their
clusters are patterned after ours) setup our clustered LDM environment, we
can either field questions via email, or we could chat by phone.
re:
> Thanks,
No worries.
Cheers,
Tom
--
****************************************************************************
Unidata User Support UCAR Unidata Program
(303) 497-8642 P.O. Box 3000
address@hidden Boulder, CO 80307
----------------------------------------------------------------------------
Unidata HomePage http://www.unidata.ucar.edu
****************************************************************************
Ticket Details
===================
Ticket ID: MBG-654954
Department: Support LDM
Priority: Normal
Status: Closed