Sorry .. I meant "real-time" not "meal-time". Probably made that mistake cause i'm hungry at the moment ;)
On 2/24/06, Arjun <[EMAIL PROTECTED]> wrote: > > Ok, let me explain in more detail :- > > I have a requirement to write gathered data from a small cluster (about 40 > computers). > The number of metrics to be gathered is low (about 6-10). These metrics > need to be stored indefinitely. > We also want to process the data in a real-time manner, that is - each > insert of data results in some operations, the results > of which are used in taking meal-time decisions about the > configuration/allocation of the cluster. > > Having the data in a Relational DB makes it far easier to perform complex > operations on the stored data. This is the > main reason why we want it in a DB. This is all for a research project > into cluster computing. > The growing size of the Data is not really an issue for us as we can > truncate it to whatever we want. > > We already know that an rrd can be dumped to xml or whatever, but we > specifically want data to be written > directly to a DB. > > I hope that answers all the queries. > :) > > Thanks and regards > Arjun > > > On 2/24/06, michal < [EMAIL PROTECTED]> wrote: > > > > Arjun napisał(a): > > > Hi Folks, > > > > > > I have a requirement to integrate ganglia to a DB such as MySql so > > that > > > gmetad should store retreived data in > > > MySql instead of an RRD. > > > > You can easyly dump RRD file to XML (rrdtool dump). > > Then parse XML file and put the data anywhere you want. > > > > But keep in mind that badatabse will growing up if you just add new > > records. You must drop old data by your self. > > > > By the way, what is wrong with RRD? > > > > Michal > > > >
