Hi， Andrew， Thanks for the info! Yes, I agree, it depends on the use case and multiple clusters might be needed for the scalability. Just want to get an idea of the size limit of a single HBase cluster from the technical point of view.
Cheers, Dan On 2019/03/22 22:57:15, Andrew Purtell <apurt...@apache.org> wrote: > This information isn’t publicly available to us either. I think our > experience is we have a distribution of cluster sizes that fall along three > orders of magnitude: > > 10-100: Most start here. > > 100-1000: Mid size and enterprise. > > 1000-10000: Peak scale internet. > > I think a cluster of size 5000 nodes or larger hasn’t ever been done but > like I said we don’t really know. > > Most organizations that use HBase at scale run more than one cluster, so > the total amount under storage can be in the peak scale range (1k-10k) but > the largest particular cluster is more like 100-500 nodes (my employer > falls into this bucket). > > A context free discussion on cluster sizing is rarely useful. What you > might need, and if you even need something like HBase, is going to depend > on your requirements. Basically: Try everything else first, and when you > need us, you’ll know when they won’t work for you. > > On Thu, Mar 21, 2019 at 8:35 AM dongda...@gmail.com <dongda...@gmail.com> > wrote: > > > Hi, Experts, > > Does anybody know what is the largest HBase cluster ever used by any > > company? The only info that we can get is a few PB of size in year 2011 by > > google. But what is the current status after so many years of technical > > evolvement of HBase(e.g., can HBase support ~100PB storage?)? Thanks so > > much! > > > > Cheers, > > Dan > > > -- > Best regards, > Andrew > > Words like orphans lost among the crosstalk, meaning torn from truth's > decrepit hands > - A23, Crosstalk >
