Perhaps it'd be useful to have performance benchmarks with an Isilon setup vs. a non-Isilon setup? I saw a graphic promising 50x better performance for [ostensibly un-released] "Project Nitro," but don't get any sense of how much faster things would be for a user today. There's also been a lot of work in the upcoming Hadoop 3.0 release to support erasure coding and it might be worthwhile to address this if one of the main selling points of Isilon is its storage efficiency.
-Dima On Thu, Sep 1, 2016 at 9:21 AM, Ted Yu <yuzhih...@gmail.com> wrote: > Interesting. > > Minor correction: > > bq. The locations of all files and regions are kept in a special metadata > table “*hbase:meta*” > > The locations of hfiles are not tracked in hbase:meta > > > > On Thu, Sep 1, 2016 at 1:52 AM, Chernov, Arseny <arseny.cher...@emc.com> > wrote: > > > Dear colleagues at User@HBase , > > > > I really value your time and thank you for attention. I understand that, > > without context setting, my mail carries a substantial risk to trigger a > > huge off-top. This is not my intention at all. > > > > So, brief context is: we at EMC Isilon see strong uptake of HBase in > > mid-market and enterprise datacenters, where sustained practices of ITIL, > > compliance, legacy architectures and cost control , -- oftentimes are not > > correlating well with true “Big Table” web-scale concepts of this > brilliant > > database. We see emergence of small, order of 100-s of TB-s, virtualised > > HBase deployments, in different pockets of enterprise, as well. > > > > I humbly ask you to review and share your feedback (1x1 or in this forum) > > your PoV-s on “Why HBase on Isilon” blogpost that I’ve just published: > > > > Why HBase on EMC Isilon – Top 5 Reasons http://bit.ly/2bSmKgc > > > > I’m very keen to learn more from this respected community, as well as > > possibly answer your questions about EMC Isilon and share (to extent of > > what could be shared) what we see. > > > > -- > > Arseny Chernov > > 陈毅誠 > > > > > > >
