Thanks for the responses. I'll wait to test with HIVE-20621 to see how much that helps. For HIVE-17194, doesn't look like that is documented outside of JIRA currently. It looks like I may have picked up over the wire compression by default. I had assumed it wasn't compressed.
Kevin Risden On Tue, Oct 16, 2018 at 12:44 AM Gopal Vijayaraghavan <gop...@apache.org> wrote: > Hi, > > > > I was looking at HiveServer2 performance going through Knox in KNOX-1524 > and found that HTTP mode is significantly slower. > > The HTTP mode does re-auth for every row before HIVE-20621 was fixed – > Knox should be doing cookie-auth to prevent ActiveDirectory/LDAP from > throttling this. > > I fixed it for the open-source JDBC drivers, but I think the proprietary > ODBC drivers are still broken (the C++ code doesn’t save cookies). > > > I also didn't see any information about reducing amount of data over the > wire with compression. > > It’s possible Knox doesn’t generate/forward the Accept-Encoding headers. > > > > https://issues.apache.org/jira/browse/HIVE-17194 > > > Cheers, > > Gopal > > > > *From: *Prasanth Jayachandran <pjayachand...@hortonworks.com> > *Reply-To: *"user@hive.apache.org" <user@hive.apache.org> > *Date: *Monday, October 15, 2018 at 12:34 PM > *To: *"user@hive.apache.org" <user@hive.apache.org>, "user@hive.apache.org" > <user@hive.apache.org> > *Subject: *Re: HiveServer2 performance references? > > > > One performance fix that is fixed recently and is related is HIVE-20621 > > > > Thanks > > Prasanth > > > > On Mon, Oct 15, 2018 at 12:11 PM -0700, " > Kevin Risden > " <kris...@apache.org> wrote: > > Are there any resources on HiveServer2 performance - specifically binary > vs HTTP mode? I would also be interested in any information about enabling > compression between HiveServer2 and a client like beeline. > > > > I was looking at HiveServer2 performance going through Knox in KNOX-1524 > and found that HTTP mode is significantly slower. I searched and didn't > find any references to performance of HiveServer2 out of the box. I also > didn't see any information about reducing amount of data over the wire with > compression. > > > > Any pointers would be very helpful. Thanks! > > > > Kevin Risden > >