Thanks JD. I will shut down one of the ZK instance. To Michael and JD,I will start another thread regarding the performances with more details.
JM 2012/6/26, Michael Segel <michael_se...@hotmail.com>: >> Network is always good to check, it's all fun and games until an >> interface negotiates 100Mb. >> >> 50ms per get sounds a bit extreme. > <mini-rant> > Funny you should mention hardware. > I did submit a talk on cluster design to Strata (NY and London) Seems it > didn't make the cut on NY, but who knows about London... > > It seems that people are now starting to get the idea that its important to > think about your hardware and cluster design before you actually start to > build a cluster. > </mini-rant> > > You're right we don't know enough about the hardware and configuration to > talk intelligently... > > Depending on the size of the row... it could cause a long time to do a > single fetch. (err get() ) > > On Jun 26, 2012, at 1:12 PM, Jean-Daniel Cryans wrote: > >> On Tue, Jun 26, 2012 at 10:56 AM, Jean-Marc Spaggiari >> <jean-m...@spaggiari.org> wrote: >>> Am I better to run it on 1? Or on 3? I just want to do some testing >>> for now. but for ZK, can I keep it in only one server for now? Or it will >>> be >>> more efficient if Iconfigure it on 3? >> >> FWIW your system will be as available is PC1 is, so just put 1 ZK on >> that node. ZK is not on the read path so whether you have 1 or 10 it >> won't change anything. >> >>> But I have issues with the performances. It's taking 20 >>> seconds to do 1000 gets with the actual configuration... I'm tracking >>> the issues. I think the network is one so I will address it this week, >>> >> >> Network is always good to check, it's all fun and games until an >> interface negotiates 100Mb. >> >> 50ms per get sounds a bit extreme. >> >>> >>> Thanks, >>> >>> JM >>> >>> 2012/6/26, Jean-Daniel Cryans <jdcry...@apache.org>: >>>> A quorum with 2 members is worse than 1 so don't put a ZK on PC2, the >>>> exception you are seeing is that ZK is trying to get a quorum on with >>>> 1 machine but that doesn't make sense so instead it should revert to a >>>> standalone server and still work. >>>> >>>> J-D >> > >
