I've 'talked' with a friend in Germany who's been doing a lot of research on SSDs and Informix (Yes they are still around within IBM...)
Lessons learned: The more expensive solutions (PCI-E) SSD cards don't perform much better than a controller and SATA SSDs. Although the price is much greater. Raid 10 still makes sense. With respect to HBase/ZK ... You could put out a couple of boxes with SSDs for ZK nodes. I don't know your budget or types of boxes... Patrick H recommends 5 ZK nodes in a production cluster. I don't disagree, but feel that you could get away with 3 and still have a window for maintenance... however... if you can measure you down time in terms of $100.00s per hour or more, the 5 ZKs are still pretty cheap. (Or rather 3ZKs and then run a ZK on the name node, and a ZK on your secondary.) (Then you only have to build out 3 boxes, and put a second set of drives on the name node and secondary) I haven't tried this out yet... my guess is that you probably won't see a large enough performance increase to justify the expense... JMHO -Mike PS. I've been watching the prices of SSDs for a different application. I now see Crucial's SATA III SSDs are actually cheaper than their SATA II drives. > From: [email protected] > To: [email protected] > Date: Thu, 8 Jul 2010 19:38:22 -0500 > Subject: RE: zookeeper & HBase > > Good to know ZK is IO intense. > Since ZK does not require much disk space and is IO intense. Has anyone > played with using solid state drives for ZK. > We have a 20 node cluster. It would be feasible to have a 3 node ZK all > configured with solid state drives. > > Thanks > Arun > > -----Original Message----- > From: Jonathan Gray [mailto:[email protected]] > Sent: Thursday, July 08, 2010 4:25 PM > To: [email protected] > Subject: RE: zookeeper & HBase > > ZK is sensitive to IO starvation which is why it is recommended to keep it on > a separate node or separate disk. In most cases, giving ZK its own disk is > sufficient and dedicated node(s) are unnecessary. > > On smallish clusters like 10 nodes, I would recommend starting with just 1 ZK > node co-located with your NameNode and HMaster, but with a dedicated disk > just for ZK. Since the NN is a SPOF, having one ZK doesn't really lower your > fault tolerance, except that it may be on a non-raided disk. I encourage > RAID usage for NN and ZK. JBOD for DN/RS. > > JG _________________________________________________________________ The New Busy think 9 to 5 is a cute idea. Combine multiple calendars with Hotmail. http://www.windowslive.com/campaign/thenewbusy?tile=multicalendar&ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_5
