7. Machines
â˘Industry Standard
â˘No RAID controller (JBOD on the slaves)
â˘Homogeneous environment is not necessary
â˘Cores, Spindles, and RAM
⢠Different conďŹgurations for different uses
7
8. Network
â˘Leverage the existing infrastructure
â˘No fancy equipment, no InďŹniband
â˘Redundancy is key, no SPOF
â˘TOR vs Core
â˘1Gb, 10Gb, and 40Gb
â˘Bonding, VIPs, other such complexities
8
37. Automation
How fast can you:
â˘Change an OS conďŹguration on 100 machines?
â˘Kill one process on said machines?
â˘Reboot all your machines?
â˘Reboot all your machines one by one, with
some added conďŹguration changes?
â˘Add 10 new fully conďŹgured nodes?
37
49. If you can manage to take your cluster offline for
possibly an hour:
1.Shutdown HBase
2.distcp to another cluster/separate folder
3.Restart HBase
* It's possible to run a distcp before shutting down, make sure you run distcp
-update -delete for the second step.
Backup - Offline
49
50. 1.Create another HBase cluster (can be remote)
2.Alter the families that need replication
3.Make sure the same tables exist on the slave
cluster
* Replication isn't done inline with the inserts in the master cluster
* See "Apache HBase Replication" with Chris Trezzo at 5:20PM
Backup - Replication
50
51. â˘Doesn't require copying data
â˘Runs in less than 60 seconds
â˘Minimal impact on performance
* See the slides from "Apache HBase Table Snapshots" with Jonathan Hsieh
& pals
Backup - Snapshot
51