https://serverfault.com/questions/12679/can-anyone-explain-precisely-what-iowait-is
http://bencane.com/2012/08/06/troubleshooting-high-io-wait-in-linux/
http://www.thattommyhall.com/2011/02/18/iops-linux-iostat/
http://www.ufsdump.org/papers/uuasc-june-2006.pdf
http://www.ufsdump.org/papers/io-tuning.pdf
http://veithen.github.io/2013/11/18/iowait-linux.html
http://www.ufsdump.org/
Sept.29 log
Monday, January 29, 2018
Sunday, January 28, 2018
Ether mining with DwarfPool
Quite happy today that i've finally setup my ether mining rigs on AWS, and joined DwardPool. Checkout my rigs performance at this DwarfPool link :)
It's not profitable with current aws pricing, but hey, it's super fun !!!
It's not profitable with current aws pricing, but hey, it's super fun !!!
Wednesday, December 27, 2017
Monday, December 25, 2017
Reading list
strace
traceroute
linux sysAdmin guides: https://linux.die.net/sag/
linux cli tool summary https://linux.die.net/Linux-CLI/
linux documentation https://linux.die.net/
traceroute
linux sysAdmin guides: https://linux.die.net/sag/
linux cli tool summary https://linux.die.net/Linux-CLI/
linux documentation https://linux.die.net/
Thursday, November 30, 2017
DB2 PureScale on AWS?
According to IBM site (link-1, link-2), pureScale minimally needs
- the IBM General Parallel File System software to create the DB2 cluster file system on the shared disk;
- fast network connectivity (TCPIP allowed since 10.5.0.4 Cancun release).
Notice IBM recently rebranded GPFS to Spectrum Scale, which is conceptually deployable to AWS (ref.link), though fine prints may apply (eg. no encryption in ebs, not sure if this is a hard requirement, yet......)
Sounds like it's quite feasible to have pureScale on AWS? will give it a try.
Also, nice guide combing QRepl and across two pureScale clusters.
As for HADR across 2 pureScale clusters, many restrictions apply, with the most significant one being no ROS, followed by (NEAR)SYNC not supported, among many others listed here.
More on GPFS, or should i say spectrumScale, please refer to this link.
Also, nice guide combing QRepl and across two pureScale clusters.
As for HADR across 2 pureScale clusters, many restrictions apply, with the most significant one being no ROS, followed by (NEAR)SYNC not supported, among many others listed here.
More on GPFS, or should i say spectrumScale, please refer to this link.
Saturday, November 25, 2017
HBase question dumps
Collection of random questions about HBase, for my own reference:
- Why prefer JBOD over RAID for slave nodes? is it b/c of the underlying assumption of using HDFS? how does HDFS replicates blocks? is it possible that replication fails before a block get replicated? How to recover from such failure, from HDFS standpoint of view?
- Not directly related to HBase, but what are the pros and cons of different fs mentioned in Chapter 2? (ext3,4, XFS, ZFS....)
- RPC call between HBase and HDFS, assume version matches, performance concerns?
- Best practice on how to avoid memory swap, especially on slaveNode.
Subscribe to:
Posts (Atom)