Translating with DrWatson… this can take a few seconds the first time.
This is a genuine, complex translation. DrWatson protects commands, error codes, and log output while naturally translating the surrounding text. It’s translated once and saved.
John Smith asked whether LRU writes should be lower than chunk writes in onstat -F. Art Kagel said it is debated; LRU writes ease checkpoints and about 70-75 percent LRU writes tuned OLTP faster, while bulk loads suit a more balanced mix. ZhiWei Cui said it depends on disk IOPS capability. Art agreed tuning depends on hardware and workload.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
John Smith — source: IBM Community (ConnectedCommunity.org) Informix forum
Hello
Reading Onstat -F, there are LRUs writes more than Chunk writes
should LRUs Writes be 0 or at least less than chunk writes ?
Thanks :)
------------------------------
John Smith
------------------------------
#Informix
↪ replying to John Smith
Art Kagel — source: IBM Community (ConnectedCommunity.org) Informix forum
John:
That's an open debate.
Chunk writes are more efficient overall for the storage system
LRU writes minimize the overhead of checkpoints and reduce the number of chunk writes which can overwhelm the system
Mostly I know that when I tune a system to be about 70-75% LRU writes it runs faster. Especially for OLTP workloads, but bulk loads type workload also to some extent.
Art
------------------------------
Art S. Kagel, President and Principal Consultant
ASK Database Management Corp.
www.askdbmgt.com
------------------------------
↪ replying to Art Kagel
ZhiWei Cui — source: IBM Community (ConnectedCommunity.org) Informix forum
I think it has more to do with the ability of disk IOPs.
If the IOPs capability of the disk is very good, you need to adjust the LRU a little larger and make use of the write efficiency of chunk.
If the IOPs performance is poor, it is recommended to adjust the LRU lower to reduce the transaction blocking caused by CKPT.
------------------------------
ZhiWei Cui
GBASE
ShiJiaZhuang
------------------------------
↪ replying to ZhiWei Cui
Art Kagel — source: IBM Community (ConnectedCommunity.org) Informix forum
Well taken points! I agree.
But my point holds that for OLTP workloads, heavier LRU writes versus chunk writes is better, while for bulk load workloads, a more balanced or even chunk write heavy configuration is better.
You are correct, one has to tune for the hardware, but also for the workload. Everything we do as DBAs involves tradeoffs.
------------------------------
Art S. Kagel, President and Principal Consultant
ASK Database Management Corp.
www.askdbmgt.com
------------------------------
We use strictly necessary cookies to make this site work. With your
consent we’d also use optional cookies for analytics and marketing. You can accept all,
reject all, or choose. Read our Cookie Policy.