or
or
By clicking below, you agree to our terms of service.
New to HackMD? Sign up
Syntax | Example | Reference | |
---|---|---|---|
# Header | Header | 基本排版 | |
- Unordered List |
|
||
1. Ordered List |
|
||
- [ ] Todo List |
|
||
> Blockquote | Blockquote |
||
**Bold font** | Bold font | ||
*Italics font* | Italics font | ||
~~Strikethrough~~ | |||
19^th^ | 19th | ||
H~2~O | H2O | ||
++Inserted text++ | Inserted text | ||
==Marked text== | Marked text | ||
[link text](https:// "title") | Link | ||
 | Image | ||
`Code` | Code |
在筆記中貼入程式碼 | |
```javascript var i = 0; ``` |
|
||
:smile: | ![]() |
Emoji list | |
{%youtube youtube_id %} | Externals | ||
$L^aT_eX$ | LaTeX | ||
:::info This is a alert area. ::: |
This is a alert area. |
On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?
Please give us some advice and help us improve HackMD.
Do you want to remove this version name and description?
Syncing
xxxxxxxxxx
HPC Storage Benchmark for Ling and Grigorios
Conclusions - text suggested for paper
Notes
Results of tests were surprisingly consistent, though affected by occasional degradations in performance, presumably due to Ceph doing some background task.
How to acknowledge CREATE in papers:
https://docs.er.kcl.ac.uk/CREATE/acknowledging/
Tests
Fio jobfiles mostly as presented at https://docs.oracle.com/en-us/iaas/Content/Block/References/samplefiocommandslinux.htm
Case 1 - sequential read
Jobfile
Result 1 - Saturday 8th Feb, lunchtime
Result 2 - Sunday 9th Feb, lunchtime
Case 2 - sequential read, 1M block size
Jobfile
Output 1 - Sunday lunchtime
With a new file, to see if it gets put on a different disk
Case 3 - two nodes at the same time, case 1 and 2 on different files
The results below show that the network connection at the node is the limiting factor for bandwidth
Case 4 - two nodes case 2 on same file
Case 5 - seq read 28 processes
Jobfile
Output
Case 6 - seq read 1 process
This case shows that we can't saturate the bandwidth with 1 fio process
Input
Output
With 2 output - around 13:05
Much slower than expected - why?
With 3 output - around 13:05
Much slower than expected - why?
Repeat of 2
Repeat of 3
Repeat of 1 - again doesn't saturate, seems to be genuine
Case 7 - random read latency
Jobfile
Output
Case 8 - random read bandwidth
Jobfile
Output
Output with 1 process
Case 9 - random rw
Input
Output
Does this show that performance of other benchmarks shows caching?
Test with a much larger file that won't be cached
Case 10 - 100GB file seq read
Jobfile
Output
Repeat after deleting file
Repeat keeping file
Repeat after deleting file
Case 11 - 100GB random read
Jobfile
Output
This is faster than I expected - were the previous tests slowed down?
Case 12 - random rw latency
Input
Output
Case 13 -
Input
Output
Case 14 - random read latency on 100G file
Input
Output