trafficserver-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From ZHAOCHUNBO [赵春波] <>
Subject RE: 答复: ram_cache for big objects
Date Mon, 18 Apr 2016 12:31:50 GMT

HI Sudheer,

Acording to the document of “volume.conf”, it seems I can not assign ramdisk to one specified
Each volume is striped across several disks to achieve parallel I/O. For example: if there
are four disks, then a 1-GB volume will have 256 MB on each disk (assuming each disk has enough
free space available).

So do you mean any other plugins?


赵春波 Thomas   CDN架构师
邮箱 :<>

From: Sudheer Vinukonda []
Sent: Saturday, April 16, 2016 10:58 AM
Subject: Re: 答复: ram_cache for big objects

I haven't had to rely on page cache with ATS (unsure it'd work reliably, given ATS's appetite
for memory - ram cache + non ram cache).

But, you can configure multiple cache partitions with the required files mapped to ramdisk
(using volumes) and the rest mapped to regular disk. Cache volumes allow mapping based on
host name (hosting.config). If the domain is not unique, you can modify the cache key to add
a special domain name for the files needed and store them in the volumes that are tied to
the ramdisk.

On Friday, April 15, 2016, 7:27 PM, ZHAOCHUNBO [赵春波] <<>>
Hi Sudheer,

Thank you very much for you suggestion.

But I think ramdisk is not suitable for my case. There are many other files in the same domain,
the storage requirement is larger than my ram storage.
Only several of them are hot files, may I have a way to make ram_cache or linux pagecache

Thanks in advance,


发件人: Sudheer Vinukonda [<javascript:return>]
发送时间: 2016年4月15日 23:47
主题: Re: ram_cache for big objects

One approach is to use a ramdisk (e.g configure tmpfs ).

On Friday, April 15, 2016 12:33 AM, ZHAOCHUNBO [赵春波] <<javascript:return>>

HI all,

I have met one problem:

There is one big file(600MB) and it is very hot, it is larger than ` ram_cache_cutoff ` setup:
CONFIG proxy.config.cache.ram_cache_cutoff INT 20M

I have noticed that the `rKB/s` on one disk is very high, and the disk usage is keeping nearly

Device:        rrqm/s  wrqm/s    r/s    w/s    rkB/s    wkB/s avgrq-sz avgqu-sz  await  svctm
sde              0.00  124.00  196.33    2.33 47210.00  505.33  480.36    35.17  182.13  5.02

It seems linux pagecache does NOT work when using ATS, and ATS ram_cache does not work either,
am I correct?
What is the suggestion value of `proxy.config.cache.ram_cache.size` and `proxy.config.cache.ram_cache_cutoff`

How should I handle this case?

Thanks in advance.

Best Regards,

View raw message