On 06/14/2013 01:03 PM, csn233 wrote:
On Fri, Jun 14, 2013 at 11:34 PM, Marcus Kool
<Marcus.Kool@xxxxxxxxxxxxxxx> wrote:
- more expensive (disks + battery-backed I/O controller)
Expensive disks/battery-backed are over-kill. More/adequate spindles
should do the job just as well. Why do you need a battery-backed
controller? Squid is not a transaction-based system - if you lose the
cache, tough, do "squid -z" and start again.
fast disks are good. multiple controllers and mutiple buses are good.
An EMC disk array is the most expensive and best option since Squid desires
a huge number of IOPS.
Battery-backed disk controllers are a good tradeoff: they are not so expensive
and give a reasonable performance boost.
You are missing the point completely. I'm not discussing the
performance in relation to good/bad hardware, or EMC or IOPS or
otherwise. I'm talking about how it relates to Internet access
performance. If you have good hardware, but the bottleneck is
somewhere else, you are barking up the wrong tree.
Well, sorry for misunderstanding. How can I understand that you talk/think
about disk vs internet when the text that you quote is about disks ???
- Squid uses more memory to index the disk cache (14 MB memory per GB disk
cache)
My memory allocation is only about 20-30% of that (formula), and
paging/swapping metrics doesn't indicate there is a problem. General
formulas may not always apply.
The 14 MB per GB is documented in the Squid wiki and based on the
observation that the avergae object size is 13 KB.
If you only have 20-30% of the formula you may have a larger average
object size or only use 20-30% of the confgured disk cache.
Yes, it may be documented. You forgot the IF's and MAY's. IF's and
MAY's are very important. IF it applies to you, or it MAY apply to
you, Try not to quote things without qualifications or understanding.
No. The FAQ is pretty clear about how many bytes are required per cached object.
No IF's or MAY's.
BTW: I co-authored this section of the FAQ.
The thread started with a web proxy for an ISP.
ISPs generally do not want to restart the proxy and/or rebuild the index.
It takes too long.
Don't assume. State the technical variables and let them decide.
We make assumptions with every word that we write here.
It is fair to assume that an ISP wants a stable proxy and my
comment mentioned the assumption. IMHO nothing bad with that.
No, memory is by far the fastest cache media. Since memory is
relatively cheap it is the best option.
No, it's not when it doesn't solve the problem, if your bottleneck is
somewhere else.
Eh? the "problem" of a web proxy is to serve HTTP/S requests as fast
as possible. As memory is the fastest medium for caching objects,
it is valid to go for a lot of memory.
The bottlenecks are network and disk.
The network bottleneck is the easiest: buy more bandwidth if required.
The disk bottleneck has two ways of solving:
1) do not use disks (memory cache only)
2) go for fast disks, fast controllers and/or fast disk arrays.
Ok, I stated it a bit aggressive. It should read
"Buy as much memory as your budget allows".
Wrong. Don't buy more than what you actually need.
That is your point of view. I prefer to be careful not to use
more than enough since it wastes memory.
I said the same thing.
Squid is hungry for a large number of IOPS. So get the best that
your budget can buy.
For low budgets this is a relatively cheap caching disk controller,
for high budgets it varies between low-end and high-end disk arrays
(the ones that have between 32 and 1000+ of spindles).
Nothing to do with low or high budgets. Buy what will provide benefits
in relation to cost.
Yes, that was the question of the person who started this thread:
what hardware will do?
And, as many will agree, there are a lot of IF's and MAY's and
there is no perfect answer. One can only try to help the person who
asks a question and help him decide. Explaining the need for
a fast I/O system is an important part of it. Mentioning
options for implementation are suggestions that people can
evaluate. Most likely the original poster of this thread will
do that. A simple "Buy what will provide benefits in relation
to cost" does not answer the question nor offers any help.
Marcus
