Does anyone has a mail host that can be used for remote download performance?

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Does anyone has a mail host that can be used for remote download performance?

ISHIKAWA,chiaki
I am looking for a mail host that can be used for pop3 download
benchmarking, and which won't block my IP address
as being a spam source even if the e-mail messages contain illegible
text such as
the following:

Az Azz, Azz 1, 1111 zz 11:11 AA, Azzzzzz Azzzzz <[hidden email]> zzzzz:
 > Az 11/11/1111 11:11 AA, A.A. Az zzzzz:
 >> Az zzzzz zz zz zzzzzz/zzzzzzz zzzz zzzzzz AAAA_AAAAA_AAAA (),
 >> zz zzz zzz AAAAA zzzzzz zzz zzzzzzz zz zzzzzz zzzzzzzzzz.
 >> Azzzz zzz zzzz zzzz zz zz AAAA_AAAAA_AAAA () zz zzz zzzz zzz
 >> zzz zzzzzz zzzzzzzzzzz, zzz zzzzzz zzzzzzzzzz zzzzz zzzzzz zzzzz zzz
 >> zzzzzzzzzz zz zzzzzz zzzz zz zz zzzz zz zz zz.
 >
 > Az, zzzz zz z zzzz zzzzz.  Azzzzz.
 >

A zz zzzzzzzz zz zz zzzzz.


--
A.A.

Background:

I am trying to investigate the performance issues of
using a cache of open file streams for pop3 download (as is the case today),
and NOT USING it.

My guess is that for remote pop3 download, the network overhead is large
enough to mask the saving of open/close per message.

Details:

I am creating a patch to use buffered output for pop3 download : bug 1242030
TB does not use buffered output currently. This means we have causes
excessive # of write system calls when TB downloads, say, a MIME
attachment that contains jpeg file. Typically TB writes 75 octet at a
time using write system call and this is performance killer from system
perspective.
So I am creating a patch set to reduce the # of write system calls by
using buffered output.

In creating this patch for 1242030, I found the use of cached open
stream is a source code maintenance burden.
I can make the tentative patch to work  for linux and OSX, but not for
windows as of now. (There is an issue of stream not closed early enough
so that moving/removing a file succeeds: under linux and OSX, it is OK
to move or delete a file which has an open file stream to ti. Under
Windows, it is NG.) I solved this earlier with a kludge, which was ugly.
It was so ugly that jorgk wanted me to re-create the patch without using
it, and at the same time he wanted to me to honor the reusable cached
open stream. I agreed since the original klduge was certainly ugly,
Unfortunately reusable cached open stream bit caused the problem under
Windows, which has not been solved in the last month or two.

So I create an alternate patch that does away with the cache of open
stream. This alternate patch works under Windows, linux, OSX as far as I
could check using tryserver. (I only develop under linux locally.)

I wonder if I should pursue the patch that uses the cache of open
streams a bit further to see if I can make it work under Windows.
But this is going to be a few months project.

So in order to see if further development would be worth while in terms
of performance,
I wanted to compare the performance of two patch sets.
At least, I can compare the performance of the two different patches
under linux NOW.
I found that my patch set using cached open stream for pop3 download
(under linux only for testing purposes) is certainly faster than not
using it when I download messages from the mail server on the SAME linux
PC where I run TB.
This is understandable since there is not much network overhead in this
setup.

But in real world, we download e-mail messages from REMOTE mail host,
and my guess is that network overhead is large enough to mask any saving
of using a cache of open streams during pop3 download. (The cache  saves
a pair of open and close per message.)

So I want to see if the speed difference between the case of using
cached open stream and not using is large when REMOTE mail host is used.

Thus, I am in search of a free mail host that allows me to create an
account and test the performance as noted above.
Readers may wonder why I am not using gmail and other popular free mail
services. A good question.

CAVEAT: In order to make the benchmarking based on a predefined set of
messages reproducible by others, I have created a set of e-mail messages
based on real e-mail messages I received on a day last year, but
ANOMYZED them by replacing the characters in the mail text and mail headers:
[A-Z] -> A
[a-z] -> z
[0-9] -> 0
This is the strange text you see near the beginning of this message.
I am benchmarking the performance by sending and receiving these
messages. I can make the messages available once I can publish the
benchmarking result.

I am using the real world messages to reflect the
very lengthy header lines attached by services like office365 or gmail.
The reason I anomized them is to let others check my benchmark if they
want to by using the same set of e-mail messages.

But you can see that the e-mail messages contain illegible text and thus
I am afraid commercial e-mail hosts would think them as strange spams
and may block my PC (or rather IP address) as spam source.
This is something I want to avoid.
That is why I am looking for a mail host other than gmail and other well
known e-mail services and a host that would test the performance by
sending 200+ messages at a time and download it.
I want a very flexible mail host that won't throttle my mail injection
of 200+ messages in succession. This is vital.
Just trying to obtain a meaningful benchmark for a C-C binary by
repeating the test three times takes time.
If injecting messages is throttled and can't be done very quickly, it
will take a LONG TIME to obtain a benmchmark result.

TIA



_______________________________________________
dev-apps-thunderbird mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-apps-thunderbird
Reply | Threaded
Open this post in threaded view
|

Re: Does anyone has a mail host that can be used for remote download performance?

ISHIKAWA,chiaki
I found a tool called "toxiproxy" that allows me to emulate
a network delay on a local host.
I will see if this is good enough for the simulation of the communication
delay of remote host.


On 2017年08月24日 10:40, ISHIKAWA,chiaki wrote:

> I am looking for a mail host that can be used for pop3 download
> benchmarking, and which won't block my IP address
> as being a spam source even if the e-mail messages contain illegible text
> such as
> the following:
>
> Az Azz, Azz 1, 1111 zz 11:11 AA, Azzzzzz Azzzzz <[hidden email]> zzzzz:
>> Az 11/11/1111 11:11 AA, A.A. Az zzzzz:
>>> Az zzzzz zz zz zzzzzz/zzzzzzz zzzz zzzzzz AAAA_AAAAA_AAAA (),
>>> zz zzz zzz AAAAA zzzzzz zzz zzzzzzz zz zzzzzz zzzzzzzzzz.
>>> Azzzz zzz zzzz zzzz zz zz AAAA_AAAAA_AAAA () zz zzz zzzz zzz
>>> zzz zzzzzz zzzzzzzzzzz, zzz zzzzzz zzzzzzzzzz zzzzz zzzzzz zzzzz zzz
>>> zzzzzzzzzz zz zzzzzz zzzz zz zz zzzz zz zz zz.
>>
>> Az, zzzz zz z zzzz zzzzz.  Azzzzz.
>>
>
> A zz zzzzzzzz zz zz zzzzz.
>
>

_______________________________________________
dev-apps-thunderbird mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-apps-thunderbird
Reply | Threaded
Open this post in threaded view
|

Re: Does anyone has a mail host that can be used for remote download performance?

ISHIKAWA,chiaki
In reply to this post by ISHIKAWA,chiaki
On 2017年08月24日 17:33, ishikawa wrote:
> I found a tool called "toxiproxy" that allows me to emulate
> a network delay on a local host.
> I will see if this is good enough for the simulation of the communication
> delay of remote host.
>

I measured the RTT (Round Trip Time) by "ping" and observed the bandwidth of
transfer by downloading a few big e-mail attachments from my local ISP pop3
server.

RTT was 5 milliseconds (which was rather good).
Bandwidth was 720KB maximum (for small e-mail, the bandwidth was much smaller.)

Using these values, I simulated the remote network condition to connect to a
remote host using toxiproxy.
It turns out the "feel" of download matches my experience quite well.

I will post some findings to bug 1242030.

TIA

>
> On 2017年08月24日 10:40, ISHIKAWA,chiaki wrote:
>> I am looking for a mail host that can be used for pop3 download
>> benchmarking, and which won't block my IP address
>> as being a spam source even if the e-mail messages contain illegible text
>> such as
>> the following:
>>
>> Az Azz, Azz 1, 1111 zz 11:11 AA, Azzzzzz Azzzzz <[hidden email]> zzzzz:
>>> Az 11/11/1111 11:11 AA, A.A. Az zzzzz:
>>>> Az zzzzz zz zz zzzzzz/zzzzzzz zzzz zzzzzz AAAA_AAAAA_AAAA (),
>>>> zz zzz zzz AAAAA zzzzzz zzz zzzzzzz zz zzzzzz zzzzzzzzzz.
>>>> Azzzz zzz zzzz zzzz zz zz AAAA_AAAAA_AAAA () zz zzz zzzz zzz
>>>> zzz zzzzzz zzzzzzzzzzz, zzz zzzzzz zzzzzzzzzz zzzzz zzzzzz zzzzz zzz
>>>> zzzzzzzzzz zz zzzzzz zzzz zz zz zzzz zz zz zz.
>>>
>>> Az, zzzz zz z zzzz zzzzz.  Azzzzz.
>>>
>>
>> A zz zzzzzzzz zz zz zzzzz.
>>
>>
>

_______________________________________________
dev-apps-thunderbird mailing list
[hidden email]
https://lists.mozilla.org/listinfo/dev-apps-thunderbird