Postfix Scalability - High Volume MTA

classic Classic list List threaded Threaded
22 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Postfix Scalability - High Volume MTA

Manish Kathuria-2
Hello,

What is the extent of scalability of Postfix  ? Are there any figures
available regarding the amount of mail volume it can handle ? I am
trying to work out the number of postfix based mail servers required to
relay about 1 million messages in a day, with each message being 50 KB
in size. The MTA would be running on a Linux distribution like RHEL /
Cent OS and the messages would be generated by multiple hosts and
delivered to the relay server through SMTP for forwarding to respective
recipients.

Would a single system be capable to handle this mail volume and what
would be the hardware requirements for such a system ? I might be wrong
but I feel that a commodity box with Core 2 Duo Processor with 2-4 GB of
RAM and fast SATA / SCSI disks on RAID -1 should be able to handle at
least 50% of the target load (i.e. 500,000 mails per day). I would
appreciate your comments and suggestions.

Thanks,

Manish Kathuria




Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

JOHN ROMAN
at my company we have the same requirement.  Postfix with capable hardware can easily on a standalone box handle this much mail

with this much mail, you should consider clustering for availability

----- Original Message -----
From: Manish Kathuria <[hidden email]>
Date: Wednesday, June 4, 2008 14:13
Subject: Postfix Scalability - High Volume MTA
To: [hidden email]

> Hello,
>
> What is the extent of scalability of Postfix  ? Are there
> any figures
> available regarding the amount of mail volume it can handle ? I am
> trying to work out the number of postfix based mail servers
> required to
> relay about 1 million messages in a day, with each message being
> 50 KB
> in size. The MTA would be running on a Linux distribution like
> RHEL /
> Cent OS and the messages would be generated by multiple hosts and
> delivered to the relay server through SMTP for forwarding to
> respectiverecipients.
>
> Would a single system be capable to handle this mail volume and what
> would be the hardware requirements for such a system ? I might
> be wrong
> but I feel that a commodity box with Core 2 Duo Processor with 2-
> 4 GB of
> RAM and fast SATA / SCSI disks on RAID -1 should be able to
> handle at
> least 50% of the target load (i.e. 500,000 mails per day). I would
> appreciate your comments and suggestions.
>
> Thanks,
>
> Manish Kathuria
>
>
>
>
>
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Wietse Venema
In reply to this post by Manish Kathuria-2
Manish Kathuria:

> Hello,
>
> What is the extent of scalability of Postfix  ? Are there any figures
> available regarding the amount of mail volume it can handle ? I am
> trying to work out the number of postfix based mail servers required to
> relay about 1 million messages in a day, with each message being 50 KB
> in size. The MTA would be running on a Linux distribution like RHEL /
> Cent OS and the messages would be generated by multiple hosts and
> delivered to the relay server through SMTP for forwarding to respective
> recipients.

Without content filtering, a 5-year old single-CPU machine could
do this easily.  However, 1 million messages in a day of 50 kbyte
each means you need 10+Mbit/s of network bandwidth.

        Wietse

> Would a single system be capable to handle this mail volume and what
> would be the hardware requirements for such a system ? I might be wrong
> but I feel that a commodity box with Core 2 Duo Processor with 2-4 GB of
> RAM and fast SATA / SCSI disks on RAID -1 should be able to handle at
> least 50% of the target load (i.e. 500,000 mails per day). I would
> appreciate your comments and suggestions.
>
> Thanks,
>
> Manish Kathuria
>
>
>
>
>
>

Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Victor Duchovni
On Wed, Jun 04, 2008 at 02:29:35PM -0400, Wietse Venema wrote:

> Manish Kathuria:
> > Hello,
> >
> > What is the extent of scalability of Postfix  ? Are there any figures
> > available regarding the amount of mail volume it can handle ? I am
> > trying to work out the number of postfix based mail servers required to
> > relay about 1 million messages in a day, with each message being 50 KB
> > in size. The MTA would be running on a Linux distribution like RHEL /
> > Cent OS and the messages would be generated by multiple hosts and
> > delivered to the relay server through SMTP for forwarding to respective
> > recipients.
>
> Without content filtering, a 5-year old single-CPU machine could
> do this easily.  However, 1 million messages in a day of 50 kbyte
> each means you need 10+Mbit/s of network bandwidth.

Postfix on modern hardware, and a cooperative I/O subsystem, can handle
a million messages an hour. The bottleneck, will be one of:

        - Disk bandwidth
        - Network bandwidth
        - CPU for any content filtering

If content filters are not required, network bandwidth is available, and
remote SMTP clients are not throttling the server, a single system is
generally sufficient for this performance level.

--
        Viktor.

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the "Reply-To" header.

To unsubscribe from the postfix-users list, visit
http://www.postfix.org/lists.html or click the link below:
<mailto:[hidden email]?body=unsubscribe%20postfix-users>

If my response solves your problem, the best way to thank me is to not
send an "it worked, thanks" follow-up. If you must respond, please put
"It worked, thanks" in the "Subject" so I can delete these quickly.
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Angelos Karageorgiou
Quoting Victor Duchovni <[hidden email]>:

> If content filters are not required, network bandwidth is available, and
> remote SMTP clients are not throttling the server, a single system is
> generally sufficient for this performance level.
>


Any good souls out there have any comparison data among various MTA's

Thank you

Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Victor Duchovni
On Wed, Jun 04, 2008 at 10:31:17PM +0300, Angelos Karageorgiou wrote:

> Quoting Victor Duchovni <[hidden email]>:
>
> >If content filters are not required, network bandwidth is available, and
> >remote SMTP clients are not throttling the server, a single system is
> >generally sufficient for this performance level.
>
> Any good souls out there have any comparison data among various MTA's

Postfix is plenty fast, reproducible benchmarks in this space are very
difficult. Some commercial appliances that sacrifice reliability for
speed appear to outperform Postfix, but become fragile when pushed to
performance limits.

If you want benchmarks for your own workload, you'll need to try the
MTAs for yourself. Postfix and Exim should both perform reasonably
well, qmail should do fine, recent Sendmail with multiple queues
may do OK too.

--
        Viktor.

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the "Reply-To" header.

To unsubscribe from the postfix-users list, visit
http://www.postfix.org/lists.html or click the link below:
<mailto:[hidden email]?body=unsubscribe%20postfix-users>

If my response solves your problem, the best way to thank me is to not
send an "it worked, thanks" follow-up. If you must respond, please put
"It worked, thanks" in the "Subject" so I can delete these quickly.
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Joe-274
In reply to this post by Angelos Karageorgiou
Angelos Karageorgiou wrote:
> Quoting Victor Duchovni <[hidden email]>:
>
>> If content filters are not required, network bandwidth is available, and
>> remote SMTP clients are not throttling the server, a single system is
>> generally sufficient for this performance level.
>>
>
>
> Any good souls out there have any comparison data among various MTA's

Several years ago we were a sendmail shop and were keeping an eye on the
alternatives. IIRC qmail was just too gratuitously different, did not
offer a performance advantage, and presented its own set of quirks and
issues that prevented us from going that route. We looked at postfix and
found it to be reasonably pain-free drop-in replacement for sendmail.
This was somewhere in the 2002-2003 time frame.

We ran one particular performance stress test with default installs of
postfix and sendmail on identical OS and hardware, and gave them copies
of the same spool of mail messages to deliver. I don't have the specific
numbers with me, but I clearly remember that postfix churned through the
work in a few minutes, and the load average on the box peaked at around
2.0. After postfix was finished, the sendmail box was thrashing badly,
the load average was around 40, and it took about half an hour to finish
processing the mails that postfix had run through in a few minutes.

That sold me on the spot, and we switched our MTAs from sendmail to
postfix shortly thereafter.

Joe


Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Jake Vickers-2
In reply to this post by JOHN ROMAN
JOHN ROMAN wrote:
> at my company we have the same requirement.  Postfix with capable
> hardware can easily on a standalone box handle this much mail
>
> with this much mail, you should consider clustering for availability
>

Can anyone offer any good links for reading on clustering Postfix?  I
searched some, but all the results I came up with recommended against
using any shared file system like NFS for storing user's mailboxes
(granted I only looked at links involving Maildir format). I'd be
curious to read any articles on successful installations for this type
of cluster. I read one or two that suggested that a cluster locking
system such as that supplied by GFS or OCFS might be okay, but would be
interested in "trusted" articles (ie: not written by someone playing
with vmware and 100 email spools for load testing)
Thanks (you can email me off-list to keep clutter down or if it's not
relevant)

Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Victor Duchovni
On Wed, Jun 04, 2008 at 09:25:25PM -0400, Jake Vickers wrote:

>
> Can anyone offer any good links for reading on clustering Postfix?

Do you want to cluster the Postfix queues of active MTAs, so that messages
resident in queue directories owned by failed CPUs are processed shortly
after the failure by non-failed CPUs (this requires SAN storage and
additional instances on the adopting system to drain the inherited queue)?

Or you do you want to cluster a mail-store (delivered mail for
POP/IMAP/...  users). The latter is really not a Postfix problem. Deploy
a cluster-friendly IMAP service, and deliver to it using SMTP or LMTP.
Don't make the storage cluster more comlex by including Postfix in the
set of clustered services. Just run multiple Postfix servers in front
of the IMAP cluster.

--
        Viktor.

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the "Reply-To" header.

To unsubscribe from the postfix-users list, visit
http://www.postfix.org/lists.html or click the link below:
<mailto:[hidden email]?body=unsubscribe%20postfix-users>

If my response solves your problem, the best way to thank me is to not
send an "it worked, thanks" follow-up. If you must respond, please put
"It worked, thanks" in the "Subject" so I can delete these quickly.
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Jake Vickers-2
Victor Duchovni wrote:
On Wed, Jun 04, 2008 at 09:25:25PM -0400, Jake Vickers wrote:

  
Can anyone offer any good links for reading on clustering Postfix?
    

Do you want to cluster the Postfix queues of active MTAs, so that messages
resident in queue directories owned by failed CPUs are processed shortly
after the failure by non-failed CPUs (this requires SAN storage and
additional instances on the adopting system to drain the inherited queue)?

Or you do you want to cluster a mail-store (delivered mail for
POP/IMAP/...  users). The latter is really not a Postfix problem. Deploy
a cluster-friendly IMAP service, and deliver to it using SMTP or LMTP.
Don't make the storage cluster more comlex by including Postfix in the
set of clustered services. Just run multiple Postfix servers in front
of the IMAP cluster.

  

I was more looking at multiple Postfix severs delivering to a central storage medium. I assumed each would handle it's own active queue. You pretty much answered my question there though.
I'm curious as to clustering active queues as well and would like to read some more on that as well. Just collecting knowledge!
Thanks!
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Manish Kathuria-2
In reply to this post by Victor Duchovni


On Wed, 2008-06-04 at 14:39 -0400, Victor Duchovni wrote:

> On Wed, Jun 04, 2008 at 02:29:35PM -0400, Wietse Venema wrote:
>
> > Manish Kathuria:
> > > Hello,
> > >
> > > What is the extent of scalability of Postfix  ? Are there any figures
> > > available regarding the amount of mail volume it can handle ? I am
> > > trying to work out the number of postfix based mail servers required to
> > > relay about 1 million messages in a day, with each message being 50 KB
> > > in size. The MTA would be running on a Linux distribution like RHEL /
> > > Cent OS and the messages would be generated by multiple hosts and
> > > delivered to the relay server through SMTP for forwarding to respective
> > > recipients.
> >
> > Without content filtering, a 5-year old single-CPU machine could
> > do this easily.  However, 1 million messages in a day of 50 kbyte
> > each means you need 10+Mbit/s of network bandwidth.
>
> Postfix on modern hardware, and a cooperative I/O subsystem, can handle
> a million messages an hour. The bottleneck, will be one of:
>
> - Disk bandwidth
> - Network bandwidth
> - CPU for any content filtering
>
> If content filters are not required, network bandwidth is available, and
> remote SMTP clients are not throttling the server, a single system is
> generally sufficient for this performance level.

Thank you for your feedback. In the same context, I have a couple of
other queries. I think the kind of internet bandwidth required would be
available and we can target for much more considering the peak
requirements. Since the mail to be relayed would be fed from the clients
on different systems in the same network, would there be any performance
gains if the incoming mail is accepted through one network interface (a
private one) and the outgoing is sent out through another one (having a
public IP) ?

Is it possible to do content filtering selectively on the incoming mails
only (which would be about 50,000 per day) ?

Thanks,

Manish


Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Victor Duchovni
In reply to this post by Jake Vickers-2
On Wed, Jun 04, 2008 at 09:51:55PM -0400, Jake Vickers wrote:

> I was more looking at multiple Postfix severs delivering to a central
> storage medium. I assumed each would handle it's own active queue. You
> pretty much answered my question there though.

Yes, separate Postfix from the mail store, you need a resilient IMAP
service anyway, as Postfix will not do that for you, find one that is
suitably clustered.

> I'm curious as to clustering active queues as well and would like to
> read some more on that as well. Just collecting knowledge!

With some critical Postfix queues we trade-off for higher recoverability
and lower uptime (clusters and SANs are more complex, and have a higher
average failure rate). Recoverability is achieved by placing the active
queue on paired machines with SAN storage, and starting a drain-only
(no SMTP input, just output) Postfix instance on the adopting host to
handle messages trapped in the active queue of a dead node.

In other cases, we KISS and expect that the active queue of a real MTA
rarely holds deliverable mail long enough to worry too much about losing
a few messages if a building burns down, ... The higher simplicity of
non-clustered systems and the fact that mail enters and leaves the
systems typically in a fraction of a second, makes it reasonable to
not bother with BCP "solutions" for a non-problem.

(Compliance archive capture and delivery systems need the BCP side of
the tradeoff).

--
        Viktor.

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the "Reply-To" header.

To unsubscribe from the postfix-users list, visit
http://www.postfix.org/lists.html or click the link below:
<mailto:[hidden email]?body=unsubscribe%20postfix-users>

If my response solves your problem, the best way to thank me is to not
send an "it worked, thanks" follow-up. If you must respond, please put
"It worked, thanks" in the "Subject" so I can delete these quickly.
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Victor Duchovni
In reply to this post by Manish Kathuria-2
On Thu, Jun 05, 2008 at 07:32:06AM +0530, Manish Kathuria wrote:

> Is it possible to do content filtering selectively on the incoming mails
> only (which would be about 50,000 per day) ?

Most modern systems have network interfaces that outperform the disk.
If you have a 1Gbps NIC, you are unlikely to need two of them. If
your peak output saturates 100 Mbps you may need two NICs, but who
will be accepting mail from your single machine at this rate?

Note that will full-duplex interfaces, incoming mail and outgoing mail
don't compete for the same bandwidth, these are mostly separate (ACKs
aside) halves of a full-duplex channel.

--
        Viktor.

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the "Reply-To" header.

To unsubscribe from the postfix-users list, visit
http://www.postfix.org/lists.html or click the link below:
<mailto:[hidden email]?body=unsubscribe%20postfix-users>

If my response solves your problem, the best way to thank me is to not
send an "it worked, thanks" follow-up. If you must respond, please put
"It worked, thanks" in the "Subject" so I can delete these quickly.
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Devdas Bhagat
In reply to this post by Manish Kathuria-2
On Wed, Jun 04, 2008 at 11:42:53PM +0530, Manish Kathuria wrote:
> Hello,
>
> What is the extent of scalability of Postfix  ? Are there any figures

Postfix scales plenty well. I have a couple of spamtraps currently
running at 800 messages/minute each of throughput, along with a 90%+
rejection rate on email already being blacklisted by other means.

uptime:
 04:13:51 up 138 days, 21:29,  3 users,  load average: 3.50, 4.05, 5.04

This is a single SATA disk, with 2 GB of RAM running a dual xeon at
2.4GHz. This host does no content filtering, merely relays mail to be
further processed. 800 concurrent smtp listeners. I could tune this
further, but that task isn't on my priority list right now.

> available regarding the amount of mail volume it can handle ? I am
> trying to work out the number of postfix based mail servers required to
> relay about 1 million messages in a day, with each message being 50 KB

This is a fairly trivial system. What you need to be looking at is peak
loading (one million messages being injected in an hour is different
from them being evenly injected through the day).

Devdas Bhagat
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Manish Kathuria-2


On Wed, 2008-06-04 at 23:38 -0500, Devdas Bhagat wrote:

> On Wed, Jun 04, 2008 at 11:42:53PM +0530, Manish Kathuria wrote:
> > Hello,
> >
> > What is the extent of scalability of Postfix  ? Are there any figures
>
> Postfix scales plenty well. I have a couple of spamtraps currently
> running at 800 messages/minute each of throughput, along with a 90%+
> rejection rate on email already being blacklisted by other means.
>
> uptime:
>  04:13:51 up 138 days, 21:29,  3 users,  load average: 3.50, 4.05, 5.04
>
> This is a single SATA disk, with 2 GB of RAM running a dual xeon at
> 2.4GHz. This host does no content filtering, merely relays mail to be
> further processed. 800 concurrent smtp listeners. I could tune this
> further, but that task isn't on my priority list right now.
> Devdas Bhagat

Thanks. The throughput numbers are quite similar to my requirements.
This is great and the first hand statistics I was looking for. Which
distribution and postfix version are you using ? I think there were some
file descriptor limitations on Linux since Postfix does not support
tuning FD_SETSIZE on glibc systems.

Are there any benchmarks / data for a high volume content filtering
system running postfix and amavisd-new ?

--
Manish

Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Devdas Bhagat
On Thu, Jun 05, 2008 at 10:48:21AM +0530, Manish Kathuria wrote:
>
> Thanks. The throughput numbers are quite similar to my requirements.
> This is great and the first hand statistics I was looking for. Which
> distribution and postfix version are you using ? I think there were some

Red Hat Enterprise Linux ES release 4 (Nahant Update 5)
postfix-2.4.6-1.rhel4

amavisd-new will generally introduce a fairly steep drop in your throughput,
depending on the options you have configured.

I suggest assuming a factor of 4 per SA instance. I am going to
investigate using OpenBSD instead of Linux for the traps.

Devdas Bhagat
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Stefan Förster-4
In reply to this post by Manish Kathuria-2
* Manish Kathuria <[hidden email]> wrote:

> On Wed, 2008-06-04 at 23:38 -0500, Devdas Bhagat wrote:
> > Postfix scales plenty well. I have a couple of spamtraps currently
> > running at 800 messages/minute each of throughput, along with a 90%+
> > rejection rate on email already being blacklisted by other means.
> >
> > uptime:
> >  04:13:51 up 138 days, 21:29,  3 users,  load average: 3.50, 4.05, 5.04
> >
> > This is a single SATA disk, with 2 GB of RAM running a dual xeon at
> > 2.4GHz. This host does no content filtering, merely relays mail to be
> > further processed. 800 concurrent smtp listeners. I could tune this
> > further, but that task isn't on my priority list right now.
> > Devdas Bhagat
>
> Thanks. The throughput numbers are quite similar to my requirements.
> This is great and the first hand statistics I was looking for. Which
> distribution and postfix version are you using ? I think there were some
> file descriptor limitations on Linux since Postfix does not support
> tuning FD_SETSIZE on glibc systems.

Postfix supports the use of kqueue(2), /dev/epoll and epoll(4). Have a
look at http://www.postfix.org/TUNING_README.html#file_limit for more
information.


Cheers
Stefan
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Angelos Karageorgiou
In reply to this post by Jake Vickers-2
Jake Vickers wrote:
>
> I was more looking at multiple Postfix severs delivering to a central
> storage medium. I assumed each would handle it's own active queue. You
> pretty much answered my question there though.
> I'm curious as to clustering active queues as well and would like to
> read some more on that as well. Just collecting knowledge!
> Thanks!



I am using 3 front end MXs with their raid1 for queuing and amavisd-new
for content scanning all delivering to an nfs server that holds the
maildirs.

The NFS server has raid 5  configuration which has proven the best for
maildirs since the NFS server spends 90% of the time sending data to the
clients under the above configuration.

Not a single glitch for approx. 100,000 accounts 1 1/2 years running.

I will go as far as saying that you can probably scale out this
configuration until you saturate the NFS network link.

angelos.vcf (543 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Manish Kathuria-2


On Thu, 2008-06-05 at 09:48 +0300, Angelos Karageorgiou wrote:

> Jake Vickers wrote:
> >
> > I was more looking at multiple Postfix severs delivering to a central
> > storage medium. I assumed each would handle it's own active queue. You
> > pretty much answered my question there though.
> > I'm curious as to clustering active queues as well and would like to
> > read some more on that as well. Just collecting knowledge!
> > Thanks!
>
>
>
> I am using 3 front end MXs with their raid1 for queuing and amavisd-new
> for content scanning all delivering to an nfs server that holds the
> maildirs.
>
> The NFS server has raid 5  configuration which has proven the best for
> maildirs since the NFS server spends 90% of the time sending data to the
> clients under the above configuration.

Which IMAP / POP3 server do you use ?

>
> Not a single glitch for approx. 100,000 accounts 1 1/2 years running.

What is the total message volume  in terms of messages received per day
and delivered per day ?


--
Manish

Reply | Threaded
Open this post in threaded view
|

Re: Postfix Scalability - High Volume MTA

Simone Felici
Manish Kathuria ha scritto:

>
> On Thu, 2008-06-05 at 09:48 +0300, Angelos Karageorgiou wrote:
>> Jake Vickers wrote:
>>> I was more looking at multiple Postfix severs delivering to a central
>>> storage medium. I assumed each would handle it's own active queue. You
>>> pretty much answered my question there though.
>>> I'm curious as to clustering active queues as well and would like to
>>> read some more on that as well. Just collecting knowledge!
>>> Thanks!
>>
>>
>> I am using 3 front end MXs with their raid1 for queuing and amavisd-new
>> for content scanning all delivering to an nfs server that holds the
>> maildirs.
>>
>> The NFS server has raid 5  configuration which has proven the best for
>> maildirs since the NFS server spends 90% of the time sending data to the
>> clients under the above configuration.
>
> Which IMAP / POP3 server do you use ?
>
>> Not a single glitch for approx. 100,000 accounts 1 1/2 years running.
>
> What is the total message volume  in terms of messages received per day
> and delivered per day ?
>
>
> --
> Manish
>
>

If this could help, this is my environment:
We have 3 frontend servers (3 MX with same priority) for more or less 100.000 mailboxes.
My front-end servers (3) are poor on HW, need to replace them (Celeron 2GHz - 1GB RAM).
BTW My servers process more or less 6.500.000 mails per day (all together). Of them ONLY 170.000 mails are accepted, the rest is, in order:
- blocked by RBLs (65%)
- blocked by greylist (30%)
- blocked by other causes (not found, mailbox full, other reasons) (5%)
Mailboxes are stored on a SAN (storage area network). The central server is not postfix, due migrtion issue... In the next time I'll change it.
Front-End postfix 2.3.3.2 with mysql support.

Bye, Simon





--
Simone Felici                    E-Mail: [hidden email]
Divisione Tecnica                Tel:    0461 030 111
Alpikom S.p.A.                   Fax:    0461 030 112
v.Fersina, 23 - 38100 Trento     URL:    http://www.alpikom.it
12