postfix load handling
in [Postfix]
|
Prev: How to reject bad hosts
Next: statvfs errors
From: Bjorn Mork on 10 Aug 2010 04:05 Hi, can POSTFIX handle load of 120k mailboxes....What would be required additional with postfix to bear such load.... BMork
From: Patrick Ben Koetter on 10 Aug 2010 04:51 * Bjorn Mork <bjron.mork(a)gmail.com>: > can POSTFIX handle load of 120k mailboxes....What would be required > additional with postfix to bear such load.... You can run multiple Postfix machines to deal with the load. But more imporant with all, but especially with 120k mailboxes the questions are not load alone: - What will be incoming traffic (average/peak)? - What will be outgoing traffic (average/peak)? - What mail management will/needs to be done? - Anti-Spam? - Anti-Virus? - Other Policies? - server side mailbox filtering - What's the average mailbox size? - What mailbox type will you use? - NFS or local disks? - Which RAID level? - How many disks will be in a RAID array? p(a)rick -- All technical questions asked privately will be automatically answered on the list and archived for public access unless privacy is explicitely required and justified. saslfinger (debugging SMTP AUTH): <http://postfix.state-of-mind.de/patrick.koetter/saslfinger/>
From: Bjorn Mork on 10 Aug 2010 06:24 p(a)tric, i have tried to answer your queris, (Please correct, if I am wrong in understanding your question...) We do have multiple IBM Blade server with 2.4 Xeon + 16GB + NAS over iSCSI protocol...... How many blades will be involved for such load....??? Can you please suggest its logical architecture ??? > - What will be incoming traffic (average/peak)? > *its a assumption, that during peak hours avg 60k-70k pop session expected... * > * *- What will be outgoing traffic (average/peak)? > *May be same or can have more value than pop sessions.... * > - What mail management will/needs to be done? > *normal management like postfixadmin type...* > - Anti-Spam? > *MailScanner + Spamassassin * > - Anti-Virus? > * MailScanner + ClamAV * > - Other Policies? > *mostly default values, but major reqirement is quotas , few accounts with 2 GB and few with 100 GB.....* > - server side mailbox filtering > *I think you are taking about squiral / courier mail.... ???* > - What's the average mailbox size? > *few accounts with 2GB and few with more space like 100GB.....* > - What mailbox type will you use? > *maildir /var/spool/mail/* > - NFS or local disks? > *NAS is available with us over iSCSI protocol......* > - Which RAID level? > *RAID-1 , for OS + Application level , because planning to store /var/www/html and /var/spool/mail on NAS* > - How many disks will be in a RAID array? > > p(a)rick > > > > -- > All technical questions asked privately will be automatically answered on > the > list and archived for public access unless privacy is explicitely required > and > justified. > > saslfinger (debugging SMTP AUTH): > <http://postfix.state-of-mind.de/patrick.koetter/saslfinger/> >
From: Patrick Ben Koetter on 10 Aug 2010 07:37 * Bjorn Mork <bjron.mork(a)gmail.com>: > i have tried to answer your queris, (Please correct, if I am wrong in > understanding your question...) > > We do have multiple IBM Blade server with 2.4 Xeon + 16GB + NAS over iSCSI > protocol...... > > How many blades will be involved for such load....??? I would have to speculate and I won't because it would be bad advice. Instead setup a test machine and use the Postfix tools smtp-sink and smtp-source and the benchmark program "mstone" to simulate your usage profile. Find out within which limits you can run your setup safely (calculate about 30% extra load on top) on one machine and then find out how many machines you will need to serve your overall load. > Can you please suggest its logical architecture ??? I, personally, think running a pair of identical Postfix machines that can failover each other is a good thing. Those machines are backed up with another pair of fallback servers. Their job is to take and deliver any message that could not be delivered at first attempt from the primary pair. > > - What will be incoming traffic (average/peak)? > > > *its a assumption, that during peak hours avg 60k-70k pop session > expected... Just for the books: Postfix is not a POP server. POP access has a different load consumption profile than a SMTP server. Do you have any figures how many messages your users send via SMTP when they pick up message via POP? > * > > > * *- What will be outgoing traffic (average/peak)? > > > *May be same or can have more value than pop sessions.... > * > > > > - What mail management will/needs to be done? > > > *normal management like postfixadmin type...* > > > > - Anti-Spam? > > > *MailScanner + Spamassassin > * > > > - Anti-Virus? > > > * MailScanner + ClamAV > * I don't use MailScanner in combination with Postfix and I don't know how that combination performs and/or scales. It might be hard to get advice on this on the Postfix mailing list because running MailScanner with Postfix is not blessed - AFAIK MailScanner bypasses Postfix queue management and uses Postfix internal API, which might change without notice leaving your system unusable after an update. > > - Other Policies? > > > *mostly default values, but major reqirement is quotas , few > accounts with 2 GB and few with 100 GB.....* You probably want a mechanism that can tell a recpients quota _during_ the SMTP session and rejects the message immediately if a mailbox is over quota, instead of accepting the message, trying to deliver it, failing and bouncing it. That would eat to many ressources when you need to watch out you don't waste 'em. > > - server side mailbox filtering > > > *I think you are taking about squiral / courier mail.... ???* I am thinking procmail, maildrop, deliver, sieve-rules etc. > > - What's the average mailbox size? > > > *few accounts with 2GB and few with more space like 100GB.....* That will have an impact on POP and IMAP sessions, but should not be a problem for SMTP. It will be a challenge for backup though... > > - What mailbox type will you use? > > > *maildir /var/spool/mail/* If you use dovecot take a look at the dbox format. It should give you a performance boost on POP/IMAP. > > - NFS or local disks? > > > *NAS is available with us over iSCSI protocol......* > > > > - Which RAID level? > > > *RAID-1 , for OS + Application level , because planning to > store /var/www/html and /var/spool/mail on NAS* What's your failover strategy? p(a)rick -- All technical questions asked privately will be automatically answered on the list and archived for public access unless privacy is explicitely required and justified. saslfinger (debugging SMTP AUTH): <http://postfix.state-of-mind.de/patrick.koetter/saslfinger/>
From: Stan Hoeppner on 10 Aug 2010 10:01
Patrick Ben Koetter put forth on 8/10/2010 6:37 AM: > * Bjorn Mork <bjron.mork(a)gmail.com>: >> i have tried to answer your queris, (Please correct, if I am wrong in >> understanding your question...) >> >> We do have multiple IBM Blade server with 2.4 Xeon + 16GB + NAS over iSCSI >> protocol...... >> >> How many blades will be involved for such load....??? 16GB should be able to handle an insane number of simultaneous SMTPd processes especially if you use Postscreen, which I'd highly recommend with the user count you're talking about. This is highly dependent on your Postfix restrictions, anti spam daemons and virus scanners. Postfix, and SMTP in general is not CPU bound, so with dual 2.4 GHz CPUs, even if they are the "NetBurst" architecture, you'll run out of other resources before CPU bandwidth. The main one being... Storage, the most important aspect of your inquiry. You mentioned iSCSI and NFS, which at face value doesn't make sense. You don't connect an NFS client to an NFS server over iSCSI. iSCSI is for block level disk access and functions almost identically to a fiber channel SAN. iSCSI works at the block device level underneath the filesystem. NFS works _at_ the filesystem level. I'll assume you're actually using NFS, not iSCSI. Given your load, you'll need at least 2 GbE links on the Postfix servers--1 for inbound SMTP connections, and other for access to the NFS server, and these should be on separate ethernet and IP networks (different switches and subnets from the incoming smtp traffic) for security and other reasons. SMTP mail is primarily a disk bound application. You need a very high seek rate--high IOPs--as the mail flow is going to be lots of small files, including queue files and files permanently written to maildir on the NFS server. Would you please post the NFS/NAS server manufacturer, NFS/NAS server model number, number of disks and their RAID (stripe) level, and what types of disks--i.e. fiber channel, SCSI, SAS, SATA, etc? The random IOPs performance of your NFS/NAS server will be the overall performance limiting factor, if it's not sufficiently beefy. -- Stan |