I must say... I have been looking forward to Stalwart v1.0 so much - and this blog post just didn't sell it for me.
It focuses on speed, and on the optimisation of db's (4 different databases!). Granted, the two things that they wanted to sort out before the big v1. However, as a big announcement, as a release-party theme, at the biggest and best selling time for v1.0, I feel they missed the boat a bit with the post. At least for me.
The issue probably stems from the fact that their front page reads like Stalwart is geared towards self-hosters. They are a nice target audience during development. They will test, slurp up every new feature, and give ready feedback.
The other part of Stalwart is the enterprise world, where they probably want to compete with Microsoft Exchange and such. That is, following Microsoft's business model (unless I am sorely mistaken), where the money is to be made.
And I am saying this with no bitterness, sarcasm, negativity at all (forgive me if it comes across as such, English is my third language...)
So, for me, as a self-hoster, this v1.0 announcement left me feeling a bit like I was not the target.
My beef - as a self-hoster - with the traditional mailer systems (and entourage) is not so much speed. Or db's. It's the fact that I have to install so many different components, and configure them, each with their own configuration language(!).
It's the fact that, besides SMTP and IMAP, I need to also do HTTP for a certificate.
It's the fact that I have to somehow get DKIM keys (or Google won't accept my mail). SPF? DMARC? DANE (which means: DNSSEC) and/or MTA-STS (which also means HTTP)? What else is there? DKIM2 on the horizon, in any event.
It's the difficulty of keeping spam out (not my biggest problem to be honest).
And, most of all, once the above is finally done and dusted: Getting my emails delivered to large mailer recipients.
And then, every so many years, an OS upgrade, with upgrades to all those components, and possibly new configurations.
In short, my beef is that I have to be a pretty (no comma!) all-round Unix sysadmin to run a mail server these days.
Look, I like doing that, to be honest, even if my official education / degree is not in IT. I like the Unix way of one software does one thing, and (I hope) does it well. But, I don't always have that time any more; and, reluctantly, I have been eyeing Stalwart with high-ish hopes. Because of speed? Because it hides four different types of databases in that one binary? Nah. Because it is easy. Because it offers it all in one, single, binary. Because, for crying out loud, it offers JMAP! Because I don't have to rotate certs. Or keys. Or any of that rot. Most of it is done for me. It's like outsourcing the sysadmin part, and still keeping my mail on my server.
(Btw, I cringe while writing this! What have I become? I guess a husband, dad-of-three, business owner... A lot has changed during those years since first I started poking around unix in the late 90's.)
From the front page, one would get the impression that Stalwart is meant for the self-hoster. After all, it reads: "One server for your email, calendars, contacts and shared files, with spam and phishing protection built in. Every standard protocol your clients already speak (JMAP, IMAP, POP3, SMTP, CalDAV, CardDAV and WebDAV) supported natively."
I wish this blog post, selling the v1.0, would have talked about just that. How it is 1 binary to install, that does everything. How it sorts out my certs, my keys for DKIM, ARC, SPF, what have you. How it offers not just mail, but calendar, address books, shared files... You know, the things we sort of have come to expect from a mail solution. And how it offers new shiny stuff, like JMAP. Oh, and it is also much faster - one can always throw that in.
And then there could have been part of that post also being for Enterprises (of two different posts?). On how they squeezed out every last bit of slowness and friction. How they optimised all the different databases. How you can just connect a bunch of Stalwart servers for redundancy. Or how emails are not stored in plain text, but in databases that can be just anywhere and everywhere. No problem at all, cool stuff!
For months and years I have been saying: There is "too much" activity on Stalwart's GitHub; too many commits; too many breaking changes between those pre-v1 releases. Understandably so! I have to wait until v1. Well, here it is! And, undoubtedly, one day I will give it a try. I already have a VPS ready :-)
I have to install so many different components, and configure them
Same but after 25 years of self-hosting email, I am not that interested in trying to migrate tens of gigabytes of email messages to Postgres. If they add a Maildir backend, it might be a plausible alternative.
(modulo all the weird hacks that 25 years have accreted on the top which are the main thing preventing me from ripping everything out and simplifying/modernising the stack.)
theperfecthobbit | 12 hours ago
I must say... I have been looking forward to Stalwart v1.0 so much - and this blog post just didn't sell it for me.
It focuses on speed, and on the optimisation of db's (4 different databases!). Granted, the two things that they wanted to sort out before the big v1. However, as a big announcement, as a release-party theme, at the biggest and best selling time for v1.0, I feel they missed the boat a bit with the post. At least for me.
The issue probably stems from the fact that their front page reads like Stalwart is geared towards self-hosters. They are a nice target audience during development. They will test, slurp up every new feature, and give ready feedback.
The other part of Stalwart is the enterprise world, where they probably want to compete with Microsoft Exchange and such. That is, following Microsoft's business model (unless I am sorely mistaken), where the money is to be made.
And I am saying this with no bitterness, sarcasm, negativity at all (forgive me if it comes across as such, English is my third language...)
So, for me, as a self-hoster, this v1.0 announcement left me feeling a bit like I was not the target.
My beef - as a self-hoster - with the traditional mailer systems (and entourage) is not so much speed. Or db's. It's the fact that I have to install so many different components, and configure them, each with their own configuration language(!). It's the fact that, besides SMTP and IMAP, I need to also do HTTP for a certificate. It's the fact that I have to somehow get DKIM keys (or Google won't accept my mail). SPF? DMARC? DANE (which means: DNSSEC) and/or MTA-STS (which also means HTTP)? What else is there? DKIM2 on the horizon, in any event. It's the difficulty of keeping spam out (not my biggest problem to be honest). And, most of all, once the above is finally done and dusted: Getting my emails delivered to large mailer recipients. And then, every so many years, an OS upgrade, with upgrades to all those components, and possibly new configurations.
In short, my beef is that I have to be a pretty (no comma!) all-round Unix sysadmin to run a mail server these days.
Look, I like doing that, to be honest, even if my official education / degree is not in IT. I like the Unix way of one software does one thing, and (I hope) does it well. But, I don't always have that time any more; and, reluctantly, I have been eyeing Stalwart with high-ish hopes. Because of speed? Because it hides four different types of databases in that one binary? Nah. Because it is easy. Because it offers it all in one, single, binary. Because, for crying out loud, it offers JMAP! Because I don't have to rotate certs. Or keys. Or any of that rot. Most of it is done for me. It's like outsourcing the sysadmin part, and still keeping my mail on my server.
(Btw, I cringe while writing this! What have I become? I guess a husband, dad-of-three, business owner... A lot has changed during those years since first I started poking around unix in the late 90's.)
From the front page, one would get the impression that Stalwart is meant for the self-hoster. After all, it reads: "One server for your email, calendars, contacts and shared files, with spam and phishing protection built in. Every standard protocol your clients already speak (JMAP, IMAP, POP3, SMTP, CalDAV, CardDAV and WebDAV) supported natively."
I wish this blog post, selling the v1.0, would have talked about just that. How it is 1 binary to install, that does everything. How it sorts out my certs, my keys for DKIM, ARC, SPF, what have you. How it offers not just mail, but calendar, address books, shared files... You know, the things we sort of have come to expect from a mail solution. And how it offers new shiny stuff, like JMAP. Oh, and it is also much faster - one can always throw that in.
And then there could have been part of that post also being for Enterprises (of two different posts?). On how they squeezed out every last bit of slowness and friction. How they optimised all the different databases. How you can just connect a bunch of Stalwart servers for redundancy. Or how emails are not stored in plain text, but in databases that can be just anywhere and everywhere. No problem at all, cool stuff!
For months and years I have been saying: There is "too much" activity on Stalwart's GitHub; too many commits; too many breaking changes between those pre-v1 releases. Understandably so! I have to wait until v1. Well, here it is! And, undoubtedly, one day I will give it a try. I already have a VPS ready :-)
ptman | 11 hours ago
mox was really easy to set up and took care of most things or guided me through them
zimpenfish | 10 hours ago
Same but after 25 years of self-hosting email, I am not that interested in trying to migrate tens of gigabytes of email messages to Postgres. If they add a Maildir backend, it might be a plausible alternative.
(modulo all the weird hacks that 25 years have accreted on the top which are the main thing preventing me from ripping everything out and simplifying/modernising the stack.)
pl | 4 minutes ago
You don't have to install PostgreSQL, I'm using the RocksDB backend that simply comes as part of the one executor you need to run.
zie | 12 hours ago
Apparently 250k messages is a lot? My unread mail count in my work inbox is 226k alone, guess I'm a large mailbox :)
pl | 11 hours ago
So
(1000/4)*25000= 62.5 Million in a Second does sound like a lot if I'm not mistaken?mccd | 14 hours ago
The website feels extremely AI generated.