That could be, but I'd rather they made the case. When SMTP came out, it was indeed very simple. In college in the 1980s, I could send email just by doing something like:
% telnet mail.umich.edu 25
MAIL FROM: <william@umich.edu>
RCPT TO: <joe.blow@umich.edu>
DATA
From: William Pietri <william@umich.edu>
To: My Skeptical Friend <joe.blow@umich.edu>
Subject: Who needs a mail client?
You can send email without a mail client!
.
It's harder now, because spammers. But to me it's amazing that something like has been working for nearly 40 years. For perspective, it came out the same year as the Commodore 64.
And because cybercrime, which is now doing staggering damage, much of it initiated via SMTP.
SMTP was designed when the Internet's topology and population were vastly different, and we desperately need to replace it, e.g. with TMTP from the "mnm" open source project (my work):
Sorry folks, I thought it was OK to resubmit links that didn't make the front page.
I never got a "Show HN" link on the prestigious Show page because someone else posted it and made the front page. So I tried some other URLs.
I don't think there's an issue with posting relevant comments with a link, but I agree I should disclose authorship.
--- Flagged post above ---
And because cybercrime, which is now doing staggering damage, much of it initiated via SMTP.
SMTP was designed when the Internet's topology and population were vastly different, and we desperately need to replace it, e.g. with TMTP from the "mnm" open source project: [link]
There is definitely an issue with posting your project over and over and over, because that's a sign you see it as relevant even when other people don't. That's certainly the case here. If you wanted to make the "and cybercrime" point, there are a zillion more relevant links than yours.
That looks like spammer behavior to me, which makes your desire to replace email absurd.
I made a stupid typo, once SMTP gets in your brain it's hard to get it to rattle its way back out I guess. It's corrected now but the old link continues to work.
That's true, but then the next generation of technology would have to be named "Simpler". Names work best when they're non-judgmental. Simple, good, fast, etc. are things you want to avoid.
Are there any sane solutions for managing lots of MIBs? In my experience the largest pain with SNMP comes from the fact that vendors usually provide their MIBs as a single huge pile of junk containing some random version of all the dependencies as well, with completely random file naming conventions.
This means that trying to use two vendors' MIBs together with eg. snmptrapd basically leads to undefined behaviour when translating traps. I don't know if there any way to actually tell which definitions will take precedence or whether the definition is correct.
It also doesn't help that often the trap's translation is just as cryptic as the bare OID.
The state of Cisco's MIB downloads kind of shows the state of things, it's just a very dusty, forgotten corner of their website that links directly to FTP. Part of me doesn't want to complain but browsers don't tend to have native FTP anymore so this becomes a minor pain, and certainly tells you that Cisco has just about forgotten about this.
In general no vendors seem to put much effort into their MIBs and that produces a lot of the frustration around these.
Some vendors can not get even the simple SNMP right and will send OIDs in wrong order in WALK, causing a standard-observing SNMP client to prematurely end a subtree walk or to loop forever.
And the SNMP is mostly read-only, usually only basic functions can be controlled (such as interface enable/disa le) but more complex configuration can be only read (vlans, firewall rules, etc.)
So really the SNMP support in datasheet means nothing.
In my experience the S in SNMP is anything but simple. Conceptually it is simple, but once you add MIBs in (which describe the data) it gets really complicated.
SNMP support in actual devices seems to be more of a box ticking exercise than anything else now-a-days. It works for the basics but trying to do anything more complicated reveals a lot of bugs in the implementation. I think there is a lot of importing an SNMP library and calling it done SNMP implementations in devices!
Take the Q-BRIDGE-MIB for an example - one I know well. That is a standard implemented by switches so they can configure VLANs for you which sounds great. Does it work? Well sort of - switch manufacturers have invented a lot of ways of getting the implementation wrong so you end up with a translation layer for every switch type.
The security of SNMP is terrible too. In theory SNMPv3 has better security, but very few devices actually implement it, and it is, you guessed it, really really complicated. Only run SNMP on a private (non internet connected) network!
as someone who has worked with SNMP for the past 5 years and who has written SNMP to websocket bridges and SNMP device simulators, SNMP was simple upto V2C, they made it extremely complex with V3 and they have shoved too many things into what was once a simple protocol.
also, SNMP is simple due to the fact that it is built mostly over UDP and that there are no acknowledgements for any action you perform.
Netconf is a much nicer interface to work with.
I know that 99.9% of your readers know exactly what SNMP is (even though SMTP is in the title), but when you write a post like this, please just put in what the initialism is actually short for.
41 comments
[ 4.6 ms ] story [ 91.8 ms ] thread>>> 2021-05-01 simple as in smtp
I suspect the idea is that neither smtp nor snmp are simple when you dig down, despite the name.
And I'd agree with the article, mibs can be maddening
And for snmpv3:
"if you are using SNMPv3 where the authentication and configuration can be amazingly, maddeningly complex for some vendors."
Indeed
The curious can read the inital RFC: https://tools.ietf.org/html/rfc821
And because cybercrime, which is now doing staggering damage, much of it initiated via SMTP.
SMTP was designed when the Internet's topology and population were vastly different, and we desperately need to replace it, e.g. with TMTP from the "mnm" open source project (my work):
https://mnmnotmail.org/
I never got a "Show HN" link on the prestigious Show page because someone else posted it and made the front page. So I tried some other URLs.
I don't think there's an issue with posting relevant comments with a link, but I agree I should disclose authorship.
--- Flagged post above ---
And because cybercrime, which is now doing staggering damage, much of it initiated via SMTP.
SMTP was designed when the Internet's topology and population were vastly different, and we desperately need to replace it, e.g. with TMTP from the "mnm" open source project: [link]
That looks like spammer behavior to me, which makes your desire to replace email absurd.
https://b3ta.com/newsletter/issue768/
Feels the same way like horrible regimes usually have a lot of positive words in their name.
Although I understand most of them are relatively simple compared to their predecessors rather than being truly simple or minimalistic.
This means that trying to use two vendors' MIBs together with eg. snmptrapd basically leads to undefined behaviour when translating traps. I don't know if there any way to actually tell which definitions will take precedence or whether the definition is correct.
It also doesn't help that often the trap's translation is just as cryptic as the bare OID.
In general no vendors seem to put much effort into their MIBs and that produces a lot of the frustration around these.
Well-written and informative.
And the SNMP is mostly read-only, usually only basic functions can be controlled (such as interface enable/disa le) but more complex configuration can be only read (vlans, firewall rules, etc.)
So really the SNMP support in datasheet means nothing.
The S stands for Simple http://harmful.cat-v.org/software/xml/soap/simple
SNMP support in actual devices seems to be more of a box ticking exercise than anything else now-a-days. It works for the basics but trying to do anything more complicated reveals a lot of bugs in the implementation. I think there is a lot of importing an SNMP library and calling it done SNMP implementations in devices!
Take the Q-BRIDGE-MIB for an example - one I know well. That is a standard implemented by switches so they can configure VLANs for you which sounds great. Does it work? Well sort of - switch manufacturers have invented a lot of ways of getting the implementation wrong so you end up with a translation layer for every switch type.
The security of SNMP is terrible too. In theory SNMPv3 has better security, but very few devices actually implement it, and it is, you guessed it, really really complicated. Only run SNMP on a private (non internet connected) network!
You're talking about NFS, right?
I've never had to work with it and didn't know!