I'm with pe1chi and the log clearing camp on this one. Basic on-box log management functionality should be brought up to par with equivalent products.
- Log rotation
- Ability to clear logs locally / manually if needed
Two features, not asking the world of the developers for that.
The other side of the coin:
That said. Any organization of any size especially int the business of providing network services should have at least a syslog server and a SNMP based monitoring solution. Any decent syslog server will let you rotate logs to match the size of disk available. Also, a 8TB SATA drive costs $230 on Amazon here in the US. If you can afford 800 MikroTiks you can afford a hard drive to cram an awful lot of syslog onto.
My MikroTik's are not chatty at all with default syslog settings so in reality you can probably go pick up a Raspberry Pi, a 16gb SD card, an Ethernet cable and some duct tape and tape it under someones desk and be at least twice as good at troubleshooting.
I can't tell you how many organizations I've been to that will complain to no end about how crappy their gear, network or life is but have 0 monitoring. Examples are the best so I'll tell one of many stories that sold me on monitoring. I'm sitting in a meeting when I worked in an enterprise environment with our systems guys and the boos. I'm the network guy. The systems guy says "hey we need to upgrade the load balancers. Why don't we go with the virtual appliances." I say heck ya! That sounds great, physical hardware is pain and is way more expensive. Great we both agree. Lets pick a size. The systems guy, bless his heart, goes, "well the old physical ones have 1 GB interfaces so we should at least get the 1 GB symmetrical license for the virtual appliance." I say well hold on the licensing can be upgraded at anytime non-disruptively. Why don't we see what we are actually using out of that 1 GB. After the initial surprise that this was even being monitored we logged into ZenOSS, what we ran at the time, and had a look. The entire box had seen a fairly sustained load of less than 5 mbps with peaks up to 10 mbps. My boss looked at the pricing and chose the 200 mbps model which was still way overkill and saved the organizations 10s of thousands of dollars both up front and year over year in maintenance.
Most people say oh well you should've known that anyways or "I could've looked at the interfaces live and told you that." The truth is you couldn't have at least with enough of a straight face to tell your boss. Monitoring is good for detecting outages and can be an excellent planning tool. It can identify out of date hardware or software and as we just learned help plan for capacity. More advanced tools can even help you manage those 800 devices through automation.