Regular Member
EricE
Posts: 580
Registered: ‎01-10-2012
Posts: 580
Kudos: 121
Solutions: 2
Registered: 01-10-2012

Re: Tolly report

The official Mikrotik support offering is direct email. Slamming the whole company because of their forum community is not a fair criticism of their support channel.


Sure it is - when it's a company sponsored forum and the companies employees are part of the problem!

I don't buy products these days, especially when they are for myself or smaller clients, if there isn't a strong and friendly community. Often I can get far better and far faster support from a friendly and knowledgeable community. And as others posted, having a vibrant community becomes your best marketing engine too. I never hear anything about MT other than price. Well, if that's all you have eventually someone is going to come along and take that away from you.

Case in point - I dabbled with MT because of the price, but I held my nose through the whole experience. In the end the cheap hardware wasn't worth the overall experience to me - luckily I don't need that many boxes, so the extra cost in building a decent atom box to run pfSense is worth my sanity.

EdgeMax looks to be even better for some uses - and the friendly and very knowledgeable forum here at UBNT is a definite plus!

I know exactly what kind of an operation MT is, and they will deserve exactly what they get in the long term unless they change their ways. It's common sense and anyone who is over 30 can see the train coming. There are some patterns of human behavior that are just well known - it's not rocket science! I do give MT some credit for things like developing their own MPLS stack, so they aren't the usual quick win artists that seem to spring up around open source, but they aren't far. Checking their ego and engaging their customers at a more base level would go a long way towards fostering community good will. The very definition of "too smart for your own good"....
SuperUser
mhoppes
Posts: 14,014
Registered: ‎06-23-2010
Posts: 14014
Kudos: 4189
Solutions: 59
Registered: 06-23-2010

Re: Tolly report

Not only that....have you emailed support? Unless you give them your birthday, SSN, first born child's blood, etc they won't help you.
Why on earth would I email support if I didn't own a valid product?
Regular Member
glendale2x
Posts: 319
Registered: ‎06-18-2010
Posts: 319
Kudos: 41
Registered: 06-18-2010

Re: Tolly report

Things like community forums and mailing lists act as a public face of the company, and for better or for worse, that's the perception people will have even if it's not the "official" support channel.

Cisco's official support channel is their TAC with pricey support contracts attached to hardware, but Cisco employees regularly participate on the cisco-nsp mailing list. It's a great community resource which is one of the reasons I keep using Cisco.
Regular Member
glendale2x
Posts: 319
Registered: ‎06-18-2010
Posts: 319
Kudos: 41
Registered: 06-18-2010

Re: Tolly report

Back to Mikrotik, the attitude on the forums makes me wounder how much worse official support in private could be, and thus not want to bother. I use RB750's for customer CPEs in a static routing role, sometimes NAT, and pretty much just basic stuff that can't possibly get messed up. I certainly wouldn't feel comfortable relying on it in a situation where I'd ever need official support because it doesn't look like a safe bet.
SuperUser
mhoppes
Posts: 14,014
Registered: ‎06-23-2010
Posts: 14014
Kudos: 4189
Solutions: 59
Registered: 06-23-2010

Re: Tolly report

Back to Mikrotik, the attitude on the forums makes me wounder how much worse official support in private could be, and thus not want to bother. I use RB750's for customer CPEs in a static routing role, sometimes NAT, and pretty much just basic stuff that can't possibly get messed up. I certainly wouldn't feel comfortable relying on it in a situation where I'd ever need official support because it doesn't look like a safe bet.

The attitude in email is the same if not worse. If you can't prove you own the equipment or you fail to provide a tiny little piece of data they deem necessary. For get it.
SuperUser
Josh_SPITwSPOTS
Posts: 18,668
Registered: ‎11-20-2011
Posts: 18668
Kudos: 5903
Solutions: 148
Registered: 11-20-2011

Re: Tolly report

It's a $99 router.
A $99 router and v1.0 OS.
I have the distinct feeling that somebody is being a little too critical for a product they haven't even used yet.
josh reynolds :: www.spitwspots.com
ubiquiti certified trainer :: ubwa | uewa
SuperUser
mrjester
Posts: 1,115
Registered: ‎06-14-2012
Posts: 1115
Kudos: 829
Solutions: 79
Contributions: 9
Registered: 06-14-2012

Re: Tolly report

Here is a question I haven't seen answered by this report...

How does the IPv6 performance compare? Is v6 even hardware accelerated? If so, is the performance equal or does it suffer due to the larger headers?

I would really like to see some numbers around this.
Contact me @UBNT-Bane
SuperUser
Josh_SPITwSPOTS
Posts: 18,668
Registered: ‎11-20-2011
Posts: 18668
Kudos: 5903
Solutions: 148
Registered: 11-20-2011

Re: Tolly report

Here is a question I haven't seen answered by this report...
How does the IPv6 performance compare? Is v6 even hardware accelerated? If so, is the performance equal or does it suffer due to the larger headers?
I would really like to see some numbers around this.

Good question. Will test if I can. Pretty sure it does.
josh reynolds :: www.spitwspots.com
ubiquiti certified trainer :: ubwa | uewa
Ubiquiti Employee
UBNT-stig
Posts: 5,335
Registered: ‎06-09-2011
Posts: 5335
Kudos: 1566
Solutions: 479
Contributions: 15
Registered: 06-09-2011

Re: Tolly report

Here is a question I haven't seen answered by this report...

How does the IPv6 performance compare? Is v6 even hardware accelerated? If so, is the performance equal or does it suffer due to the larger headers?

I would really like to see some numbers around this.
IPv6 currently is not hardware accelerated.
EdgeMAX Router Software Development
SuperUser
Josh_SPITwSPOTS
Posts: 18,668
Registered: ‎11-20-2011
Posts: 18668
Kudos: 5903
Solutions: 148
Registered: 11-20-2011

Re: Tolly report

IPv6 currently is not hardware accelerated.

do you know if that is possible with a firmware upgrade down the road?
-sent via tapatalk
josh reynolds :: www.spitwspots.com
ubiquiti certified trainer :: ubwa | uewa
Ubiquiti Employee
UBNT-ancheng
Posts: 6,884
Registered: ‎06-10-2011
Posts: 6884
Kudos: 2159
Solutions: 829
Contributions: 2
Registered: 06-10-2011

Re: Tolly report

It is certainly a possibility, depending on development resources of course.
Member
Network_Pro
Posts: 131
Registered: ‎07-11-2009
Posts: 131
Kudos: 8
Registered: 07-11-2009

Re: Tolly report

Hello Ubiquiti.
Please provide CERTAINTY for your followrs by stating things like
"IPv6 accelerated update is scheduled for Q1 2013"
Currently it looks like you are on vacation and hanging out with us in the forum.
Ubiquiti Employee
UBNT-ancheng
Posts: 6,884
Registered: ‎06-10-2011
Posts: 6884
Kudos: 2159
Solutions: 829
Contributions: 2
Registered: 06-10-2011

Re: Tolly report

Captain's log, stardate 201210.4

This is day 491 of our peaceful, relaxing vacation. HW team finally fixed the PLL multiplier issues in the new prototypes, and JTAG is finally working correctly! Writing to NOR flash actually works without issues! After tweaking the UART initialization values, serial console is working as well! Exciting! (Note to self: Before getting too excited, don't forget to hang out with the wonderful forum members for a bit. Remember, we're on vacation after all.)

Created the initial files for new board support in u-boot based on eval board. Reduce image size by removing certain components not needed right now. Booting the u-boot image works until the point when it tries to relocate to DRAM, where it simply hangs.

DDR initialization appears to be working but the symptoms look like memory issue. Hmm, SPD data seems to be read correctly from the DIMM. Added simple memory test before u-boot relocating to DRAM. Memory test reports many errors, which seem to indicate incorrect timing parameters or something. Doesn't make sense... The DDR initialization should be using the SPD data automatically, why would the parameters be incorrect? Doesn't make sense at all! Wait, what is that?! Oh NO! What the...
SuperUser
mrjester
Posts: 1,115
Registered: ‎06-14-2012
Posts: 1115
Kudos: 829
Solutions: 79
Contributions: 9
Registered: 06-14-2012

Re: Tolly report

IPv6 currently is not hardware accelerated.


This is most definitely not what I wanted to hear. What is the performance difference?
Contact me @UBNT-Bane
Senior Member
twinkletoes
Posts: 2,857
Registered: ‎08-26-2009
Posts: 2857
Kudos: 715
Solutions: 3
Registered: 08-26-2009

Re: Tolly report

Only 999,999 pps instead of 1,000,000. Luckily, the FPGA wasn't really that important!
SuperUser
mrjester
Posts: 1,115
Registered: ‎06-14-2012
Posts: 1115
Kudos: 829
Solutions: 79
Contributions: 9
Registered: 06-14-2012

Re: Tolly report

Only 999,999 pps instead of 1,000,000. Luckily, the FPGA wasn't really that important!


Actually, in my testing, it had a pretty big impact on performance. Probably won't hurt in a typical home use setting, but if you need bi-directional performance, you will probably want to wait till the v6 is accelerated.
Contact me @UBNT-Bane
Established Member
rebelwireless
Posts: 1,783
Registered: ‎02-09-2010
Posts: 1783
Kudos: 424
Solutions: 10
Registered: 02-09-2010

Re: Tolly report

I would assume a huge performance hit. The MIPS cpu in these things (if it is what is assumed) is not very powerful on it's own, it needs that extra hardware to push these kinds of pps.

I suspect that a similarly clocked mikrotik board will be a perfect comparison. Mikrotik is all software based on the standard linux kernel routing code.
SuperUser
mrjester
Posts: 1,115
Registered: ‎06-14-2012
Posts: 1115
Kudos: 829
Solutions: 79
Contributions: 9
Registered: 06-14-2012

Re: Tolly report

These were the numbers I got from iperf.


  • Two laptops connected directly to the ERL running Ubuntu 12.04 x86 live CD and configured with static IPs.
  • I used the default flags in iperf and only specified the transport and if single or bi directional.
  • Offload Disabled means the ASIC offload for IPv4 was disabled.
  • The ERL was only configured with the IP addresses. No other features were setup.


v4 single direction - v4 1.09Gbytes @ 936Mb/s
v4 bi-directional - 1.09GB @ 932Mb/s, 1.0GB @ 862Mb/s

v6 single direction - 994Mbytes @ 834Mb/s
v6 bi-directional - 878MB @ 736Mb/s, 527MB @ 443Mb/s

v4 Offload Disabled, single direction - 1.09GB @ 936Mb/s
v4 Offload Disabled, bi-directional - 825MB @ 692Mb/s, 837MB @ 701Mb/s


* And before you ask, no I don't have any more details and no I am not really interested in setting up the test again. It is a pita as it was all run from the LiveCD and iperf isn't installed by default.
Contact me @UBNT-Bane
Established Member
Idea
Posts: 1,322
Registered: ‎10-28-2010
Posts: 1322
Kudos: 555
Solutions: 15
Registered: 10-28-2010

Re: Tolly report

Captain's log, stardate 201210.4

This is day 491 of our peaceful, relaxing vacation. HW team finally fixed the PLL multiplier issues in the new prototypes, and JTAG is finally working correctly! Writing to NOR fl..


Yes! More of that will keep the unwashed masses happy.
Member
gsloop
Posts: 164
Registered: ‎03-10-2011
Posts: 164
Kudos: 34
Registered: 03-10-2011

Re: Tolly report

@mrjester
I know, I know, you said no more details...BUT :manhappy:
Could you spare a measly second of your attention to tell me if you know what the packet size was? :manhappy:

One packet to rule them all, One ping to find them,
One broadcast to bring them all and in the darkness route them

-Greg