Mikrotik CCR2004-1G-12S+2XS (12x sfp+, 2x sfp28) router msrp 595$

[RTM]

So I figure some of you might find it interesting, that Mikrotik has released details of their new CCR2004 router.
Link to more information on their website

It has 12x 10g SFP+ ports, 2x 25g SFP28 ports and a 1G port (management usually), all priced very nicely at a MSRP of $595.
As far as I can tell it is the same SoC(or same series) (according to Mikrotik it is the AL-32400) as what Ubiquiti put in the Dream machine pro (AL-324). I've found prices listed here in the EU for around 450 EUR not including VAT.

I find it noteworthy to mention that according to the benchmark, it is unlikely to be able to route the full bandwidth of the ports, according to their benchmark the best case without rules is around 34Gbps (which is still pretty damn awesome IMHO).

I assume they will be able to offload some of the traffic (say if you assign some ports to act like a small switch) to the builtin Marvell chip (which according to the block diagram is a Passive Intelligent Port Extender, whatever that is), but it would be interesting to know a little more of how it is supposed to work. @Patrick consider this a request to test the device (preferably in comparison with the dream machine pro)

 

[RTM]

I forgot to bring up, that in the marketing materials, they are listing that the device is unable to use all the memory with RouterOS 6 (slightly less than 2G) and 4G with RouterOS 7. Other than the obvious issue with a possible inability to use what you bought fully, it brings up the potential question, if it is going to be a gimped device without RouterOS 7, which in turn may be too unstable to actually use in production(it is in beta right now).

Another thing worth mentioning, is that more details were leaked about the CCR2004 series on the Mikrotik forum, among that it looks like there may be a 16 core version of this.

ccr2004-1g-12s+2xs
ccr2016-1g-12xs-2xq
ccr2004-2g-2xc-2s+
ccr2004-2g-2xc-2s+/r2
ccr2004-2g-2xc-2s+/r2-60

All referenced in a kernel object file in v7.0b5...

 

[PigLover]

The CCR2004 is also detailed in their May newsletter: https://mikrotikdownload.s3.eu-west-1.amazonaws.com/news/news_95.pdf

I liked the reference and links to Patrick's review of their PoE+ product on page 6 of the newsletter.

 

[Patrick]

@PigLover I saw that this morning. STH got a whole page.

We are trying to do a bit of a router/ firewall test revamp. Maybe we test when that is ready.

 

[Eong]

I find it noteworthy to mention that according to the benchmark, it is unlikely to be able to route the full bandwidth of the ports, according to their benchmark the best case without rules is around 34Gbps (which is still pretty damn awesome IMHO).

Does it mean the two SFP28 ports will eat all the 34Gbps bandwidth when they work at full speed?
Connect the SFP28 to my QSFP switch is another issue.

 

[RTM]

Does it mean the two SFP28 ports will eat all the 34Gbps bandwidth when they work at full speed?
Connect the SFP28 to my QSFP switch is another issue.

If you look at the block diagram (I posted a link earlier), you can see that the PIPE chip has an uplink of 2x25 to the SoC, so it seems likely that if all processing needs to happen on the SoC the SFP28's will indeed eat up all the bandwidth.

At the end of the day, I think it all comes down to what that PIPE chip can do, it seems a little anemic to have 34G (in absolute best case scenario) routing performance with that amount of ports and max. bandwidth if all traffic has to pass the SoC. Then again perhaps it is just bias on my end, wanting to use it as a core router in a network like a L3 switch, perhaps there is a different use case that I am not seeing.

 

[RTM]

So the final verdict is still out (at least there is nothing official), but it would seem that the consensus on the Mikrotik forum is that the PIPE chip is fairly dumb and unlikely to be useful as anything resembling a switch.

Just wanted to put this here, so people don't start spending hard earned money on what may be a pipe dream (no pun intended), it is definitely best to wait for an official response.

 

[Deslok]

I'm fairly excited to see this if for no other reason than it means we're going to see a revamp of the CCR 1xxx lineup soon(well, over the next year or two) I just setup a CCR1009 for a small office and it was fantastic. I'm a little suprised to see SFP28 ports on this when they've been so gung-ho about 40Gb ports on their larger switches(although if I were to guess the ccr2016-1g-12xs-2xq is going to be 12 25gb ports and 2 40gb ports which would be impressive)

looking at their test results it's performing better than the old CCR1016 and CCR1009 while being priced between them with a massive increase in potential port configurations(with 64byte being an exception for some reason...).

 

[newabc]

@PigLover I saw that this morning. STH got a whole page.

We are trying to do a bit of a router/ firewall test revamp. Maybe we test when that is ready.

I am excited to see it.

I think the test result of CCR2004-1G-12S+2XS in Mikrotik's page is quite acceptable than the specs of UDM pro.

It is doubtful about the IPS/IDS throughput of UDM pro with a quad-core 1.7Ghz ARM and 4G RAM in the real world, maybe its specs is the result of using a very very low numbers of signatures/rules.