OpenFlow 1.1 in hardware: I was wrong (again)

Earlier this month I wrotewe’ll probably have to wait at least a few years before we’ll see a full-blown hardware product implementing OpenFlow 1.1.” (and probably repeated something along the same lines in during the OpenFlow Packet Pushers podcast). I was wrong (and I won’t split hairs and claim that an academic proof-of-concept doesn’t count). Here it is: @nbk1 pointed me to a 100 Gbps switch implementing the latest-and-greatest OpenFlow 1.1.

The trick lies in the NP-4 network processors from EZchip. These amazing beasts are powerful enough to handle the linked tables required by OpenFlow 1.1; the researchers “just” had to implement the OpenFlow API and compile OpenFlow TCAM structures into NP-4 microcode.

I have to admit I’m impressed (and as some people know, that’s not an easy task). It doesn’t matter whether the solution can handle full 100 Gbps or what the pps figures are; they got very far very soon using off-the-shelf hardware, so it shouldn’t be impossibly hard to repeat the performance and launch a commercial product. The only question is the price of the NP-4 chipset (including associated TCAM they were using) – can someone build a reasonably-priced switch out of that hardware?

Along the same lines: would anyone know which vendors/products are using the NP-4 chips? That information might give us some insight into how pricey an NP-4-based switch could be.

7 comments:

  1. I found this .pdf of an EZChip presentation from February 2011. Page 11 does not list specific hardware but does list current vendors who are customers of EZChip. It may help in narrowing down a search for specific products.

    www.ezchip.com/Images/pdf/EZCH-investorsQ410-110210.pdf
  2. So far all I see is the NP-4 mentioned with Cisco's ASR900 and associated line cards.

    some good info here

    "....NP-4 will most likely go into production in 4Q11 and we now do not expect the NP-4c based ASR9000 line-card to go into production before 1Q12. In our view, longer term prospects remain quite positive as our checks suggest that service provider secular trends are healthy and also as multiple tier1 customers including CSCO have already signed on to use the next generation NP-5 platfor..."

    http://www.streetinsider.com/Special+Reports/Auriga+Maintains+a+Buy+on+EZchip+Semiconductor+%28EZCH%29%3B+JNPR+MX+Sales+Positive,+but+NP-4+Ramp+Likely+Delayed/6441161.html
  3. I looked into EZchip several years ago and really liked their capabilities. Good to see now that their chips are powerful enough for those speeds and flexible enough to run such generic software.
  4. Thanks for the link. Looks like Juniper is a major customer (and we know their gear is reasonably-priced).

    Also, based on the rough calculations they make in the PDF, the NP-x chipsets are not a major cost factor.
  5. http://groups.geni.net/geni/attachment/wiki/GEC10DemoSummary/UQAM-Openflow-Genidemo-Poster_V_1.0.pdf?format=raw
  6. The Juniper SRX, MX, and M-series employ the EZCH NP-2, but no subsequent EZCH generations. The largest NP-3 deployment is in the Cisco ASR 9000 and also in the new line cards on their CAT 65xx and 76xx switches. The next largest NP-3 deployment is in ZTE devices. The next generation NP-4 will be in the same Cisco boxes and more around 1Q2012. The NP-4 will also be in boxes from ZTE, Huawei, and Ericsson around 4Q2011. Some people are predicting EZCH is on the path to becoming the Intel of the network.
  7. @Seth: Thank you! Let's hope EZCH gets there.
Add comment
Sidebar