re: Charlotte's Networks, Take 2"Any suggestion that CWNT was at a three-month disadvantage is false. There was a five-day gap. Any other time issues are of CWNT's making and no one else's."
Dave, that's a misleading statement at best. The ACTUAL gap was NOT 5 days but was more like 2 1/2 months if I understand you correctly.
Apparently Charlotte's was given the OPPORTUNITY to test with a gap as small as 5 days but for whatever reasons (most likely the revised SW code was not done and or they had logistical conflicts) they choose not to retest with the 5 day gap. That is not the same as saying "There WAS a five-day gap."
Hell, Lightreading published CWNT's plans for a retest way back on March 2nd, the same day they published the test results from the first test.
"CharlotteGÇÖs Networks has indicated that it will commission a private test this spring that reproduces all the events in the Light Reading project"
Seems to me that your 5 day comment would be analogous to saying that both CISCO and Juniper had a ZERO day gap for the second test since they were afforded the same OPPORTUNITY that CWNT had to conduct their own retest and they for whatever reason choose not to participate!
Dave that's a misleading statement at best. The ACTUAL gap was NOT 5 days but was more like 2 1/2 months if I understand you correctly.
Apparently Charlotte's was given the OPPORTUNITY to test with a gap as small as 5 days but for whatever reasons (most likely the revised SW code was not done and or they had logistical conflicts) they choose not to retest with the 5 day gap. That is not the same as saying "There WAS a five-day gap."
Hell, Lightreading published CWNT's plans for a retest way back on March 2nd, the same day they published the test results from the first test.
"CharlotteGÇÖs Networks has indicated that it will commission a private test this spring that reproduces all the events in the Light Reading project"
Seems to me that your 5 day comment would be analogous to saying that both CISCO and Juniper had a ZERO day gap for the second test since they were afforded the same OPPORTUNITY that CWNT had to conduct their own retest and they for whatever reason choose not to participate! Maybe they weren't ready to retest.
re: Charlotte's Networks, Take 2"Seems to me that your 5 day comment would be analogous to saying that both CISCO and Juniper had a ZERO day gap for the second test since they were afforded the same OPPORTUNITY that CWNT had to conduct their own retest and they for whatever reason choose not to participate! Maybe they weren't ready to retest."
We offered all vendors in the LR test the same opportunity to supply devices for testing at the same point, plus or minus 5 days.
The fact that CWNT opted not to do so is CWNT's issue, not ours.
re: Charlotte's Networks, Take 2SPASM wrote: >>I don't understand the focus on the "3 months". >> The bottom line is this - Who has >> the better box now? Charlotte has a better box >> when comparing OC-48....
If you assume that over the last 3++ months, neither Juniper nor Cisco have done anything to improve the performance of their products, then that is a very narrow view of the world where your competitors sit around, twiddle their thumbs and do nothing while they wait for you to catch up.
I am respecting your opinion very much. I am looking on the MPLs test report (page 7 ) on www.cwnt.com. I see the following:
Charlotte 18,375+ LSPs Juniper 10,000 LSPs Cisco 5,000 LSPs
Is this number important at all? Is MPLs important? In the report (page 1) I see some issues of latencies vs. throughput. Can you as an expert comment on the above issues, please.
re: Charlotte's Networks, Take 2Have you tested those routers again Mr.Newman? Has anyone? Same conditions,same lab,same people? Yeah,I know,apples and oranges....
The bias here is OVERLY apparent.
And I haven't a clue who Perry Piedmont is....
That's exactly his point. Same conditions, same lab, same people. The exact thing that didn't happen in the retest of cwnt equipment. It's the same as 2 baseball teams playing a game. One loses, then goes back to the stadium later and says, " Look, we hit 6 home runs when you weren't here!" Not the best analogy, I admit, but you see my point. As an engineer (test engineer), I can see dnewman's point, but not yours. I can't see the bias, either. His view was neutral, stating the obvious caveats inherent in the test scenario. As you already stated," Same conditions, same lab, same people."
re: Charlotte's Networks, Take 2If you can't answer THAT question, phrased in THAT way, then at least explain why you won't.
Simple. We can all see what statement you are trying to wheedle out of him. I wouldn't answer a question phrased that way, either. It would ruin his credibility. I wouldn't vouch for any test results I didn't verify, either. Comparison testing is done the same way the world over, and not just in the CompSci field. All the engineering methodology here is based on what has gone before, because it's proven to be accurate and reproducible.
re: Charlotte's Networks, Take 2This wasn't directed toward me, but I am going to answer the question anyway:
Charlotte 18,375+ LSPs Juniper 10,000 LSPs Cisco 5,000 LSPs
Is this number important at all? ---------------------------------------
This number alone is meaningless. All three numbers are way too small and there is no architectural limit in any of the three systems that would prevent them from quickly delivering a higher number.
And I will tell you that this is a number where Charlotte's having the test results of Juniper and cisco likely allowed them to tune their system to a winning number. And that other vendors could probably match this number with a new software load.
This number says nothing about MPLS features, the quality of the implementation or (most importantly) the ability of the MPLS software of each vendor to work well with that of other vendors. If I was going to evaluate a router, those things would be important. ========
Is MPLs important? ------------------- It depends on who the router is being sold to. To the companies that are only deploying MPLS as a toy in their labs, it doesn't matter at all.
To many other people, it depends on which MPLS features are actually being delivered with the product. It depends on how the customer intends to use MPLS.
re: Charlotte's Networks, Take 2David Newman you dont understand the difference between olympic games and a world record. Four competitors show up to the olympic games to run the 100meter. One competitor takes third place in the olympics. Now three months later that same competitor shows up to a track meet in the same stadium with the same referee(Spirent). That competitor runs the fastest time in the world with the same referee present(Spirent). Thus you have one competitor with the label gold metal winner and one with the label world record holder. Is it fair to compare a gold metal winner with a world record holder???
Flawed analogy. Four runners run at the Olympics, one sets the Olympic record, one comes in third. 3 months later, the 3rd place finisher sets the World record. What you have is 1 Olympic record holder, and 1 World record holder. If you watch the Olympics, you'll see that's how it is represented. Here we have a 4th place finisher in a race of four, and a first place finisher in a race of one( already been stated as such). Is that such a hard concept for everyone to wrap their intellects around? It doesn't mean that the cwnt router sucks, it means that they didn't do very well in the heads up comparison. Plain & simple, can't reduce it much more than that.
Please answer this. Who was responsible for the integrity of the tests and test scripts that were used in the LR tests?
I assume it is you.