re: Cisco to Pay $89M for Procket AssetsI agree. The report should publish the filter configs. I am really surprised that 10 filters could screw up the T640 so bad. The packet sizes really do not matter. No matter what the packet size, if these results are true, they are pretty awful.
re: Cisco to Pay $89M for Procket Assetsindianajones wrote: "The packet sizes really do not matter. No matter what the packet size, if these results are true, they are pretty awful."
Wrong, packet sizes matter VERY much. Most equipment today uses some form of cellification to ship packets across the switch fabric. The size of these cells determines what the fabric speedup needs to be for the worst case. For example, if you used a 70-byte cell size and you were sent 71-byte packets, you would need 2 cells for the packet, but the second cell would be mostly padding. This would require a 2x fabric speedup to ensure that line rate could be achieved.
It is well known that Juniper has a problem with 71-byte packets because of this issue. It does not suprise me the packet size used is not published; Cisco is using the competition's worse case with their best case to skew the results in their favor. The fact that they only used 10 filters proves this - they found the worst possible case for Juniper to filter on. I've personally seen 100k line filters on Juniper run with production traffic with no performance hit. I've also seen a 10 line ACL bring a GSR to its knees; it's just a matter of marketing the other guy's weaknesses against your strengths and no one does it better than Cisco.
re: Cisco to Pay $89M for Procket AssetsThank's Tony for the common sense response to the earlier post re: waiting for the fire sale.
Only idiotic companies (and VC speculators) try to buy assets (or just IP) without people. If a company is really interested in the assets, and they know what they are doing, they always make a deal for at least a few core people. If they are interested in small parts of the product line, they make an offer to the sales guy who was selling it...and pick up the customer relationship in so doing. They talk to the customer(s) and see if their acquisition of the assets will change buying patterns.
Etc.
Etc.
The ill will Cisco would have picked up by waiting for the fire sale or tightening the screw down harder would have come back to haunt them in the form of a blown acquisition. They would have a few patents and some chip designs, and no practical way to implement them.
Stated another way, it would have cost them more than $89M and a couple of years to have re-built a crew in-house. And by the time they finished, it might be obsolete.
I did not mean that "packet sizes do not matter". Sorry. What I wanted to say was that irrespective of what packet sizes were used, the T640 seems to perform extremely poorly for certain packet sizes which would be unacceptable for carriers, methinks.
It would be nice if Mier reports on the packet sizes, filter configs used etc. I agree with you that I have seen T640 configured with 100's of filters and still perform well. I would be extremely interested in getting to know what ACLs caused T640 to croak so badly. Just curious :-) Mier guys, want to oblige?
re: Cisco to Pay $89M for Procket Assets >Only idiotic companies (and VC speculators) try >to buy assets (or just IP) > without people. If a company is really > interested in the assets, and they know > what they are doing, they always make a deal > for at least a few core people.
Juniper bought Nexsi chips long after it closed the door, it's been used in their product now.
re: Cisco to Pay $89M for Procket AssetsIf you want a target, pick on Randall. It's unfathomable he has escaped the criticism that he is due. He spent money like he was a prince and created the "fear to speak out" atmosphere. So posters of lightreading, enough of the Procket management bashing. They worked hard and made mistakes, but my guess is that none of you big talkers out there would have done it any differently given the board and first CEO.
Are you saying your first CEO was better than Randall? If so, what happened?
re: Cisco to Pay $89M for Procket AssetsAgree with Why: without the team, there is no value with a product this complex. However, this is not the case always. There are a lot of restarts happening right now and in many cases, the fewer the people, the better, if the product is relatively simple and relatively complete (caveat being that you need a well documented development history in any case).
The key here is that a lot of the time, you have a hangover effect from people who tried really hard to get something built, and turning their attention into either a new product direction or new market is often very difficult and cumbersome. In those situations, the fewer the people, the better. In Procket's case, Cisco is making a smart move hanging on to this team because say what you want, but this was the best engineering team in the communications business, hands down. That is a fact.
Elsewhere...I hate to say I told you so on my post for the 60 reasons to buy Procket. But I did. Cisco was really the only choice that made sense for both sides. If I were Juniper, I would be concerned. They can no longer take their technology advantage for granted, especially if Cisco picked up Dino in this move (does anyone know?). Dino's better than anyone Juniper has, who's still working. So now Juniper's 60 customers may be re-thinking whether to ditch Cisco, especially as Juniper still doesn't have any switching or transport gear and still can't sell a true portfolio.
re: Cisco to Pay $89M for Procket Assets>In Procket's case, Cisco is making a smart >move hanging on to this team because say what >you want, but this was the best engineering >team in the communications business, hands >down. That is a fact.
Best at what?
Best at making noise? Best at public outbursts? Best at in-fighting? Best at self-promotion? Best at leaking info? Best at burning cash? Best at burning bridges?
If that is what you are refereeing to, then I agree 100%!
Given all that, and given the product focus changes, market focus changes, and management changes I am actually shocked that Procket was actually able to get a box out the door at all. The box is not revolutionary enough for carriers to risk buying from a startup; however, it does out perform Juniper a little. I would agree that the Procket team is above average, however, my money is elsewhere for best comm team.
Get over yourself!