Some people are concerned that when a host moves to a different subnet, it could still be tracked because the host part of the address remains the came. In IPv4 there is no simple way to track a host across subnets.
This is a big step forward for IPv6 adoption, but I think the next major step will be by the cable companies. They want every set-top-box or cable TV to have two way communication and be fully addressable. Where else would they get the address space needed for that?
10.0.0.0/8? It fits their "walled garden" business model perfectly.
Virtually every application and router must be updated to support IPv6. Addresses are longer and harder to remember. Packet headers are larger, so less data fits in each packet. Multihoming still hasn't been sorted out. Certain default configurations allow anyone to see your MAC address and thus track your computer more easily (but fixes for this are known). Administration of a dual-stack network may cost almost twice as much as administration of a pure IPv4 network.
I ignored the network neutrality stuff because it's all up in the air right now. But it appears to be the same as my other points: right now it looks like most of the broadband industry is opposed to neutrality, so such complaints have nothing to do with FIOS per se.
Most of your complaints are irrelevant, because FIOS is a form of consumer broadband and whether we like it or not, consumer broadband is totally different from business-class dedicated Internet access.
they also have great difficulties with VLANs, and IPV6
DSL and cable ISPs don't support VLANs or IPv6 either.
can they update their hardware to accommodate multiple concurrent IPTV QoS-based streams at HD raster/frame/color levels?
A HD H.264 stream is only 10Mbps, so FIOS can fit roughly 62 streams per fiber, with a 32:1 split ratio that's about 2 streams per customer worst-case. In real-world situations it will be better.
Can you join an MPLS network
I've never heard of any consumer broadband ISPs that support MPLS, so FIOS is hardly special.
Can you get them to do an SLA?
Yet again, no consumer ISPs offer SLAs. If you want an SLA get a fractional T3 or dedicated Ethernet connection and be happy.
FTTH runs at 622-2400 Mbps, while cable modems are 30-60 Mbps IIRC. Thus the fiber ISPs have more room to increase their bandwidth caps without having to install new equipment. I expect telcos and cablecos to keep leapfrogging each other in terms of Mbps/$ over the next few years.
There's a big difference between having IPv6 available in equipment and using it. Virtually all new routers now support IPv6, but the ISPs don't turn it on.
I've heard IPv6 is big in Japan (they get everything first), but it appears to be more for political posturing than practical reasons....there appear to be a bunch of compatibility issues for running a pure IPv6 network...
That's because you're not supposed to go pure IPv6; you should run dual-stack for 10-20 years before dropping IPv4.
Re:Not sure about the book...
on
User Mode Linux
·
· Score: 1
I saw a company in the UK offering Zones hosting, but I lost the URL. Joyent is probably going to be offering SPARC Zones hosting at some point.
I'd have to guess that a distributed streaming system is a long way off in being possible, since timely delivery is crucial, and people further away from downloading from the source would not be getting the data fast enough in real time.
Regular Internet streaming has ~10 seconds of buffering; would it make that much of a difference if P2P streaming had 30-60 seconds of buffering?
There is a lot of academic research about P2P live streaming: Yoid, ESM, TMesh, SplitStream, Bullet, Chainsaw, etc. A company called Blue Falcon (no kidding) built a P2P live streaming system (and then ran out of money). It's only a matter of time before this technology becomes common.
Re:Not sure about the book...
on
User Mode Linux
·
· Score: 2, Insightful
VPSes are cool, but UML is almost the least efficient way to implement them. I anticipate that Linode will be dethroned by Linux-VServer, OpenVZ, and Solaris Zones hosting soon enough.
If you want some sort of wireless P2P phone system, you'd probably be better off starting with a PDA with a high-power 802.11 card in it. SDR sounds like overkill.
Network interface throughput (in bytes/s) divided by floating-point operations per second; this represents the compute/communication balance that tends to be inherent in scientific cluster applications. So if you double the number of processors but keep the same number of network interfaces, your bytes/flop is 2X worse, which may limit application performance.
I thought the current dual-socket motherboards (eg this board) could already accept dual-core Athlon (well, Opteron) chips (eg: the 270 series) to make a quad-core machine ?
They can; 4x4 appears to be a new marketing label for the same thing. (Just as "Athlon" and "Opteron" are the same chip already.)
There's also Guba, a site that downloads videos from Usenet and converts them to Flash.
In retrospect, it makes sense. Flash is small and easy to install, so it has huge market share. And it loads fast, has a modern codec, and allows the player to be customized, so Web developers like it.
Slashdot Mirror
Some people are concerned that when a host moves to a different subnet, it could still be tracked because the host part of the address remains the came. In IPv4 there is no simple way to track a host across subnets.
If a host uses stateless autoconfiguration, then by default the host part of its global IPv6 address will contain its MAC address.
This is a big step forward for IPv6 adoption, but I think the next major step will be by the cable companies. They want every set-top-box or cable TV to have two way communication and be fully addressable. Where else would they get the address space needed for that?
10.0.0.0/8? It fits their "walled garden" business model perfectly.
Virtually every application and router must be updated to support IPv6.
Addresses are longer and harder to remember.
Packet headers are larger, so less data fits in each packet.
Multihoming still hasn't been sorted out.
Certain default configurations allow anyone to see your MAC address and thus track your computer more easily (but fixes for this are known).
Administration of a dual-stack network may cost almost twice as much as administration of a pure IPv4 network.
Yes, several analyses of IPv4 address usage over time have been made, although they don't agree with each other:
Geoff Huston (2003)
Tony Hain (2005)
are the local telcos (ie, the ones actually providing access to users) paying for transit or do they have peering arangements?
It depends on the size. The largest last-mile ISPs (e.g. AT&T) can probably negotiate settlement-free peering. Small ISPs buy transit.
It would seem to me that they would make out in the transit arrangement already because they are receiving more data than they are sending.
In transit, doesn't the smaller ISP always pay?
Of course, SSL does not prevent and was never intended to prevent spamming and phishing; it foils sniffing (which generally doesn't happen anyway).
I ignored the network neutrality stuff because it's all up in the air right now. But it appears to be the same as my other points: right now it looks like most of the broadband industry is opposed to neutrality, so such complaints have nothing to do with FIOS per se.
Most of your complaints are irrelevant, because FIOS is a form of consumer broadband and whether we like it or not, consumer broadband is totally different from business-class dedicated Internet access.
they also have great difficulties with VLANs, and IPV6
DSL and cable ISPs don't support VLANs or IPv6 either.
can they update their hardware to accommodate multiple concurrent IPTV QoS-based streams at HD raster/frame/color levels?
A HD H.264 stream is only 10Mbps, so FIOS can fit roughly 62 streams per fiber, with a 32:1 split ratio that's about 2 streams per customer worst-case. In real-world situations it will be better.
Can you join an MPLS network
I've never heard of any consumer broadband ISPs that support MPLS, so FIOS is hardly special.
Can you get them to do an SLA?
Yet again, no consumer ISPs offer SLAs. If you want an SLA get a fractional T3 or dedicated Ethernet connection and be happy.
FTTH runs at 622-2400 Mbps, while cable modems are 30-60 Mbps IIRC. Thus the fiber ISPs have more room to increase their bandwidth caps without having to install new equipment. I expect telcos and cablecos to keep leapfrogging each other in terms of Mbps/$ over the next few years.
That's why VMware gave up on the hypervisor wars and is trying to provide better system management than Xen.
That's the IETF plan, but I suspect the telco/cable duopoly plan will be different.
There's a big difference between having IPv6 available in equipment and using it. Virtually all new routers now support IPv6, but the ISPs don't turn it on.
I've heard IPv6 is big in Japan (they get everything first), but it appears to be more for political posturing than practical reasons. ...there appear to be a bunch of compatibility issues for running a pure IPv6 network...
That's because you're not supposed to go pure IPv6; you should run dual-stack for 10-20 years before dropping IPv4.
I saw a company in the UK offering Zones hosting, but I lost the URL. Joyent is probably going to be offering SPARC Zones hosting at some point.
I'd have to guess that a distributed streaming system is a long way off in being possible, since timely delivery is crucial, and people further away from downloading from the source would not be getting the data fast enough in real time.
Regular Internet streaming has ~10 seconds of buffering; would it make that much of a difference if P2P streaming had 30-60 seconds of buffering?
There is a lot of academic research about P2P live streaming: Yoid, ESM, TMesh, SplitStream, Bullet, Chainsaw, etc. A company called Blue Falcon (no kidding) built a P2P live streaming system (and then ran out of money). It's only a matter of time before this technology becomes common.
VPSes are cool, but UML is almost the least efficient way to implement them. I anticipate that Linode will be dethroned by Linux-VServer, OpenVZ, and Solaris Zones hosting soon enough.
If you want some sort of wireless P2P phone system, you'd probably be better off starting with a PDA with a high-power 802.11 card in it. SDR sounds like overkill.
The old version of MovieBeam was using dNTSC, but the new version seems to be using ATSC.
Has AMD started enabling multiple hypertransport links in the Athlon chips?
That's what they just announced. What AMD has crippled, they can uncripple whenever they feel like...
Network interface throughput (in bytes/s) divided by floating-point operations per second; this represents the compute/communication balance that tends to be inherent in scientific cluster applications. So if you double the number of processors but keep the same number of network interfaces, your bytes/flop is 2X worse, which may limit application performance.
I don't see what the fuss is about; 2-socket motherboards have been available for years and most scientific clusters already use them.
This will reduce by a factor of four the number of expensive low-latency network interconnects needed to build a cluster of a given size.
Not if you want to maintain constant bytes/flop.
I thought the current dual-socket motherboards (eg this board) could already accept dual-core Athlon (well, Opteron) chips (eg: the 270 series) to make a quad-core machine ?
They can; 4x4 appears to be a new marketing label for the same thing. (Just as "Athlon" and "Opteron" are the same chip already.)
There's also Guba, a site that downloads videos from Usenet and converts them to Flash.
In retrospect, it makes sense. Flash is small and easy to install, so it has huge market share. And it loads fast, has a modern codec, and allows the player to be customized, so Web developers like it.
Of course, the latest Athlon 64 FX is both faster and cheaper than an Opteron 875. Same for Intel.