I think something needs to be explained here. Apparently from the threads I've been reading in the article most people do not realize that VoIP is a high bandwidth application. It's true. Consider this, we allot 8kbps per user for our b-band offering which is 125:1 if you're a network person or 128:1 if you're a retarded systems administrator. This typically leaves us with a surplus of bandwidth. For businesses it all depends on the SLA. If they want 1:1 then they'll be paying about $100/meg. Our cheapest upstream is $75/meg plus our own network costs (we just sunk $750k in a new core in one POP and we only replaced 2 devices). How do I know this? I run an ISP.
Going back to the original topic. Skype, Vonage and VoIP offerings built into IM clients, FPS and role-playing games (or the addons) consume between 32 and 64kps, depending on the codec and utilization of the voice frequencies (ie, my phone calls consume around 32kbps but a call between my aunt and mother run much closer to 64kbps). Contrary to popular misbelief just because an audio codec like G.711 claims to only use up to 64kbps does not mean it won't consume more bandwidth with more voice traffic, ie both people talking simultaneously. The voice traffic is many times the average transfer rate of most consumers. While surfing the web and checking email most users will barely make a blip on a I/O graph of their CM or their DSL modem. Most of the VoIP apps I've worked with use G.711 by default instead of G.729 or some other less demanding codec. I haven't even touched on IP/UDP overhead for VoIP traffic. A G.711 64kbps stream is around 84kbps with IP/UDP overhead. This overhead is even greater if you're putting the traffic onto a VPN tunnel of some sort. GRE adds 24; IPSec adds 40 IIRC. Depending on your method VPN implementation you could even be pushing IPSec over TCP adds another 20+, depending on header options. Your VoIP call could be close to the upstream limits of your b-band connection and you don't even realize it, depending on your setup of course.
So in short, yes, VoIP is considered a high bandwidth application when compared to the atypical "95%" user. These are the users that we base on bandwidth allotments on. P2P, NNTP, and porn downloaders fall into the "5%" category. The unused excess from the "95%" users generally takes care of these users. We also run with a fairly substantial buffer, just in case. We have now decided to push for up to 100Mbps to the doorstep over the course of the next 3-5 years. We're rolling out ADSL2+ in some areas as a stop-gap measure and have started on a FTTH project for the remaining areas. We anticipate that more of the "95%" users will be become bandwidth consumers as IPTV, video-on-demand and online movie rental products become more prevalent. The trick is to not overbuild the network before users are ready to use it. We can't pass along the increased costs until they're ready for improved service. Raising cable bills by $5/month will piss alot of people off, even when we've deployed $50mil of plant and network upgrades.
That's a good point. 3Ware's management app takes acre of this for you. It even warns you on CRC errors. It's pretty slick. I'm sold on 3Ware cards. I have 7 or 8 in production in my own personal equipment.
As for hardware RAID, I would not necessarily recommend that either, as it moves the single point of failure without resolving the problem. Replacing a broken controller with something compatible some years down the road can prove impossible, especially with onboard controllers. There's also the fact that a number of RAID controller cards are buggy and others do most os the work in software drivers anyway! Performance is also no longer a reason to use a pure hardware RAID solution, especially now that multi-core machines are available cheaply.
What the fuck are you smoking? Software RAID vs hardware RAID isn't about CPU performance. It's about I/O. It's about reliability. You will never reach the same levels of I/O using onboard or expansion controllers as you will using hardware controller. EVER. Software RAID will never be able to touch the performance of ASICs for calculating parity. Software RAID can't utilize a BBU to protect the I/O stream to save it to disk upon power restoration. Software RAID can't benefit from the multi-channel read performance available with RAID 1, 0+1, 1+0, 5, 10, 30 or 50.
Your comment about buggy RAID controllers is pure bullshit. Name one buggy hardware RAID controller. I've used hardware RAID controllers from Adaptec, AMI, LSI, and 3Ware on 4 different platforms. I've even used fake hardware RAID controllers from Promise and Highpoint (I've got a Highpoint 404 going real cheap). The real hardware RAID controllers have always worked flawlessly. I have yet to encounter a buggy hardware RAID controller, even back when 3Ware support wasn't yet in the kernel and the driver was a canned offering from 3Ware. The fake hardware RAID controllers from Promise and Highpoint suck. Like you said, the driver does all the work on the non-hardware RAID controllers. The driver does nothing on the real hardware RAID controllers.
Bush and friends started a war in the middle of a bunch of hothead countries that simply live for a reason to fight. Bush and friends started a war in a country that hates us for not finishing the job the last time we were there. Bush and friends started a war in those conditions over oil. What makes you think that they wouldn't do the same thing to Russia? You think they won't attack Russia because they have nukes? That's a pretty naive assessment of our current administration's willingness to piss off other countries.
Lets look at this from yet another angle. There's a lot of oil in Siberia, isn't there? And Siberia is in Russia, isn't it? And Bush and friends love their oil, don't they? Now, is it much of a stretch to think that March 20, 2003">war to acquire oil? Na, it could never happen, again.
Of course they are pitching a good or service. They are pitching their services to whomever answers the phone. They wouldn't be doing this if there wasn't money in it for them. Even if by some stretch of the law they aren't considered telemarketers they are still obligated, as we all are, to stop calling when asked lest they be accused of harassment which is exactly what they're doing.
You mean like sending pictures of "Tub Girl" or the "Goat Sex" guy in an infinite loop? I would think that would get their attention. It's probably digital anyhow.
That's pretty slick. Could you point me to any docs or howtos on that? I'd love to try it. That would be enough reason for me to build my first Asterisk box. Thanks
If you don't mind the added expense you could re-route the calls back to the company that's calling you. All you have to do is spend enough time on the phone with a recruiter or two to either get the name of their company or their number so you can call them back as soon as you get some with something really important (like setting up the call re-routing). Send their trash calls right back to them, preferrably on a toll-free number too.
He was a programmer/DBA/general IT guy for a Dot-Bomb startup. He built the mission-critical website and DB backend for the company. My sister and he went on vacation for a week or so months after they were in production. He took a company cell, a company pager, and his own cell with him so they could contact him if something came up. While he was gone something went wrong. Rather than call him so he could work on the problem the owner's son decided to muck around in the DB to see if he could figure it out. He seriously fucked it up then. When my brother-in-law came back to work (I don't think they told him about the problem until he came back) they fired him. The blame was laid on his shoulders rather than the company or the idiot that management let work on the broken DB. Brilliance in action. They've since gone under. What a shame.
Germany? Can you elaborate on this? I've read a little bit about CoS but I wasn't aware that Germany was doing anything in particular. I'm curious now.
Could you describe your software solution in more detail? What package are you using. Does it run on Linux or is it a commercial Windows app? I'd be interested in setting up something similar. Thanks
My consulting company recently added security as a service we offer. That made it a separate group within our company. Since that time all manner of security-related things have been pushed off onto this group. I see this as a very bad way of practicing security. We don't need a special group of people to come in and harden or servers & network after we deployed it. We need to utilize good security practices as we're building the infrastructure. A secure architecture isn't something you can tack on after the project is 99% complete. It's something that has to be designed into the project from day one. "How do we achieve our goal in a secure fashion?" or at the very least, "How do we achieve our goal with the minimal acceptable risk to our security" because everything has an implied security risk. I believe this new direction for us will ultimately lead to replication of work as the security people try to rebuild our product after the fact. This obviously increases the number of billable hours on a project. Inevitably since they aren't network engineers or systems engineers they will over-secure something to the point of not working. This in turn creates even more billable hours to the customer. Perhaps that's what my company is really after....
It's naive to think that the RF bleed would only interfere with actual electrical devices. You forget that there are 135 miles of wire in a typical 747, 238 in Air Force One, and over 300 miles of wire in an A380. The wires are everywhere. You're no more than half the height of the cabin from wires (since wires run over the cabin and under the cabin, hell all around the cabin). All modern air planes are "fly by wire" planes. Man over hydraulic controls are only seen in a few minor or secondary systems. Everything else is digital. It is extremely naive to think that interference on the wires themselves can't cause problems.
Apparently you've never been to a gym. I've been to 2 over these past few months. One was an exclusive gym and the other was one of the area YMCAs. While there were a few ugly ducklings at each place both sites by and large were easily 90% filled with very attractive people.
That's interesting. One of our customers, for which we are the outsourced IT department, was bought by Accenture last year. This company actually had its founding roots in McDonalds and Pizza Hut. Interesting stuff...
I've actually never pronounced SQL as "sequel". I've always pronounced it as "squall" as in "my squall". I don't generally say MS Squall though. I usually just spell it out. That's just me of course but that's always the most common way I've heard it said around these parts. People say "sequel" if they mean to say SQL by itself and they say "squall" when part of MySQL.
Slashdot Mirror
You must be new here.
Going back to the original topic. Skype, Vonage and VoIP offerings built into IM clients, FPS and role-playing games (or the addons) consume between 32 and 64kps, depending on the codec and utilization of the voice frequencies (ie, my phone calls consume around 32kbps but a call between my aunt and mother run much closer to 64kbps). Contrary to popular misbelief just because an audio codec like G.711 claims to only use up to 64kbps does not mean it won't consume more bandwidth with more voice traffic, ie both people talking simultaneously. The voice traffic is many times the average transfer rate of most consumers. While surfing the web and checking email most users will barely make a blip on a I/O graph of their CM or their DSL modem. Most of the VoIP apps I've worked with use G.711 by default instead of G.729 or some other less demanding codec. I haven't even touched on IP/UDP overhead for VoIP traffic. A G.711 64kbps stream is around 84kbps with IP/UDP overhead. This overhead is even greater if you're putting the traffic onto a VPN tunnel of some sort. GRE adds 24; IPSec adds 40 IIRC. Depending on your method VPN implementation you could even be pushing IPSec over TCP adds another 20+, depending on header options. Your VoIP call could be close to the upstream limits of your b-band connection and you don't even realize it, depending on your setup of course.
So in short, yes, VoIP is considered a high bandwidth application when compared to the atypical "95%" user. These are the users that we base on bandwidth allotments on. P2P, NNTP, and porn downloaders fall into the "5%" category. The unused excess from the "95%" users generally takes care of these users. We also run with a fairly substantial buffer, just in case. We have now decided to push for up to 100Mbps to the doorstep over the course of the next 3-5 years. We're rolling out ADSL2+ in some areas as a stop-gap measure and have started on a FTTH project for the remaining areas. We anticipate that more of the "95%" users will be become bandwidth consumers as IPTV, video-on-demand and online movie rental products become more prevalent. The trick is to not overbuild the network before users are ready to use it. We can't pass along the increased costs until they're ready for improved service. Raising cable bills by $5/month will piss alot of people off, even when we've deployed $50mil of plant and network upgrades.
That's a good point. 3Ware's management app takes acre of this for you. It even warns you on CRC errors. It's pretty slick. I'm sold on 3Ware cards. I have 7 or 8 in production in my own personal equipment.
What the fuck are you smoking? Software RAID vs hardware RAID isn't about CPU performance. It's about I/O. It's about reliability. You will never reach the same levels of I/O using onboard or expansion controllers as you will using hardware controller. EVER. Software RAID will never be able to touch the performance of ASICs for calculating parity. Software RAID can't utilize a BBU to protect the I/O stream to save it to disk upon power restoration. Software RAID can't benefit from the multi-channel read performance available with RAID 1, 0+1, 1+0, 5, 10, 30 or 50.
Your comment about buggy RAID controllers is pure bullshit. Name one buggy hardware RAID controller. I've used hardware RAID controllers from Adaptec, AMI, LSI, and 3Ware on 4 different platforms. I've even used fake hardware RAID controllers from Promise and Highpoint (I've got a Highpoint 404 going real cheap). The real hardware RAID controllers have always worked flawlessly. I have yet to encounter a buggy hardware RAID controller, even back when 3Ware support wasn't yet in the kernel and the driver was a canned offering from 3Ware. The fake hardware RAID controllers from Promise and Highpoint suck. Like you said, the driver does all the work on the non-hardware RAID controllers. The driver does nothing on the real hardware RAID controllers.
Bush and friends started a war in the middle of a bunch of hothead countries that simply live for a reason to fight. Bush and friends started a war in a country that hates us for not finishing the job the last time we were there. Bush and friends started a war in those conditions over oil. What makes you think that they wouldn't do the same thing to Russia? You think they won't attack Russia because they have nukes? That's a pretty naive assessment of our current administration's willingness to piss off other countries.
Lets look at this from yet another angle. There's a lot of oil in Siberia, isn't there? And Siberia is in Russia, isn't it? And Bush and friends love their oil, don't they? Now, is it much of a stretch to think that March 20, 2003">war to acquire oil? Na, it could never happen, again.
http://www.thebulletin.org/minutes-to-midnight/
How many of you folks actually remember this thing?
Of course they are pitching a good or service. They are pitching their services to whomever answers the phone. They wouldn't be doing this if there wasn't money in it for them. Even if by some stretch of the law they aren't considered telemarketers they are still obligated, as we all are, to stop calling when asked lest they be accused of harassment which is exactly what they're doing.
You mean like sending pictures of "Tub Girl" or the "Goat Sex" guy in an infinite loop? I would think that would get their attention. It's probably digital anyhow.
That's pretty slick. Could you point me to any docs or howtos on that? I'd love to try it. That would be enough reason for me to build my first Asterisk box. Thanks
If you don't mind the added expense you could re-route the calls back to the company that's calling you. All you have to do is spend enough time on the phone with a recruiter or two to either get the name of their company or their number so you can call them back as soon as you get some with something really important (like setting up the call re-routing). Send their trash calls right back to them, preferrably on a toll-free number too.
This is your mother. Stop posting on this silly website and go clean your room.
Hey, that's a really good idea. Maybe I should be French instead. That might help. ;-)
He was a programmer/DBA/general IT guy for a Dot-Bomb startup. He built the mission-critical website and DB backend for the company. My sister and he went on vacation for a week or so months after they were in production. He took a company cell, a company pager, and his own cell with him so they could contact him if something came up. While he was gone something went wrong. Rather than call him so he could work on the problem the owner's son decided to muck around in the DB to see if he could figure it out. He seriously fucked it up then. When my brother-in-law came back to work (I don't think they told him about the problem until he came back) they fired him. The blame was laid on his shoulders rather than the company or the idiot that management let work on the broken DB. Brilliance in action. They've since gone under. What a shame.
Same reason why Orin Hatch has been around for so long. He's got deep pockets thanks to his corporate overlords.
*gasp* Can it be? The chosen has returned!
All hail Plover, idol of Slashdot, leader of the unbetrothed.
But seriously, congrats. I'm still firing off email to random people looking for me one true e-pen-pal.
Germany? Can you elaborate on this? I've read a little bit about CoS but I wasn't aware that Germany was doing anything in particular. I'm curious now.
Could you describe your software solution in more detail? What package are you using. Does it run on Linux or is it a commercial Windows app? I'd be interested in setting up something similar. Thanks
My consulting company recently added security as a service we offer. That made it a separate group within our company. Since that time all manner of security-related things have been pushed off onto this group. I see this as a very bad way of practicing security. We don't need a special group of people to come in and harden or servers & network after we deployed it. We need to utilize good security practices as we're building the infrastructure. A secure architecture isn't something you can tack on after the project is 99% complete. It's something that has to be designed into the project from day one. "How do we achieve our goal in a secure fashion?" or at the very least, "How do we achieve our goal with the minimal acceptable risk to our security" because everything has an implied security risk. I believe this new direction for us will ultimately lead to replication of work as the security people try to rebuild our product after the fact. This obviously increases the number of billable hours on a project. Inevitably since they aren't network engineers or systems engineers they will over-secure something to the point of not working. This in turn creates even more billable hours to the customer. Perhaps that's what my company is really after....
It's naive to think that the RF bleed would only interfere with actual electrical devices. You forget that there are 135 miles of wire in a typical 747, 238 in Air Force One, and over 300 miles of wire in an A380. The wires are everywhere. You're no more than half the height of the cabin from wires (since wires run over the cabin and under the cabin, hell all around the cabin). All modern air planes are "fly by wire" planes. Man over hydraulic controls are only seen in a few minor or secondary systems. Everything else is digital. It is extremely naive to think that interference on the wires themselves can't cause problems.
Mammaries? What are these mammaries that you speak of?
Remember, this is Slashdot. You're going to have to explain yourself a little better. Here, let me help you.
I read about this stuff nearly a decade ago. Why is this news now?
Apparently you've never been to a gym. I've been to 2 over these past few months. One was an exclusive gym and the other was one of the area YMCAs. While there were a few ugly ducklings at each place both sites by and large were easily 90% filled with very attractive people.
That's interesting. One of our customers, for which we are the outsourced IT department, was bought by Accenture last year. This company actually had its founding roots in McDonalds and Pizza Hut. Interesting stuff...
I've actually never pronounced SQL as "sequel". I've always pronounced it as "squall" as in "my squall". I don't generally say MS Squall though. I usually just spell it out. That's just me of course but that's always the most common way I've heard it said around these parts. People say "sequel" if they mean to say SQL by itself and they say "squall" when part of MySQL.