" But Jonathan Rosenberg, vice president of the products group at Google, said the Gmail announcement was legitimate. He did concede that the company did get caught up in the spirit of April Fool's Day in its press release. "
There are a bunch of areas in Linux that can help you. Only some of them are routing based.
The first thing I would try would be to setup one of your lines with 'tc' and bandwidth shape the line with CBQ and SFQ. CBQ will let you set the outbound "rate" for the line, and SFQ will enforce "fairness" between different "connections". This should keep ftp uploads from swamping upstream traffic and pushing your ping times thru the sky. You can do some similar things with 'tc' ingress policies to shape the incoming traffic, but this is less effective.
If you still want to try two lines, here is the basic setup.
You need a Linux box that has three network interfaces. One for each of your T-1s, and one for your local LAN. The Linux box's IP address is the default gateway for everyone on your local LAN.
You setup a firewall on the Linux box with something like:
ip ru add fwmark 1 table 10
ip route add default via IP_ADDRESS_OF_T1#1 dev eth1
ip route add default via IP_ADDRESS_OF_T1#2 dev eth2 table 10
This is far from complete (and I haven't tested it), but it should set "fwmark" to 1 for HTTP traffic. The router table should then take traffic with FWMARK set to 1 and use routing table 10 instead of the default table, which can have a different default route. In that both eth1 and eth2 are MASQed, both will NAT.
You will need a lot more here to be fully functional. You need to completely filter the traffic you don't want, and probably classify a bunch more stuff along the way.
Re: the 91 express lanes, it may be true that there are other private projects in CA, but this one is particularily troublesome. It seems that in order to get the private funding, the state had to agree not to expand the 91 and other adjacent roads. The state wants to do just that, but this would threaten the profit of the toll road. The owners of the 91 lanes have sued the state and have won, preventing other road projects. In the end, the state will end up buying the 91 lanes back from the private owners at massive profits to the owners. This is easily one of the biggest boondoggle's in the history of road construction.
I agree that this is not the owners fault (the state really screwed up on how they contracted this), but the profit motive for the road has still gotten in the way of transportation. And you will note that none of this has anything to do with "congestion pricing".
The other toll roads in the area are completely different from the 91 in how the ownership works. They are basically government owned, private, corporations that self finance. This is important. With the 91, excess profits go to the owners. With the other toll ways, profits above the bonds that built the roads stay in the road agency and are thus actually owned by the state. The "profit" incentive is just not the same.
In terms of "patents", I cannot disagree with you more. The "Intellectual Property" system is a "legal" racket that no longer serves it's intended purpose. The idea behind copyright and patent law is to promote creativity by providing a "limited term" monopoly. This "balancing act" is now so far out of balance as to be laughable, and only benefits the large corporations and their lawyers. Copyright law has been abused and extended to the point that nothing has entered the public domain in 50 years. Patents have been issued for the trivial and absurd. Only the lawyers and big companies win. And all this is over the ownership of "ideas". It is not a part of the constitution that "ideas" can always be owned forever and building a system of laws that allows ideas to be owned oppresses the masses. Businesses should make money by making "things", not by buying bankrupt companies for their patent portfolio.
Private roads are a disaster. The incentive of a road is to move people. This is not the incentive of a private road. The incentive of a private road is to make as much money as possible, even if this means leaving 80% of the people at home.
If you don't think this works this way, just look at the only private "road" in California. The '91 express lanes'. This is an example of the government, and thus the people, being abused by a private road company that want's to guarantee their profits.
Private enterprise is not always any more efficient than public works. In fact, if private enterprise has a monopoly, it is usually much less efficient than public projects. Take 20% of the money currently spent on pharmacuticals and pump it into university research projects that are not patented and you get 100% of the current rate of discoveries. Levitra costs $4/month instead of $400/month. The bottom line is that the current "montra" of "private enterprise is always more efficient" is literally killing thousands of people.
I am part owner in a hoster and we run UML for just about everything that we do. Even our in-house mail and web servers are running under UML.
Pricing UML servers is a strange "artform". In your case, wanting 60G of disk pushes you up into the land of dedicated servers at $79 to $125 per month. This is probably a lot more processing power than you need.
The best offer that we would have for you is:
15 Gig disk
256Meg RAM (128 Meg core and 128 Meg swap running out of tmpfs space)
40G/month of transfer
$50/month pre-pay 1yr + $0 setup or
$60/month billed quarterly + $100 setup
Our servers are available either located in Philadelphia with internet connectivity on Level-3, Verio, or AT&T (your choice), or in Los Angeles with internet connectivity BGP'd between Level-3, Yipes, and Williams. The underlying servers are AMD Athlon XP2200's with 1Gig of ram, Raid-1 80Gig drives, and 100Mbit switched connectivity that is not bandwidth limited. In general, bursting to a couple of megabits from individual connections is pretty easy.
Our servers are designed to not be overcommitted. We basically cut a server up into 20 increments. Each increment gets 32 Meg of ram and 32 Meg of swap. The swap is run out of a "tmpfs" mount on the underlying host, so it is usually ram as well. Running this way, the only users that have ever complained about performance were running applications that ran their virtuals out of RAM, not the underlying host.
The servers themselves have an SSH based console that lets you see processes, kill your virtual, restart your virtual, and "become" the virtuals console upon a reboot. This lets you fix filesystem errors and watch bootup, etc.
We have options for local and off-site backups, multi-patch connectivity for green-screen applications, additional IP addresses, including IP addresses on multiple backbones, local and remote replicated file systems, and a bunch of other stuff.
In about 2 weeks, we are reworking our website with new UML offerings that are more flexible, allowing users to buy RAM and disk seperately, configure server pairs as hot-backup systems, buy dedicated physical systems with UML pre-configured so you can sell your own UML "sub hosts", and more. Of particular interest to a lot of users will be user-manageable external firewalls, and managed security services so that you don't need to worry about patches, etc.
You can call our offices (yes there are real humans here) at (610) 237-2000 or (800) 470-2756. Our AUP is pretty simple, don't spam and we are happy. Depending on which of our networks you want to be on, IRC servers are allowable, but check with us first so that we are sure they will work for you.
One last point for people considering getting into the UML hosting business. There is an unbelievable amount of fraud out there. We get >75% bogus orders. There are orders with real credit card numbers with real names and phone numbers presumably coming from Russian or Far East organized crime. We have setup our system to "require" 100% phone authorization before we turn a server on.
Doug Dumitru
EasyCo LLC
610 237-2000
doug@easyco.com (I get so much spam, a little more can't hurt)
If you want to limit BW by IP address, this might be doable depending on what the server is. If the server is a Linux or "virtual Linux" box, you can probably use 'tc' (Traffic Control) in the kernel to meter bandwidth by subnet or address. This works pretty well. Look at the advanced routing howto for info. It is a bear to setup, but actually works quite well.
The problem is that if these are bots grabbing your whole site, slowing them down to 10K/sec won't actually reduce the amount of traffic they pull from you. They may take all day to get the pages, but the bytes will still move.
Some options that you have.
* If the user really doesn't need the data, block their address entirely.
* consider blocking the 'bots' "client" signature. You can do this in
Apache. "Respected" bots don't lie about who they are. If a bot does
lie, then it is a DOS attack in disguise.
* Contact the users, if you can.
* If you want the user to get a mirror, setup something to actually do the
mirror that is effective. I would recommend running rsync.
This is an HTML w/o style tags mirror of the article itself (no posted comments).
Thanks Pamela et.al. Great Article http://sd-mirror.dumitru.com/gl.htm
Dial UP 28K up 53.6K down 295Meg/day up 564meg/day down
DSL 128K up 384K down 1.3Gig/day up 4Gig/day down
Cable 256K up 3M down 2.7Gig/day up 31gig/day down
So using the Cox numbers (Cox is who I have and I want to compliment them on giving out honest numbers), this is:
Upload: 67 hours of max uploads/month or 9% duty cycle
Download: 39 hours of max downloads/month or 5% duty cycle
So they are working against about a 5 to 10% duty cycle. If you are using the service for "interactive" usage and not "automated" usage, then the limits are "way out there". If you want to run bittorent or kazaa, then you are hosed, but these are not "interactive" usage.
For non-server usage this is a lot. Lets say you listen to internet radio at 48K/sec. Even at 24 hours/day, this is only 14Gig or less than 50% of the usage limit.
Patents are meant to protect new "inventions", not "innovations". To patent something is to ask the government to force it's citizens, under threat of imprisionment, to enforce a monopoly. The theory behind patents is to encourage "inventors" to "invent" stuff.
A patent should not be based on what something does, but on how it does it. Tivo has patented a set of functions, not how the functions are implemented. If you gave a hypothetical competent engineer, at the time that Tivo "invented" their PVR, a list of the features that Tivo does and asked them how to implement this, that engineer would very likely come up with the same "techniques" that Tivo claims they invented. What a Tivo does is interesting engineering, but it is not an "invention".
Before the days of digital video recording, broadcasters wanted "instant replay" for their sports events. In 1978, I interned for ABC sports as a college "assitant" during a college bowl game. They had this neat device that recorded video onto a magnetic platter. Total recording time was only about 30 seconds, but you could play back from anywhere, at any speed, while the recording was still in progress. This is the 100% analog solution that lets you zip around a video stream.
There were obviously limitations of the technology. First, was is only 30 seconds. Second, they were notoriously flakey. Third, they cost a lot more than a car.
Tivo is an obvious example of using new technology to do something that a lot of people have wanted to do for a long time. None of the individual elements of Tivo are new. The encoding of the video/audio is uninteresting. The creating of sliding windows is trivial and obvious. Tivo did not invent the hard disk, the encoding algorithsm, the D/A converts. Come to think of it, they are just an integrator putting together obvious elements.
Regardless of the economics of Linux and free software in the OS space, it cannot die because there is no real alternative.
Consider the marketplace outside of i86/PCs.
Embedded systems devices are heading for Linux in droves. The rule here is "cost" "cost" "cost". Saving 25 cents is important. Any license from Microsoft or anyone else is a natural for the axe. Device manufacturers are not too worried about the GPL and losing their "advantage" in the software area (except maybe for Cisco which already outright owns an OS). If you are building a settop box and don't have a marketting tie-in with Microsoft, you will probably use Linux or BSD. Anything else is just too long to market and adds too much of a drag without giving anything back to the project.
In-house OS's will die as well. With new "router" projects from Linux/BSD, the few remaining in-house OS's will have trouble maintaining their own weight. Cisco is the last player here and expect their competitors to build embedded Linux and/or BSD based products that are real competitors at every market nitch except for the really high-end stuff (and maybe there as well). If Cisco were smart, they would re-engineer IOS as a low-level "switch matrix" service and drive it from Linux or BSD.
Open source will continue to create an environment of Microsoft and Open Source with little room for anyone else. Netware has already been killed in this way. If Linux/Samba did not exist, then Novell would still have a proprietary file/print server market. The same holds true for compiler tools. If GNU compilers were not around, Borland might be.
On-line services will continue to use Linux in droves. This is a dog eat dog business that will only pay Microsoft when the customer demands (and pays for) it. For everything else, free is it. And these companies do use the source code as well. They fix patches manually and often contribute back.
Unix is dead. Even without the SCO mess, Unix was in trouble. The tipping point was about 18 months ago when Linux began to equal Unix in the mid-market in terms of performance, reliability, and features. Actually, the performance and feature part is definately a Linux win right now. The reliability part is getting there. With the SCO mess, can you possibly image a "new" Unix licensee starting up. If you are a current Unix vendor, you can either continue pouring massive amounts of money into your own *nix or work to make Linux run on your box. In the past, customer perception was that the proprietary *nix systems were better and worth the money. Now, they may have this opinion about your hardware, but the OS is not assigned a lot of value unless you are at the very high end. Sounds like time to transition the *nix development/support divisions to supporting/contributing to Linux.
Open source is very hard to kill. If you run Linux, you personally have rights to it. No one can take those rights away from you. Even Linus, the copyright holder, cannot take back the GPL. If a company steps away from Linux, there is still more than enough to keep it moving forward. The only way to kill Linux is to have it die by stagnation. If DRM becomes so pervasive and it's implementation so restrictive that Linux is locked out, then Linux might decline over a very large number of years. If governments step in and mandate proprietary software in order to "protect" it's industries, then the same might happen. More likely is the proposition that whole industries, governments, and societies will start to mandate open practices that just happen to be best implemented in open source. Governments will require documents be stored in freely readable data formats. Inter-system communications will happen over non-proprietary protocols. We already see this trend en-masse. Just look at the internet with TCP/IP, HTTP, XML, VOIP, etc. And finally, international entities will continue to look for any way out of the Microsoft tax.
All in all, the Forbe's authors seem to be mimicking their boss. It is called being an ideol
In all fairness, you would probably need a healthy system running Asterisk to keep up with 96 simultanious VOIP codecs, but still.
I personally think that Digium is doing themselves a bit of a disservice by not shipping a hardware-less / VOIP-only config as a stock distribution. This lets you play without any hardware at all, which is what it takes to get up to speed.
You might think that the Digium hardware is a bit expensive, but this depends on your reference point. VOIP phones for $400 to $65 each are expensive compared to analog phones at $5.95. And yes you do need a PC, but compared to the other VOIP solutions out there, Asterisk is a great solution at an unheard of price.
If you need programmable, supported, commercial PBX functions, then you can always run Artisoft. VOIP channels are $200+ each and a loaded system can hit $10K in a heartbeat. They work, but are definately not open.
The Digium hardware appears to be competitively priced for a "nitch market" product. $100 per port is not at all out of line. Compared to some VOIP products, it is downright cheap. For example, one company offers an 8-port VOIP to analog box for $2500. For larger installations, Digium's cards are also pretty good. A T1/E1 interface for $600 (24 lines, so $25/line) and a quad-T1/E1 for about $1500 (96 lines at $15.63/line) is a pretty low cost per line.
I think that the issue here is that Asterisk is not targetted for a "home PBX". On the other hand, being able to register with SipPhone, Free World Dialup, etc for inbound and outbound calls, all from your DSL line is "very interesting". And if you are interested in a "home PBX", you can run Asterisk completely without Digium (or any other telephony) hardware. SIP/IAX run completely in software over IP. You can setup Asterisk, free softphones like X-lite, and free SIP services like Sipphone and Free World Dialup all without spending any money at all. You just need a PC on-net running Linux. What is not to like.
A couple of people have mentioned Asterisk, but have not given much in terms of details (I am currently setting up an Asterisk box for my business so I have read far too much).
Asterisk is a software package (GPL) for Linux (and BSD, although less supported) that provides PBX functionality with decent "call routing" decision making abilities. It is all command-line controlled with text files in "win-ini" format.
To drive your phone lines, you use a card called an FXO card. * is developed by a company called Digium and they sell hardware. From Digium, an FXO is about $99/line and requires a PCI slot. This allows * to answer your analog line, read the caller-id info, etc. * can also direct outbound calls to your line.
To drive analog phones in your house, you can use an FXS card. From Digium, they have a card with 1-4 daughter boards supporting 1-4 lines. Cost is $125 for 1 and about $325 for 4 line support.
You can also drive analog phones by using a VOIP "gateway". These are small devices that have 1 or 2 FSX (phone) ports and a network connection. Single port units from Grandstream are about $75 and there is a new 2-phone unit for about $120. One neat part about using VOIP for local phones is that the hardware is relocatable anywhere.
Of course, actual VOIP phones can be used as well. These range from $400 from companies like Cisco to ~$65 for the Grandstream Budgetone (which I haven't tested yet).
There are also a number of "softphones" available for Windows and X. I have used X-lite but there are a bunch of others.
Your * box itself should not be behind a firewall if you want to reach it over the internet. In that it is a linux box, my suggestion would be to use it "AS" your firewall. Having a static IP address is helpful, but dynamic DNS works as well so that your box can be located from the outside.
Having the * box on the internet lets you add "extension" phones running VOIP nearly anywhere with a net connection. The "client" phones usually pierce firewalls and NAT routers without problems.
* can also "register" with a number of SIP compatible service providers. I have not tried this with Vonage, but my test * box can talk with "Free World Dialup" and "Sipphone". This lets you place calls to these services, plus receive calls from them. Both of these services are free. Other, for fee services are available that offer low cost LD, inbound calling, etc.
* boxes themselves can network to each other using a protocol called IAX (Internet Asterisk Exchange). This is a similar format to VOIP, but combines multiple calls into single packets. This lowers the packet overhead to support multiple connections.
Configuring * requires some planning. It is a PBX with a "dial plan". This controls what you dial and where calls go to. It does support VoiceMail, fallover trunking, and a bunch of other stuff.
A really good source for * info in the "voip-wiki" at http://www.voip-info.org/wiki-Asterisk
I just pulled up/usr/include/asm/errno.h from a SCO OpenServer V system I
have access to (which has a legal license). I compared this with/usr/src/linux-2.4.22/include/asm-i386/errno.h
I will not post the file here, but a couple of points are obvious.
1. The comments between the two versions are different. Even specific error
message comments vary, such as:
SYMBOL - Linux - SCO
EPERM - Operation Not Permitted - Not Owner
E2BIG - Argument list too long - Arg list too long
EDOM - Math argument out of domain - Argument out of domain
and on and on.
2. There are a bunch of defines in Linux that don't exist in SCO.
3. Some of the defines are completely for different things:
SCO - EBADE 50/* Bad Exchange Descriptor */
Linux - ENOCSI 50/* No CSI structure available */
4. The SCO file has a bunch of errors for things like TCP errors that aren't
in the Linux file at all.
5. The formatting of comments is very different.
In general, there is no way that the Linux code is a simple cut and paste of
the SCO code, at least at this level.
Maybe the code started out closer, but all that is left is symbols and numbers.
The numbers are arbitrary and vary from target to target. The symbols are a
part of the POSIX spec. The files are available under a BSD license. Just how
is this infringing on SCO's copyright and even if it were, just what are the
damages.
Now come on. Any co-lo worth it's weigth in salt can throw 200+ megabits at a target without their own users even noticing. The fact that it was only 20mbits implies ~2000 zombies (100kbits/sec each, average).
The scary thought is the zombies that are in co-lo. Back during NIMDA, I had worm scans from one of Intel's software distribution servers. That one system could have hit SCO at 20Mbits and still handled it's normal user load without breaking a sweat.
You should ask the same questions and perform the same due-dilligence as if you were paying for the service.
o Keep you own backups of your site out of the hands of the hoster.
o Make sure that any contracts you sign keep your stuff yours.
o Keep your domain names under your control.
o Have a contingency plan in case things don't work out.
If you were paying for service, you would probably be asking for SLA and reviewing technical details about the hosting plant. With free service, it is a bit harder to ask for too much. Plus, how much of a rebate do you want against zero.
Once you are up and running, be sure to give the hoster appropriate thanks and credit. A link on your home page is probably the least you should do. Also, don't do things that cost the hoster money and/or aggrevation. Be self-sufficient with your servers and applications. Use the bandwidth that you are getting sparingly. Don't overdue the photos and graphics. Just because you are getting something for free doesn't mean you should use as much as possible.
Also, try not to attract riff-raff with your project. You are getting stuff for free. Returning the favor with a DDOS attack is probably not the best idea.
You could argue that there is a secret agenda, to destroy commercial software. The FSF does have this as an agenda and it is obvious that Darl thinks this is a criminal conspiracy that threatens his way of life.
In some respects, Darl is right that others are conspiring against him. A more accurate description is that others are participating in a community and making what he does utterly uncompetitive. While there is no "free" software as Darl describes it, why would anyone pay for SCO OpenServer as opposed to paying less for Linux from a variety of vendors. There is absolutely no business reason for a customer to use SCO compared to Linux. And the business case for AIX and Solaris, while not as bleak as OpenServer, are still on the wrong end of the slope.
I am now quite convinced that Darl thinks he is doing the software industry a service by "killing off" open source. He appears to be truely convinced that the GPL builds an industry that precludes using copyright for financial profit, and that the purpose of copyright law is to protect profit. Thus the GPL violates copyright law. Mind you, he is on drugs (a metaphore, not an accusation) and if his lawyers argue this in court, they should be escourted to the door by a big guy with a gun never to return.
The real problem with Darl is that the GPL was written very carefully to prevent the extortion that is being attempted. Darl wants to get money for what he thinks he owns, which is all of Unix. He thinks that IBM should pay him more royalties for Unix even though they have a paid up license. He thinks that Linux "is" Unix and that he controls everything that "looks" and "smells" like Unix because there is no way that hobbiests could ever produce something that threatens him without stealing. Unfortunately for Darl, he may prevail is showing some limited cases where code was moved improperly. Along the way, there are a lot of dirty hands and it will be near impossible to get real money.
Consider:
o Linux users really have done nothing wrong. They don't have a license with SCO. They did not steal code. Yet is is Linux users that Darl wants money from.
o Linux developers have offered to remove offending code, and the GPL requires that they do so.
o SCO wants the offending code to remain so that they can collect big bucks. The GPL does not allow this. Somehow Darl thinks that removing the offending code does not cure the problem. He is right, removing the offending code does not cure his problem, which is a lack of revenue.
I almost hate to write this, as SCO's lawyers might actually use it.
About the only attack on GPL that makes sense is that GPL creates a community which is colluding with each other in an anti-competitive manner artificially driving down prices. GPL code constitutes a "criminal cartel" with the members conspiring with each other to profit from the conspiracy and thus unfairly competing with closed-source products.
You can even argue that the GPL acts to preclude and punish users that don't participate in the community by pushing commercial products out of the marketplace with artificially deflated prices.
Thus, the FSF movement comprises the largest anti-trust entity ever to exist and this must be squashed in order to protect the just interests of world commerce.
The take of the open letter is that "Copyright exists to support profit" and that "GPL precludes using copyright to enforce profit, and even competes with non-GPL making profit for difficult". What is missed is that Copyright exists to enforce property rights. If the author chooses to leverage his copyright with thousands of others in a cooperative work, then this is the free choice of the author. If non-GPL vendors find this disturbing, then this just indicates that competition works.
And the whole take that the copyright laws and constitution are designed to equally protect business and "the people" is bunk. The people "vote", businesses only "lobby". Contrary to business belief, they don't own the people, yet.
When updating an older server (7.1, I think), the RH RPM failed with a GLIBC dependency. The updates for RH are identical for 7.1 - 9, so you might have a problem here.
My easiest workaround was to rebuild the rpm from source with:
Get the rsync-2.5.7-0.9.src.rpm from RedHat ftp server updates.redhat.com
Install the source rpm with:
rpm -ivh/tmp/rsync-2.5.7-0.9.src.rpm
Build a new complete, clean set of RPMs with:
cd/usr/src/redhat/SPECS
rpm -bb rsync.spec
The new installable binary for your current lib versions is in/usr/src/redhat/RPMS/i386, so you can install it with:
For those that don't use rsync, this is easily one of the most useful utilities on the box. I particularily like "modules" mode over ssh. Setup an ssh key and have the key auto-run rynnc --daemon. You get modules and ssh. Really cool.
Slashdot Mirror
Apparently, this is not an April Fools joke, although
l e_ email/index.htm?cnn=yes
http://money.cnn.com/2004/04/01/technology/goog
Quoting:
" But Jonathan Rosenberg, vice president of the products group at Google, said the Gmail announcement was legitimate. He did concede that the company did get caught up in the spirit of April Fool's Day in its press release. "
There are a bunch of areas in Linux that can help you. Only some of them are routing based.
The first thing I would try would be to setup one of your lines with 'tc' and bandwidth shape the line with CBQ and SFQ. CBQ will let you set the outbound "rate" for the line, and SFQ will enforce "fairness" between different "connections". This should keep ftp uploads from swamping upstream traffic and pushing your ping times thru the sky. You can do some similar things with 'tc' ingress policies to shape the incoming traffic, but this is less effective.
If you still want to try two lines, here is the basic setup.
You need a Linux box that has three network interfaces. One for each of your T-1s, and one for your local LAN. The Linux box's IP address is the default gateway for everyone on your local LAN.
You setup a firewall on the Linux box with something like:
LAN on eth0
T1 on eth1
T1 on eth2
iptables -i eth0 --dport 80 --state NEW,ESTABLISHED --set-mark 1
iptables -t nat -o eth1 -j MASQUERADE
iptables -t nat -o eth2 -j MASQUERADE
ip ru add fwmark 1 table 10
ip route add default via IP_ADDRESS_OF_T1#1 dev eth1
ip route add default via IP_ADDRESS_OF_T1#2 dev eth2 table 10
This is far from complete (and I haven't tested it), but it should set "fwmark" to 1 for HTTP traffic. The router table should then take traffic with FWMARK set to 1 and use routing table 10 instead of the default table, which can have a different default route. In that both eth1 and eth2 are MASQed, both will NAT.
You will need a lot more here to be fully functional. You need to completely filter the traffic you don't want, and probably classify a bunch more stuff along the way.
Good luck.
Maybe they agreed to host sco.com
With all of the DDos attacks and bad press, I would charge them $1M/month for that.
It is kinda like the premium that hosters charge spammers. How about some $100/Gig BW.
Re: the 91 express lanes, it may be true that there are other private projects in CA, but this one is particularily troublesome. It seems that in order to get the private funding, the state had to agree not to expand the 91 and other adjacent roads. The state wants to do just that, but this would threaten the profit of the toll road. The owners of the 91 lanes have sued the state and have won, preventing other road projects. In the end, the state will end up buying the 91 lanes back from the private owners at massive profits to the owners. This is easily one of the biggest boondoggle's in the history of road construction.
I agree that this is not the owners fault (the state really screwed up on how they contracted this), but the profit motive for the road has still gotten in the way of transportation. And you will note that none of this has anything to do with "congestion pricing".
The other toll roads in the area are completely different from the 91 in how the ownership works. They are basically government owned, private, corporations that self finance. This is important. With the 91, excess profits go to the owners. With the other toll ways, profits above the bonds that built the roads stay in the road agency and are thus actually owned by the state. The "profit" incentive is just not the same.
In terms of "patents", I cannot disagree with you more. The "Intellectual Property" system is a "legal" racket that no longer serves it's intended purpose. The idea behind copyright and patent law is to promote creativity by providing a "limited term" monopoly. This "balancing act" is now so far out of balance as to be laughable, and only benefits the large corporations and their lawyers. Copyright law has been abused and extended to the point that nothing has entered the public domain in 50 years. Patents have been issued for the trivial and absurd. Only the lawyers and big companies win. And all this is over the ownership of "ideas". It is not a part of the constitution that "ideas" can always be owned forever and building a system of laws that allows ideas to be owned oppresses the masses. Businesses should make money by making "things", not by buying bankrupt companies for their patent portfolio.
You are clueless.
Private roads are a disaster. The incentive of a road is to move people. This is not the incentive of a private road. The incentive of a private road is to make as much money as possible, even if this means leaving 80% of the people at home.
If you don't think this works this way, just look at the only private "road" in California. The '91 express lanes'. This is an example of the government, and thus the people, being abused by a private road company that want's to guarantee their profits.
Private enterprise is not always any more efficient than public works. In fact, if private enterprise has a monopoly, it is usually much less efficient than public projects. Take 20% of the money currently spent on pharmacuticals and pump it into university research projects that are not patented and you get 100% of the current rate of discoveries. Levitra costs $4/month instead of $400/month. The bottom line is that the current "montra" of "private enterprise is always more efficient" is literally killing thousands of people.
Pricing UML servers is a strange "artform". In your case, wanting 60G of disk pushes you up into the land of dedicated servers at $79 to $125 per month. This is probably a lot more processing power than you need.
The best offer that we would have for you is:
- 15 Gig disk
- 256Meg RAM (128 Meg core and 128 Meg swap running out of tmpfs space)
- 40G/month of transfer
- $50/month pre-pay 1yr + $0 setup or
- $60/month billed quarterly + $100 setup
Our servers are available either located in Philadelphia with internet connectivity on Level-3, Verio, or AT&T (your choice), or in Los Angeles with internet connectivity BGP'd between Level-3, Yipes, and Williams. The underlying servers are AMD Athlon XP2200's with 1Gig of ram, Raid-1 80Gig drives, and 100Mbit switched connectivity that is not bandwidth limited. In general, bursting to a couple of megabits from individual connections is pretty easy.Our servers are designed to not be overcommitted. We basically cut a server up into 20 increments. Each increment gets 32 Meg of ram and 32 Meg of swap. The swap is run out of a "tmpfs" mount on the underlying host, so it is usually ram as well. Running this way, the only users that have ever complained about performance were running applications that ran their virtuals out of RAM, not the underlying host.
The servers themselves have an SSH based console that lets you see processes, kill your virtual, restart your virtual, and "become" the virtuals console upon a reboot. This lets you fix filesystem errors and watch bootup, etc.
We have options for local and off-site backups, multi-patch connectivity for green-screen applications, additional IP addresses, including IP addresses on multiple backbones, local and remote replicated file systems, and a bunch of other stuff.
In about 2 weeks, we are reworking our website with new UML offerings that are more flexible, allowing users to buy RAM and disk seperately, configure server pairs as hot-backup systems, buy dedicated physical systems with UML pre-configured so you can sell your own UML "sub hosts", and more. Of particular interest to a lot of users will be user-manageable external firewalls, and managed security services so that you don't need to worry about patches, etc.
Our website is: http://mirroredsolutions.com
You can call our offices (yes there are real humans here) at (610) 237-2000 or (800) 470-2756. Our AUP is pretty simple, don't spam and we are happy. Depending on which of our networks you want to be on, IRC servers are allowable, but check with us first so that we are sure they will work for you.
One last point for people considering getting into the UML hosting business. There is an unbelievable amount of fraud out there. We get >75% bogus orders. There are orders with real credit card numbers with real names and phone numbers presumably coming from Russian or Far East organized crime. We have setup our system to "require" 100% phone authorization before we turn a server on.
Doug Dumitru
EasyCo LLC
610 237-2000
doug@easyco.com (I get so much spam, a little more can't hurt)
I thought it was supposed to be called:
Unobtainium
or
Reallyexpensium
If you want to limit BW by IP address, this might be doable depending on what the server is. If the server is a Linux or "virtual Linux" box, you can probably use 'tc' (Traffic Control) in the kernel to meter bandwidth by subnet or address. This works pretty well. Look at the advanced routing howto for info. It is a bear to setup, but actually works quite well.
The problem is that if these are bots grabbing your whole site, slowing them down to 10K/sec won't actually reduce the amount of traffic they pull from you. They may take all day to get the pages, but the bytes will still move.
Some options that you have.
* If the user really doesn't need the data, block their address entirely.
* consider blocking the 'bots' "client" signature. You can do this in
Apache. "Respected" bots don't lie about who they are. If a bot does
lie, then it is a DOS attack in disguise.
* Contact the users, if you can.
* If you want the user to get a mirror, setup something to actually do the
mirror that is effective. I would recommend running rsync.
This is an HTML w/o style tags mirror of the article itself (no posted comments).
Thanks Pamela et.al. Great Article
http://sd-mirror.dumitru.com/gl.htm
All internet accounts are limited:
Dial UP 28K up 53.6K down 295Meg/day up 564meg/day down
DSL 128K up 384K down 1.3Gig/day up 4Gig/day down
Cable 256K up 3M down 2.7Gig/day up 31gig/day down
So using the Cox numbers (Cox is who I have and I want to compliment them on giving out honest numbers), this is:
Upload: 67 hours of max uploads/month or 9% duty cycle
Download: 39 hours of max downloads/month or 5% duty cycle
So they are working against about a 5 to 10% duty cycle. If you are using the service for "interactive" usage and not "automated" usage, then the limits are "way out there". If you want to run bittorent or kazaa, then you are hosed, but these are not "interactive" usage.
For non-server usage this is a lot. Lets say you listen to internet radio at 48K/sec. Even at 24 hours/day, this is only 14Gig or less than 50% of the usage limit.
Bunk,
Patents are meant to protect new "inventions", not "innovations". To patent something is to ask the government to force it's citizens, under threat of imprisionment, to enforce a monopoly. The theory behind patents is to encourage "inventors" to "invent" stuff.
A patent should not be based on what something does, but on how it does it. Tivo has patented a set of functions, not how the functions are implemented. If you gave a hypothetical competent engineer, at the time that Tivo "invented" their PVR, a list of the features that Tivo does and asked them how to implement this, that engineer would very likely come up with the same "techniques" that Tivo claims they invented. What a Tivo does is interesting engineering, but it is not an "invention".
Before the days of digital video recording, broadcasters wanted "instant replay" for their sports events. In 1978, I interned for ABC sports as a college "assitant" during a college bowl game. They had this neat device that recorded video onto a magnetic platter. Total recording time was only about 30 seconds, but you could play back from anywhere, at any speed, while the recording was still in progress. This is the 100% analog solution that lets you zip around a video stream.
There were obviously limitations of the technology. First, was is only 30 seconds. Second, they were notoriously flakey. Third, they cost a lot more than a car.
Tivo is an obvious example of using new technology to do something that a lot of people have wanted to do for a long time. None of the individual elements of Tivo are new. The encoding of the video/audio is uninteresting. The creating of sliding windows is trivial and obvious. Tivo did not invent the hard disk, the encoding algorithsm, the D/A converts. Come to think of it, they are just an integrator putting together obvious elements.
I hope they get squashed.
Regardless of the economics of Linux and free software in the OS space, it cannot die because there is no real alternative.
Consider the marketplace outside of i86/PCs.
Embedded systems devices are heading for Linux in droves. The rule here is "cost" "cost" "cost". Saving 25 cents is important. Any license from Microsoft or anyone else is a natural for the axe. Device manufacturers are not too worried about the GPL and losing their "advantage" in the software area (except maybe for Cisco which already outright owns an OS). If you are building a settop box and don't have a marketting tie-in with Microsoft, you will probably use Linux or BSD. Anything else is just too long to market and adds too much of a drag without giving anything back to the project.
In-house OS's will die as well. With new "router" projects from Linux/BSD, the few remaining in-house OS's will have trouble maintaining their own weight. Cisco is the last player here and expect their competitors to build embedded Linux and/or BSD based products that are real competitors at every market nitch except for the really high-end stuff (and maybe there as well). If Cisco were smart, they would re-engineer IOS as a low-level "switch matrix" service and drive it from Linux or BSD.
Open source will continue to create an environment of Microsoft and Open Source with little room for anyone else. Netware has already been killed in this way. If Linux/Samba did not exist, then Novell would still have a proprietary file/print server market. The same holds true for compiler tools. If GNU compilers were not around, Borland might be.
On-line services will continue to use Linux in droves. This is a dog eat dog business that will only pay Microsoft when the customer demands (and pays for) it. For everything else, free is it. And these companies do use the source code as well. They fix patches manually and often contribute back.
Unix is dead. Even without the SCO mess, Unix was in trouble. The tipping point was about 18 months ago when Linux began to equal Unix in the mid-market in terms of performance, reliability, and features. Actually, the performance and feature part is definately a Linux win right now. The reliability part is getting there. With the SCO mess, can you possibly image a "new" Unix licensee starting up. If you are a current Unix vendor, you can either continue pouring massive amounts of money into your own *nix or work to make Linux run on your box. In the past, customer perception was that the proprietary *nix systems were better and worth the money. Now, they may have this opinion about your hardware, but the OS is not assigned a lot of value unless you are at the very high end. Sounds like time to transition the *nix development/support divisions to supporting/contributing to Linux.
Open source is very hard to kill. If you run Linux, you personally have rights to it. No one can take those rights away from you. Even Linus, the copyright holder, cannot take back the GPL. If a company steps away from Linux, there is still more than enough to keep it moving forward. The only way to kill Linux is to have it die by stagnation. If DRM becomes so pervasive and it's implementation so restrictive that Linux is locked out, then Linux might decline over a very large number of years. If governments step in and mandate proprietary software in order to "protect" it's industries, then the same might happen. More likely is the proposition that whole industries, governments, and societies will start to mandate open practices that just happen to be best implemented in open source. Governments will require documents be stored in freely readable data formats. Inter-system communications will happen over non-proprietary protocols. We already see this trend en-masse. Just look at the internet with TCP/IP, HTTP, XML, VOIP, etc. And finally, international entities will continue to look for any way out of the Microsoft tax.
All in all, the Forbe's authors seem to be mimicking their boss. It is called being an ideol
My understanding is that 100 VOIP channels with non-trivial codecs is Dual-Xeon land. While this is not "exotic", it is usually above the $1K land.
For 20-50 channels, you can run $50 AMDs.
So just how much is such a GW from Cisco, $30K.
In all fairness, you would probably need a healthy system running Asterisk to keep up with 96 simultanious VOIP codecs, but still.
I personally think that Digium is doing themselves a bit of a disservice by not shipping a hardware-less / VOIP-only config as a stock distribution. This lets you play without any hardware at all, which is what it takes to get up to speed.
You might think that the Digium hardware is a bit expensive, but this depends on your reference point. VOIP phones for $400 to $65 each are expensive compared to analog phones at $5.95. And yes you do need a PC, but compared to the other VOIP solutions out there, Asterisk is a great solution at an unheard of price.
If you need programmable, supported, commercial PBX functions, then you can always run Artisoft. VOIP channels are $200+ each and a loaded system can hit $10K in a heartbeat. They work, but are definately not open.
The Digium hardware appears to be competitively priced for a "nitch market" product. $100 per port is not at all out of line. Compared to some VOIP products, it is downright cheap. For example, one company offers an 8-port VOIP to analog box for $2500. For larger installations, Digium's cards are also pretty good. A T1/E1 interface for $600 (24 lines, so $25/line) and a quad-T1/E1 for about $1500 (96 lines at $15.63/line) is a pretty low cost per line.
I think that the issue here is that Asterisk is not targetted for a "home PBX". On the other hand, being able to register with SipPhone, Free World Dialup, etc for inbound and outbound calls, all from your DSL line is "very interesting". And if you are interested in a "home PBX", you can run Asterisk completely without Digium (or any other telephony) hardware. SIP/IAX run completely in software over IP. You can setup Asterisk, free softphones like X-lite, and free SIP services like Sipphone and Free World Dialup all without spending any money at all. You just need a PC on-net running Linux. What is not to like.
A couple of people have mentioned Asterisk, but have not given much in terms of details (I am currently setting up an Asterisk box for my business so I have read far too much).
Asterisk is a software package (GPL) for Linux (and BSD, although less supported) that provides PBX functionality with decent "call routing" decision making abilities. It is all command-line controlled with text files in "win-ini" format.
To drive your phone lines, you use a card called an FXO card. * is developed by a company called Digium and they sell hardware. From Digium, an FXO is about $99/line and requires a PCI slot. This allows * to answer your analog line, read the caller-id info, etc. * can also direct outbound calls to your line.
To drive analog phones in your house, you can use an FXS card. From Digium, they have a card with 1-4 daughter boards supporting 1-4 lines. Cost is $125 for 1 and about $325 for 4 line support.
You can also drive analog phones by using a VOIP "gateway". These are small devices that have 1 or 2 FSX (phone) ports and a network connection. Single port units from Grandstream are about $75 and there is a new 2-phone unit for about $120. One neat part about using VOIP for local phones is that the hardware is relocatable anywhere.
Of course, actual VOIP phones can be used as well. These range from $400 from companies like Cisco to ~$65 for the Grandstream Budgetone (which I haven't tested yet).
There are also a number of "softphones" available for Windows and X. I have used X-lite but there are a bunch of others.
Your * box itself should not be behind a firewall if you want to reach it over the internet. In that it is a linux box, my suggestion would be to use it "AS" your firewall. Having a static IP address is helpful, but dynamic DNS works as well so that your box can be located from the outside.
Having the * box on the internet lets you add "extension" phones running VOIP nearly anywhere with a net connection. The "client" phones usually pierce firewalls and NAT routers without problems.
* can also "register" with a number of SIP compatible service providers. I have not tried this with Vonage, but my test * box can talk with "Free World Dialup" and "Sipphone". This lets you place calls to these services, plus receive calls from them. Both of these services are free. Other, for fee services are available that offer low cost LD, inbound calling, etc.
* boxes themselves can network to each other using a protocol called IAX (Internet Asterisk Exchange). This is a similar format to VOIP, but combines multiple calls into single packets. This lowers the packet overhead to support multiple connections.
Configuring * requires some planning. It is a PBX with a "dial plan". This controls what you dial and where calls go to. It does support VoiceMail, fallover trunking, and a bunch of other stuff.
A really good source for * info in the "voip-wiki" at http://www.voip-info.org/wiki-Asterisk
Have fun.
I will not post the file here, but a couple of points are obvious.
1. The comments between the two versions are different. Even specific error message comments vary, such as:
and on and on.
2. There are a bunch of defines in Linux that don't exist in SCO.
3. Some of the defines are completely for different things:
4. The SCO file has a bunch of errors for things like TCP errors that aren't in the Linux file at all.
5. The formatting of comments is very different.
In general, there is no way that the Linux code is a simple cut and paste of the SCO code, at least at this level.
Maybe the code started out closer, but all that is left is symbols and numbers. The numbers are arbitrary and vary from target to target. The symbols are a part of the POSIX spec. The files are available under a BSD license. Just how is this infringing on SCO's copyright and even if it were, just what are the damages.
Sorry, that is 200 zombies (not 2000). Too little sleep.
Now come on. Any co-lo worth it's weigth in salt can throw 200+ megabits at a target without their own users even noticing. The fact that it was only 20mbits implies ~2000 zombies (100kbits/sec each, average).
The scary thought is the zombies that are in co-lo. Back during NIMDA, I had worm scans from one of Intel's software distribution servers. That one system could have hit SCO at 20Mbits and still handled it's normal user load without breaking a sweat.
Normally, I would say that they need to get their servers out of house, and into a co-lo. On the other hand, who would have them.
You should ask the same questions and perform the same due-dilligence as if you were paying for the service.
o Keep you own backups of your site out of the hands of the hoster.
o Make sure that any contracts you sign keep your stuff yours.
o Keep your domain names under your control.
o Have a contingency plan in case things don't work out.
If you were paying for service, you would probably be asking for SLA and reviewing technical details about the hosting plant. With free service, it is a bit harder to ask for too much. Plus, how much of a rebate do you want against zero.
Once you are up and running, be sure to give the hoster appropriate thanks and credit. A link on your home page is probably the least you should do. Also, don't do things that cost the hoster money and/or aggrevation. Be self-sufficient with your servers and applications. Use the bandwidth that you are getting sparingly. Don't overdue the photos and graphics. Just because you are getting something for free doesn't mean you should use as much as possible.
Also, try not to attract riff-raff with your project. You are getting stuff for free. Returning the favor with a DDOS attack is probably not the best idea.
You could argue that there is a secret agenda, to destroy commercial software. The FSF does have this as an agenda and it is obvious that Darl thinks this is a criminal conspiracy that threatens his way of life.
In some respects, Darl is right that others are conspiring against him. A more accurate description is that others are participating in a community and making what he does utterly uncompetitive. While there is no "free" software as Darl describes it, why would anyone pay for SCO OpenServer as opposed to paying less for Linux from a variety of vendors. There is absolutely no business reason for a customer to use SCO compared to Linux. And the business case for AIX and Solaris, while not as bleak as OpenServer, are still on the wrong end of the slope.
I am now quite convinced that Darl thinks he is doing the software industry a service by "killing off" open source. He appears to be truely convinced that the GPL builds an industry that precludes using copyright for financial profit, and that the purpose of copyright law is to protect profit. Thus the GPL violates copyright law. Mind you, he is on drugs (a metaphore, not an accusation) and if his lawyers argue this in court, they should be escourted to the door by a big guy with a gun never to return.
The real problem with Darl is that the GPL was written very carefully to prevent the extortion that is being attempted. Darl wants to get money for what he thinks he owns, which is all of Unix. He thinks that IBM should pay him more royalties for Unix even though they have a paid up license. He thinks that Linux "is" Unix and that he controls everything that "looks" and "smells" like Unix because there is no way that hobbiests could ever produce something that threatens him without stealing. Unfortunately for Darl, he may prevail is showing some limited cases where code was moved improperly. Along the way, there are a lot of dirty hands and it will be near impossible to get real money.
Consider:
o Linux users really have done nothing wrong. They don't have a license with SCO. They did not steal code. Yet is is Linux users that Darl wants money from.
o Linux developers have offered to remove offending code, and the GPL requires that they do so.
o SCO wants the offending code to remain so that they can collect big bucks. The GPL does not allow this. Somehow Darl thinks that removing the offending code does not cure the problem. He is right, removing the offending code does not cure his problem, which is a lack of revenue.
I almost hate to write this, as SCO's lawyers might actually use it.
About the only attack on GPL that makes sense is that GPL creates a community which is colluding with each other in an anti-competitive manner artificially driving down prices. GPL code constitutes a "criminal cartel" with the members conspiring with each other to profit from the conspiracy and thus unfairly competing with closed-source products.
You can even argue that the GPL acts to preclude and punish users that don't participate in the community by pushing commercial products out of the marketplace with artificially deflated prices.
Thus, the FSF movement comprises the largest anti-trust entity ever to exist and this must be squashed in order to protect the just interests of world commerce.
The take of the open letter is that "Copyright exists to support profit" and that "GPL precludes using copyright to enforce profit, and even competes with non-GPL making profit for difficult". What is missed is that Copyright exists to enforce property rights. If the author chooses to leverage his copyright with thousands of others in a cooperative work, then this is the free choice of the author. If non-GPL vendors find this disturbing, then this just indicates that competition works.
And the whole take that the copyright laws and constitution are designed to equally protect business and "the people" is bunk. The people "vote", businesses only "lobby". Contrary to business belief, they don't own the people, yet.
RedHat has also released 2.5.7 RPMs for the fix.
/tmp/rsync-2.5.7-0.9.src.rpm
/usr/src/redhat/SPECS
/usr/src/redhat/RPMS/i386, so you can install it with:
/usr/src/redhat/RPMS/i386/rsync-2.5.7-0.9.i386.rpm
When updating an older server (7.1, I think), the RH RPM failed with a GLIBC dependency. The updates for RH are identical for 7.1 - 9, so you might have a problem here.
My easiest workaround was to rebuild the rpm from source with:
Get the rsync-2.5.7-0.9.src.rpm from RedHat ftp server updates.redhat.com
Install the source rpm with:
rpm -ivh
Build a new complete, clean set of RPMs with:
cd
rpm -bb rsync.spec
The new installable binary for your current lib versions is in
rpm -Fvh
---
For those that don't use rsync, this is easily one of the most useful utilities on the box. I particularily like "modules" mode over ssh. Setup an ssh key and have the key auto-run rynnc --daemon. You get modules and ssh. Really cool.