The quality of a datacenter has less to do with the equipment (although thats important), and more to do with who designed and is running the equipment.
Most of the datacenter outages I have been a part of in one way or another (Customer, or Provider) have been caused by:
Poor planning Human Error Poor design
As a normal customer, there is no way to know if any of these problems exist. The solution? Ask for references that utilize that datacenter. Make sure they don't give you a customer that utilizes another data center from the same provider. Data center design varies greatly, even across the same provider. Ask that reference how long they have been there, how many problems they have had, and the companies response to those issues. Look for a customer with a long history in that data center (3+ years, 5 would be better).
Don't rule out a data center because they had an outage. Outages will happen, no matter how redundant their systems are. Their response to it is very important. If you find out about a previous outage, ask to see the root cause analysis they provided their customers. If they can't or won't produce it, even under NDA, then walk away.
I've spend a lot of time looking at this type of problem. I had a customer that wanted to transfer data at greater then 10 mbps across the internet, across the country. Lets just say with windows this is impossible.
The problem has to do with TCP algorithms. I found the ones in windows are optimized for common cases. Linux has multiple TCP/IP algorithms you can choose from. Most are significantly better the one used in windows.
The "problem" with TCP is it has to assume that packet loss equals network congestion. This is a good thing for an over-loaded network link. As the link fills up, it starts dropping packets. As the computers on each end of a TCP connection see this packet loss, they start "Backing off". They slow down their transmission rates until the packet loss is gone. In most cases they back way off, and then slowly increase the speed until they start seeing a little packet loss. The methods they use to determine what is congestion, how much they slow down, and how they recover from it greatly effects total usable bandwidth.
The bottom line: TCP Algorithms greatly effect transfer speed, and no algorithm is good for every situation. Linux gives you flexibility in this area (And by default uses a better one), and windows gives you zero.
To test raw bandwidth, you have to saturate a link with UDP data, and count how much data is received. This is pretty pointless as its not the useable bandwidth, but it does tell you the "raw" potential. The problem is the "raw" potential can be subverted by a small amount of packet loss.
I have worked for 3 hosting companies. My experiance has been: If you are not considered a risk, you are allowed to work your final weeks with full access. If you are REMOTELY considered a risk, you are imediately walked out, although you are paid for your final weeks.
Any good admin/manager knows if you have physical access, you might as well have root/admininistrator access.
Out door rated, $79 a peice. They are powerful, outdoor, directional access points. You can mount one outside of each home and point them at each other. With line of sight, they will easily do 500 meters. They will plug in ethernet on each side of the connection.
You may need to mount them high to get around trees and hills. It all depends. The higher you are, the better off you'll be.
I have personaly used these devices. They are amazing for the price. I'd pay double if I had to. Nothing comes close.
The mac sites reporting from the meeting at apple got it wrong... according to the Xserve site that just went up, there are 3 PCI slots, 1 is used, so you get 2 free! Wow... I'm shocked... Thats very very good...
Its not a RAID server, its a external storage unit, you'll need to plug into a server... using the fibre channel connections...
Its a RAID box, IDE drives, Fibre channel backhaul....
Apple is doing alot right... IDE veruses SCSI - IDE is right for what they're doing (small servers), on the RAID box, I'd go SCSI. I think as they build out thier server lines, they'll build some with SCSI some with IDE...
IDE can be as fast as SCSI, but you can't get 15K RPM IDE drives, you can with SCSI, and SCSI drives are assumed to be run 24x7, IDE isn't... (Although that doesn't mean IDE drives can't last as long, just SCSI drives are designed for more use)...
I had lunch with a Cisco sales rep, and apparently thier demo for this stuff includes several feet of barb wire. They unhook it, and re hook it up, to prove its working.
The demo starts out with like 1000 feet of Cat 3, then cat 1, then lamp cable, then the barb wire, then more cable, reavaling each section as they talk about it to wow you. I haven't seen it, the sales guy just told me about it.
Sounds pretty interesting... it says 10 Mbps at 5000 feet... I assume you get less Mbps the farther you go out... actaully the sales rep was supposed to get me this info, and never did... I'll get on his back about it....
I seriously doubt any ISP is sniffing your packets to figure out where your going... its just not worth it... and nearly imposible on a large scale... To much equipment/money/time... not worth the 2 cents they MIGHT get for the info on where you're going....
Now, the government singling you out, well thats another thing... they could... but if they are sniffing your packets either A) your doing something illegal B) your not doing something illegal, so they won't find anything, other then your obbession with porn sites.
first off, your an idiot, and you don't know what your talking about....
That said, I do agree with your basic point, people whine about 1.5 Mbps should shut up... its plenty fast...
A T1, point to point, in the same metro area, is generally around $300, from the LEC (SWBell, PacBELL, etc). People like sprint, and other guys that buy wholesale from the LEC, or that just buy the copper and do it them selves, can do it for $150 a month (Sprint will charge you $120 for a local loop, around here, with internet access).
Bandwidth at T1 rates will cost you ~$650 from a internet provider.
What do you get from a T1 leased line that you don't get from DSL or cable modems? Guarentteed bandwidth. If you buy a point to point T1 (With no internet access), you get guarenteed bandwidth. Period. (Well unless you get into frame relay, but thats something else). DSL or Cable modems, nothing is guarenteed. Period. And if they say it is, thier lying.
The reason is T1's have been around for ever, and you get 1.544 Mbps or nothing (Well, almost). DSL and cable modems compensate for distance, lowering your bit rate the farther out you get. Plus with cable modems your sharing bandwidth. And with DSL, most likely your going into a DSLAM, with 5,000 other customers, then hopping on a ATM T3. So really, your sharing bandwidth too.
You pay out the nose for a T1 (Compaired to DSL or cable) because with a T1, your guarenteed something. Plus when you buy internet access on top of that T1, they guarentee you'll get 1.544 Mbps all the time, both ways.
DSL/Cable == Consumer solution (For the most part)
T1 == Business/enterprise solution (for the most part)
Anyway... I'm ranting, and I forgot my basic point....
Let me clairify, you always pay for phone support.
I'm actually anti microsoft, pro linux... I just get paid for the microsoft work.... Plus microsoft has its place... it does have its uses...
I don't consider anyone but a live person "support". Newsgroups are helpful, but thats not real support...
But, I rarely contact "Real support". So don't get me wrong, the newsgroups are important, but everything has a newsgroup, so whats the point your trying to make?
The package was bought by our operations group for the IT department... it was a joke all around... We were told about it 6 months into the contract. No one even knew how to call em...
One of the factors is if you're calling them at random, or you have a support aggrement. You ALWAYS pay for support from microsoft. It doesn't come with any product.
THe last place I worked at, we had a microsoft select agreement. Boy is that a deal. (Hahah). We got 150 incedents for $50,000. Sounds crazy, but, it was worth it... To bad we could never use 150 incedents, even if we tried. (150 people in the company, 5 IT people).
The cool thing about the select agreement, is you get a TAM (Technical account manager) that can esclate your call. Plus, he has like 10 customers, so he pays close attention to every case. Its kinda cool when he checks in to see if you were happy with a case.
With a select agreement, you get access to subscriber downloads, which rocks. You can download anything microsoft ever released (Well almost). Wanted to try BOB? go for it. MSDOS 5 in chinese, its there.
Some of thier best support people are in thier exchange support group. The reason being, exchange is a POS that needs alot of attention, and fixing database curruption is a bitch.
A dumbed down SQL engine? It is JET aka Access. Thats not a dumbed down SQL engine. Thats a flat file database. If you get corruption, your in trouble. I should know, I spent 3 days straight (No sleep) on the phone with microsoft support, trying to get it cleaned up. What we had to do is export every mail box to a PST file, and reimport them to a brand new database. We lost all permisions, and all server based rules. Luckly, microsoft has a utility to export each mailbox, for just an occassion.
It scales well only accross many servers. Even then, its a bitch.
The only way to support large installations is to have many dedicated support people...
Sungard doesn't really do collocation. They do disaster recovery services... they are not even close....
You pay sungard a monthly fee, kind of insurance, if you datacenter burns to the ground, or otherwise, you declair a disaster, and go there and rebuild. They supply space, servers, everything you need to rebuild your business, including desks/desktop computers if you want to pay for it...
I talk from experence, I worked for a company that used sungard, and about 6 months ago we went to thier AZ facility to run a Disaster Recovery test.... We ran everything on very large NT servers, and they had to do some specail stuff for us, because they've never had to provide NT servers so big... they mainly do mainframes and large unix boxes....
-Tripp
You've gotten a half baked answer from them, and everyone one this thread:
The real answer is this:
Most likely what is going on, is your being serviced by a "sub-station". Thats not the real name, but I don't know the real one.
Basicly what they do, is put a mini-switch out in the feild in a envormentaly controled valt, either below ground, or above. So you are most likely being serviced by a "sub-station" (not the real name). Its not a fiber/hybrid.
What SBC is doing (Southwestern Bell, Pacbell, Ameritech, etc) is putting NEW "sub-stations" out in the field where people are to far from the CO, or where they are already serviced by a"sub-station" . These are equipment cabinets, in the field that take the copper, and put it into a DSLAM to provide DSL services, and into a switch interface, to provide phone service, and then backhaul it all with fiber.
SBC is spending something like 5 Billion on this project, and so far nothing has come of it, other then I have seen alot of these above ground cabinets pop up.
Anyway, I was quite annoyed that no one gave a straight answer (that was moderated up, I didn't read EVERY message) although I'm probably to late to get moderated up.
The Internet was never truly redundant. If you ever thought it was, you where mistaken..
First off, we're lucky it works at all. The fact that I can get to slashdot every day is supprising.
On the polical front: There is no regulation, there are no rules. Peering is a joke. You can only peer if your one of the top 10 providers. Everyone else is buying from everyone else.
No one has the power to say your packets will always get from point A to point B. If one ISP is mad at the other, it can remove the route through thier network.
On the technical front: Most of the time, your packets will take the same path every time. If that link goes down, normaly, it will reroute (eventualy). But not in real time. And the path it just reroute to, may be sub optimal (IE Your packets take a 30 second round trip over a already overloaded link.)
Another problem is everyone is sharing fiber runs. This saves dollars, but one backhoe can (And has) put a huge black hole on the internet.
Anyway, thats my babble. I haven't looked into this stuff in a while, so my statements may be outdated.
-Tripp
PS I didn't proof read this, so don't insult my bad english.
Have you thougth about using HAM radio and packet modems? I don't know much about the technologies, although I know it will work at ranges up to 50 miles...
It has POTENTAL for hacking. Thats the point... its neat... thats all... If you don't like it, don't read the full story. Get over it. I liked it... I wanted to know about it. If you don't like it, well tuff....
Its the way slashdot has always been. If the guys (CmdrTaco and friends) like it, it gets posted. The only reason the quality has gone down is because they've done stories on most of the cool stuff already. Someone doesn't hack some new hardware every day.
Redhat only lost 1.8 million, on close to 20 million in revenue. They are close to profitablity! They are proving the model.... sure they're not going to over take microsoft any time soon, but they are not going away. So what if they closed a office of 25 people... The story just had quotes from a bunch of laid off, disgruntled employee's....
I think it all depends on what kind of person you are.
I dropped out of high school, and went into computers. But don't get me wrong, I didn't drop out of high school for computers. Why I dropped out is another story.
I went to work for a small computer shop at 17. 9 months later, I was hired by a nation-wide computer training company as a contract Technical Writer. After my 2 month contract, I left, got my GED, and a few certs.
I then got a job as a Sys Admin at a small (250 people) and growing billing software company. I am now a senior systems administrator, making good money, learning alot, and teaching people 10-30 years older then me everything I know.
My point is, it all depends who you are, and what oppertunities come your way. I didn't want to drop out of high school, but I had to. I don't regret it, but I would have liked to go to college. But for me, it was the right thing. I learned more in the 3+ years I've been doing this then I would have learned in 4 years of college. I'm also further ahead in my field.
I think I made it because I'm good at this. I think of myself as a natural. Most people that go to college for computers, didn't know enough about computers to get a job anywhere anyway. The first place I worked, they hired me because college grads didn't know what a hard drive was when they where showed one. I knew what it was, could set SCSI jumpers, and build a whole system in about a half hour.
The reason MySQL was slower was because they used the ODBC drivers. The MySQL ODBC drivers are known to be significately slower then the native drivers. Although I do agree it is probably the best way to test a large number of databases using ODBC drivers, I would like to see the results of all the databases tested using native drivers.
We run RAID on over 80 developemnt servers and 20 production servers. We run NT and MS SQL 7.0, but also do things like bill generation which involves alot of raw file access. Currenty our best setup runs like this (We use HP Intel hardware):
We have 2 raid controllers (each has 3 channels, but you won't need that much for your setup) running a RAID 10 arrray on each. RAID 10 is about the best performace you can get out of RAID. Basicly the idea here is that most RAID controlers are A) Slow B) Can only handle so many I/0's per second, and thats always slower then a modern system can handle.
If you don't know what RAID 10 is, basicly you have 2 or more mirrored drive sets. Then you stripe across those drive sets. This means you A) Need atleast 4 drives B) You lose 1/2 of your useable drive space in the mirroring. But this also means you can do 2 seperate reads across 2 different sets of striped disks which is very speedy (In theory, anyway)
So if you spread your spool and mail across 2 raid controlers running RAID 10, thats probably the best performance your going to get. 10,000 RPM drives will help alot too. The only problem, this is also the most expensive way to do it.... ohh well....
Slashdot Mirror
The quality of a datacenter has less to do with the equipment (although thats important), and more to do with who designed and is running the equipment.
Most of the datacenter outages I have been a part of in one way or another (Customer, or Provider) have been caused by:
Poor planning
Human Error
Poor design
As a normal customer, there is no way to know if any of these problems exist. The solution? Ask for references that utilize that datacenter. Make sure they don't give you a customer that utilizes another data center from the same provider. Data center design varies greatly, even across the same provider. Ask that reference how long they have been there, how many problems they have had, and the companies response to those issues. Look for a customer with a long history in that data center (3+ years, 5 would be better).
Don't rule out a data center because they had an outage. Outages will happen, no matter how redundant their systems are. Their response to it is very important. If you find out about a previous outage, ask to see the root cause analysis they provided their customers. If they can't or won't produce it, even under NDA, then walk away.
I've spend a lot of time looking at this type of problem. I had a customer that wanted to transfer data at greater then 10 mbps across the internet, across the country. Lets just say with windows this is impossible.
The problem has to do with TCP algorithms. I found the ones in windows are optimized for common cases. Linux has multiple TCP/IP algorithms you can choose from. Most are significantly better the one used in windows.
The "problem" with TCP is it has to assume that packet loss equals network congestion. This is a good thing for an over-loaded network link. As the link fills up, it starts dropping packets. As the computers on each end of a TCP connection see this packet loss, they start "Backing off". They slow down their transmission rates until the packet loss is gone. In most cases they back way off, and then slowly increase the speed until they start seeing a little packet loss. The methods they use to determine what is congestion, how much they slow down, and how they recover from it greatly effects total usable bandwidth.
The bottom line: TCP Algorithms greatly effect transfer speed, and no algorithm is good for every situation. Linux gives you flexibility in this area (And by default uses a better one), and windows gives you zero.
To test raw bandwidth, you have to saturate a link with UDP data, and count how much data is received. This is pretty pointless as its not the useable bandwidth, but it does tell you the "raw" potential. The problem is the "raw" potential can be subverted by a small amount of packet loss.
I have worked for 3 hosting companies. My experiance has been: If you are not considered a risk, you are allowed to work your final weeks with full access. If you are REMOTELY considered a risk, you are imediately walked out, although you are paid for your final weeks.
Any good admin/manager knows if you have physical access, you might as well have root/admininistrator access.
You want two of these:
http://ubnt.com/products/ns2.php
Out door rated, $79 a peice. They are powerful, outdoor, directional access points. You can mount one outside of each home and point them at each other. With line of sight, they will easily do 500 meters. They will plug in ethernet on each side of the connection.
You may need to mount them high to get around trees and hills. It all depends. The higher you are, the better off you'll be.
I have personaly used these devices. They are amazing for the price. I'd pay double if I had to. Nothing comes close.
-Tripp
The mac sites reporting from the meeting at apple got it wrong... according to the Xserve site that just went up, there are 3 PCI slots, 1 is used, so you get 2 free! Wow... I'm shocked... Thats very very good...
-Tripp
Its not a RAID server, its a external storage unit, you'll need to plug into a server... using the fibre channel connections...
Its a RAID box, IDE drives, Fibre channel backhaul....
Apple is doing alot right... IDE veruses SCSI - IDE is right for what they're doing (small servers), on the RAID box, I'd go SCSI. I think as they build out thier server lines, they'll build some with SCSI some with IDE...
IDE can be as fast as SCSI, but you can't get 15K RPM IDE drives, you can with SCSI, and SCSI drives are assumed to be run 24x7, IDE isn't... (Although that doesn't mean IDE drives can't last as long, just SCSI drives are designed for more use)...
Actually, Cogent just bought PSI for 10 million.
Thats a deal in anyones book....
-Tripp
"Fashionable programming languages don't equal useful software" - Eve Andersson
Slightly off topic..... but not at the same time....
-Tripp
If you read the article all the way, it says $1299... I don't know where they got the 1800 (CDN, maybe, i dunno... poorly written in my opinion).
-Tripp
Funny you should mention this.
I had lunch with a Cisco sales rep, and apparently thier demo for this stuff includes several feet of barb wire. They unhook it, and re hook it up, to prove its working.
The demo starts out with like 1000 feet of Cat 3, then cat 1, then lamp cable, then the barb wire, then more cable, reavaling each section as they talk about it to wow you. I haven't seen it, the sales guy just told me about it.
Sounds pretty interesting... it says 10 Mbps at 5000 feet... I assume you get less Mbps the farther you go out... actaully the sales rep was supposed to get me this info, and never did... I'll get on his back about it....
-Tripp
Youve got to be kidding me....
I seriously doubt any ISP is sniffing your packets to figure out where your going... its just not worth it... and nearly imposible on a large scale... To much equipment/money/time... not worth the 2 cents they MIGHT get for the info on where you're going....
Now, the government singling you out, well thats another thing... they could... but if they are sniffing your packets either A) your doing something illegal B) your not doing something illegal, so they won't find anything, other then your obbession with porn sites.
-Tripp
first off, your an idiot, and you don't know what your talking about....
That said, I do agree with your basic point, people whine about 1.5 Mbps should shut up... its plenty fast...
A T1, point to point, in the same metro area, is generally around $300, from the LEC (SWBell, PacBELL, etc). People like sprint, and other guys that buy wholesale from the LEC, or that just buy the copper and do it them selves, can do it for $150 a month (Sprint will charge you $120 for a local loop, around here, with internet access).
Bandwidth at T1 rates will cost you ~$650 from a internet provider.
What do you get from a T1 leased line that you don't get from DSL or cable modems? Guarentteed bandwidth. If you buy a point to point T1 (With no internet access), you get guarenteed bandwidth. Period. (Well unless you get into frame relay, but thats something else). DSL or Cable modems, nothing is guarenteed. Period. And if they say it is, thier lying.
The reason is T1's have been around for ever, and you get 1.544 Mbps or nothing (Well, almost). DSL and cable modems compensate for distance, lowering your bit rate the farther out you get. Plus with cable modems your sharing bandwidth. And with DSL, most likely your going into a DSLAM, with 5,000 other customers, then hopping on a ATM T3. So really, your sharing bandwidth too.
You pay out the nose for a T1 (Compaired to DSL or cable) because with a T1, your guarenteed something. Plus when you buy internet access on top of that T1, they guarentee you'll get 1.544 Mbps all the time, both ways.
DSL/Cable == Consumer solution (For the most part)
T1 == Business/enterprise solution (for the most part)
Anyway... I'm ranting, and I forgot my basic point....
-Tripp
Let me clairify, you always pay for phone support.
I'm actually anti microsoft, pro linux... I just get paid for the microsoft work.... Plus microsoft has its place... it does have its uses...
I don't consider anyone but a live person "support". Newsgroups are helpful, but thats not real support...
But, I rarely contact "Real support". So don't get me wrong, the newsgroups are important, but everything has a newsgroup, so whats the point your trying to make?
-Tripp
Hahaha dude
The package was bought by our operations group for the IT department... it was a joke all around... We were told about it 6 months into the contract. No one even knew how to call em...
-Tripp
Microsoft support can be good. It all depends...
One of the factors is if you're calling them at random, or you have a support aggrement. You ALWAYS pay for support from microsoft. It doesn't come with any product.
THe last place I worked at, we had a microsoft select agreement. Boy is that a deal. (Hahah). We got 150 incedents for $50,000. Sounds crazy, but, it was worth it... To bad we could never use 150 incedents, even if we tried. (150 people in the company, 5 IT people).
The cool thing about the select agreement, is you get a TAM (Technical account manager) that can esclate your call. Plus, he has like 10 customers, so he pays close attention to every case. Its kinda cool when he checks in to see if you were happy with a case.
With a select agreement, you get access to subscriber downloads, which rocks. You can download anything microsoft ever released (Well almost). Wanted to try BOB? go for it. MSDOS 5 in chinese, its there.
Some of thier best support people are in thier exchange support group. The reason being, exchange is a POS that needs alot of attention, and fixing database curruption is a bitch.
-Tripp
You're mentally ill.
A dumbed down SQL engine? It is JET aka Access. Thats not a dumbed down SQL engine. Thats a flat file database. If you get corruption, your in trouble. I should know, I spent 3 days straight (No sleep) on the phone with microsoft support, trying to get it cleaned up. What we had to do is export every mail box to a PST file, and reimport them to a brand new database. We lost all permisions, and all server based rules. Luckly, microsoft has a utility to export each mailbox, for just an occassion.
It scales well only accross many servers. Even then, its a bitch.
The only way to support large installations is to have many dedicated support people...
-Tripp
FYI
Sungard doesn't really do collocation. They do disaster recovery services... they are not even close....
You pay sungard a monthly fee, kind of insurance, if you datacenter burns to the ground, or otherwise, you declair a disaster, and go there and rebuild. They supply space, servers, everything you need to rebuild your business, including desks/desktop computers if you want to pay for it...
I talk from experence, I worked for a company that used sungard, and about 6 months ago we went to thier AZ facility to run a Disaster Recovery test.... We ran everything on very large NT servers, and they had to do some specail stuff for us, because they've never had to provide NT servers so big... they mainly do mainframes and large unix boxes....
-Tripp
You've gotten a half baked answer from them, and everyone one this thread:
The real answer is this:
Most likely what is going on, is your being serviced by a "sub-station". Thats not the real name, but I don't know the real one.
Basicly what they do, is put a mini-switch out in the feild in a envormentaly controled valt, either below ground, or above. So you are most likely being serviced by a "sub-station" (not the real name). Its not a fiber/hybrid.
What SBC is doing (Southwestern Bell, Pacbell, Ameritech, etc) is putting NEW "sub-stations" out in the field where people are to far from the CO, or where they are already serviced by a"sub-station" . These are equipment cabinets, in the field that take the copper, and put it into a DSLAM to provide DSL services, and into a switch interface, to provide phone service, and then backhaul it all with fiber.
SBC is spending something like 5 Billion on this project, and so far nothing has come of it, other then I have seen alot of these above ground cabinets pop up.
Anyway, I was quite annoyed that no one gave a straight answer (that was moderated up, I didn't read EVERY message) although I'm probably to late to get moderated up.
-Tripp
The Internet was never truly redundant. If you ever thought it was, you where mistaken..
First off, we're lucky it works at all. The fact that I can get to slashdot every day is supprising.
On the polical front: There is no regulation, there are no rules. Peering is a joke. You can only peer if your one of the top 10 providers. Everyone else is buying from everyone else.
No one has the power to say your packets will always get from point A to point B. If one ISP is mad at the other, it can remove the route through thier network.
On the technical front: Most of the time, your packets will take the same path every time. If that link goes down, normaly, it will reroute (eventualy). But not in real time. And the path it just reroute to, may be sub optimal (IE Your packets take a 30 second round trip over a already overloaded link.)
Another problem is everyone is sharing fiber runs. This saves dollars, but one backhoe can (And has) put a huge black hole on the internet.
Anyway, thats my babble. I haven't looked into this stuff in a while, so my statements may be outdated.
-Tripp
PS I didn't proof read this, so don't insult my bad english.
Have you thougth about using HAM radio and packet modems? I don't know much about the technologies, although I know it will work at ranges up to 50 miles...
-Tripp
It has POTENTAL for hacking. Thats the point... its neat... thats all... If you don't like it, don't read the full story. Get over it. I liked it... I wanted to know about it. If you don't like it, well tuff....
Its the way slashdot has always been. If the guys (CmdrTaco and friends) like it, it gets posted. The only reason the quality has gone down is because they've done stories on most of the cool stuff already. Someone doesn't hack some new hardware every day.
-Tripp
Some of you are missing the point....
Redhat only lost 1.8 million, on close to 20 million in revenue. They are close to profitablity! They are proving the model.... sure they're not going to over take microsoft any time soon, but they are not going away. So what if they closed a office of 25 people... The story just had quotes from a bunch of laid off, disgruntled employee's....
-Tripp
I think it all depends on what kind of person you are.
I dropped out of high school, and went into computers. But don't get me wrong, I didn't drop out of high school for computers. Why I dropped out is another story.
I went to work for a small computer shop at 17. 9 months later, I was hired by a nation-wide computer training company as a contract Technical Writer. After my 2 month contract, I left, got my GED, and a few certs.
I then got a job as a Sys Admin at a small (250 people) and growing billing software company. I am now a senior systems administrator, making good money, learning alot, and teaching people 10-30 years older then me everything I know.
My point is, it all depends who you are, and what oppertunities come your way. I didn't want to drop out of high school, but I had to. I don't regret it, but I would have liked to go to college. But for me, it was the right thing. I learned more in the 3+ years I've been doing this then I would have learned in 4 years of college. I'm also further ahead in my field.
I think I made it because I'm good at this. I think of myself as a natural. Most people that go to college for computers, didn't know enough about computers to get a job anywhere anyway. The first place I worked, they hired me because college grads didn't know what a hard drive was when they where showed one. I knew what it was, could set SCSI jumpers, and build a whole system in about a half hour.
Maybe I'm wrong, but I'm not doing to bad.
-Tripp
The reason MySQL was slower was because they used the ODBC drivers. The MySQL ODBC drivers are known to be significately slower then the native drivers. Although I do agree it is probably the best way to test a large number of databases using ODBC drivers, I would like to see the results of all the databases tested using native drivers.
-Tripp
We run RAID on over 80 developemnt servers and 20 production servers. We run NT and MS SQL 7.0, but also do things like bill generation which involves alot of raw file access. Currenty our best setup runs like this (We use HP Intel hardware):
We have 2 raid controllers (each has 3 channels, but you won't need that much for your setup) running a RAID 10 arrray on each. RAID 10 is about the best performace you can get out of RAID. Basicly the idea here is that most RAID controlers are A) Slow B) Can only handle so many I/0's per second, and thats always slower then a modern system can handle.
If you don't know what RAID 10 is, basicly you have 2 or more mirrored drive sets. Then you stripe across those drive sets. This means you A) Need atleast 4 drives B) You lose 1/2 of your useable drive space in the mirroring. But this also means you can do 2 seperate reads across 2 different sets of striped disks which is very speedy (In theory, anyway)
So if you spread your spool and mail across 2 raid controlers running RAID 10, thats probably the best performance your going to get. 10,000 RPM drives will help alot too. The only problem, this is also the most expensive way to do it.... ohh well....
-Tripp