PDA

View Full Version : SKA news



AtW
7th September 2011, 13:28
Considering buying replacement switch, looking at £5k minimum cost :eek:

PAH
7th September 2011, 13:54
Do what Google does and use consumer grade hardware. A few cheap switches daisy-chained on baking trays will do.

BrilloPad
7th September 2011, 13:56
Considering buying replacement switch, looking at £5k minimum cost :eek:

But with your earnings that should be no problem?

AtW
7th September 2011, 14:01
The first switches for SKA were just £50. :frown

BrilloPad
7th September 2011, 14:02
The first switches for SKA were just £50. :frown

That is the result of QE on inflation for you.

eek
7th September 2011, 14:03
Do what Google does and use consumer grade hardware. A few cheap switches daisy-chained on baking trays will do.

Thats fine for google as it doesn't really matter if a few machines go offline, the rest can pick up the slack until the switch is replaced. I don't think SKA is at that level of hardware redundancy yet.

darmstadt
7th September 2011, 14:03
Buster Bloodvessel was 53 yesterday :ohwell

AtW
7th September 2011, 14:06
Thats fine for google as it doesn't really matter if a few datacenters go offline, the rest can pick up the slack until the switch is replaced. I don't think SKA is at that level of hardware redundancy yet.

FTFY

d000hg
7th September 2011, 14:10
Sheesh AtW, get with the cloud.

PAH
7th September 2011, 14:11
Move it into the cloud so you're no longer responsible for any hardware.

d000hg
7th September 2011, 14:12
Move it into the cloud so you're no longer responsible for any hardware.I like your thinking. Where do you get your ideas?

AlfredJPruffock
7th September 2011, 14:13
Buster Bloodvessel was 53 yesterday :ohwell

Walking on Sunshine was one of my faovs ... good old Bbuster

eek
7th September 2011, 14:20
Move it into the cloud so you're no longer responsible for any hardware.

If only that was the case. Amazon is painful as you end up managing virtual servers all the time, Azure is not quite as bad but the SQL server costs can be scary.

wobbegong
7th September 2011, 14:32
Considering buying replacement switch, looking at £5k minimum cost :eek:

£5K?? They saw you coming. (http://www.diy.com/diy/jsp/bq/nav.jsp?action=detail&fh_secondid=9289663&fh_view_size=10&fh_location=%2f%2fcatalog01%2fen_GB%2fcategories%3 C{9372015}%2fcategories%3C{9372042}%2fcategories%3 C{9792017}%2fcategories%3C{9792034}%2fspecificatio nsProductType%3dmk_white_range&fh_eds=%C3%9F&isSearch=false)

:eyes

PAH
7th September 2011, 14:47
I like your thinking. Where do you get your ideas?

Same place as everyone else, the Borg. :eyes

PAH
7th September 2011, 14:51
If only that was the case. Amazon is painful as you end up managing virtual servers all the time, Azure is not quite as bad but the SQL server costs can be scary.


Seeing as (IIRC) SKA was/is built upon p2p clients trawling the web on his behalf, maybe AtW could roll out an add-on for p2p cloud services rather than faffing about with centralized servers.

[As always, my ideas come with a 10% charge for usage. Cheaper than getting a consultant in. :D]

Paddy
7th September 2011, 14:53
£5K?? They saw you coming. (http://www.diy.com/diy/jsp/bq/nav.jsp?action=detail&fh_secondid=9289663&fh_view_size=10&fh_location=%2f%2fcatalog01%2fen_GB%2fcategories%3 C{9372015}%2fcategories%3C{9372042}%2fcategories%3 C{9792017}%2fcategories%3C{9792034}%2fspecificatio nsProductType%3dmk_white_range&fh_eds=%C3%9F&isSearch=false)

:eyes

Don’t be silly, you need at least a double pole double throw to do proper switching.

Jog On
7th September 2011, 15:14
ON the plus side that £5k is tax deductable and might cancel out your corporation tax liability for the year :smokin

AtW
7th September 2011, 15:15
ON the plus side that £5k is tax deductable and might cancel out your corporation tax liability for the year :smokin

Only if you do not exceed your Annual Investment Allowance which goes down next year to pathetic £25k :eyes

Paddy
7th September 2011, 15:23
Only if you do not exceed your Annual Investment Allowance which goes down next year to pathetic £25k :eyes

Your offshore company in Panama owns the router. They rent it to your UK company for £20k per year.

As it is rented you get the full taxable allowance of £20k

You pay flat rate tax of £200 to your Offshore Company.

"Expenses" for running the company are made through the offshore ATM card.

AtW
7th September 2011, 15:29
Your offshore company in Panama owns the router. They rent it to your UK company for £20k per year.

As it is rented you get the full taxable allowance of £20k

You pay flat rate tax of £200 to your Offshore Company.

"Expenses" for running the company are made through the offshore ATM card.

:eek

eek
7th September 2011, 15:34
Only if you do not exceed your Annual Investment Allowance which goes down next year to pathetic £25k :eyes

And there you highlight an actual reason for moving to the cloud. You end up with a company whose only asset is some intellectual property providing a charged for service using cheap server space in Dublin or the USA.

The question you're left asking is where in the world do I get paypal to send the subscription payments to.

AtW
7th September 2011, 15:40
Do you think many clouds offer 10 GBE level networking (between servers, not switches) with appropriately specced boxes connected to it at a cost that is not crazy?

The tax system is certainly completely ****ed when companies get charged corp tax on buying essential machinery for business.

doodab
7th September 2011, 15:51
The tax system is certainly completely ****ed when companies get charged corp tax on buying essential machinery for business.

Yes but don't worry because the government will support entrepreneurs by abolishing the 50% income tax band.

It clearly hasn't occurred to them that sorting out the ******* mess of bureaucracy and making it easier to become a 50% tax payer in the first place would do more to encourage entrepreneurs.

AtW
7th September 2011, 15:54
Yes but don't worry because the government will support entrepreneurs by abolishing the 50% income tax band.

Entrepreneurs don't get reward in salary of £150k usually, it's CGT that's more important and as you say lack of general red tape to actually make business successful, to be fair I would not say it's too bad in this country, high taxes certainly far more important to deal with.

PAH
7th September 2011, 16:03
Do you think many clouds offer 10 GBE level networking (between servers, not switches) with appropriately specced boxes connected to it at a cost that is not crazy?


Welcome to the Performance, Reliability, Cost paradigm where you can choose only two.

AtW
7th September 2011, 16:10
Welcome to the Performance, Reliability, Cost paradigm where you can choose only two.

B0ll0x.

You can choose High Performance, High Reliability and High Cost at the same time.



:rolleyes:

MarillionFan
7th September 2011, 16:40
Considering buying replacement switch, looking at £5k minimum cost :eek:

:eek:

WTF!!!!!

http://www.basilrathbone.net/films/davidcopperfield/dc25a.jpg

d000hg
7th September 2011, 16:43
If only that was the case. Amazon is painful as you end up managing virtual servers all the time.I don't know that's the case is it, with the APIs and whatnot?

Anyway, if Apple can run their system on the cloud I reckon SKA is a candidate too: Apple's iCloud runs on Microsoft and Amazon services ? The Register (http://www.theregister.co.uk/2011/09/02/icloud_runs_on_microsoft_azure_and_amazon/)

Have Google got on the cloud bandwagon yet - I know they have cloudy apps but not the same thing? Seems inevitable and this is an area they would actually know something about after all.

d000hg
7th September 2011, 16:44
B0ll0x.

You can choose High Performance, High Reliability and High Cost at the same time.

Well done AtW. You made your first joke.

DodgyAgent
7th September 2011, 17:17
Entrepreneurs don't get reward in salary of £150k usually, it's CGT that's more important and as you say lack of general red tape to actually make business successful, to be fair I would not say it's too bad in this country, high taxes certainly far more important to deal with.

Yes but you don't have a proper business. you don't manufacture anything just write a few programs

AtW
7th September 2011, 18:09
Yes but you don't have a proper business. you don't manufacture anything just write a few programs

I can give you 2 references that would prove otherwise :eyes

MarillionFan
7th September 2011, 18:10
I can give you 2 references that would prove otherwise :eyes

Family don't count.:laugh

darmstadt
7th September 2011, 18:17
Who needs loads of Intel boxes when you can do this (http://www.sinenomine.net/sites/www.sinenomine.net/files/s9202sca.pdf)on only one IBM (back in 2002 this was - Successfully deployed 97,943 Linux images on a dedicated 9672-ZZ7 with VM. Imagine what you you could do with a z196!!!?

conned tractor
7th September 2011, 18:42
Here you go, a 3K switch (http://www.ebay.co.uk/itm/HP-8-port-10-GbE-SFP-v2-zl-Module-Switch-SFP-M-270902-/170690369390?pt=UK_Computing_Networking_SM&hash=item27bdf0576e), saving you 2K. I'll send the invoice down the business park.

AtW
7th September 2011, 18:47
Here you go, a 3K switch (http://www.ebay.co.uk/itm/HP-8-port-10-GbE-SFP-v2-zl-Module-Switch-SFP-M-270902-/170690369390?pt=UK_Computing_Networking_SM&hash=item27bdf0576e), saving you 2K. I'll send the invoice down the business park.

It's only 8 ports, probably not even stackable so keep it.

conned tractor
7th September 2011, 18:49
It's only 8 ports, probably not even stackable so keep it.

This one? (http://www.ebay.co.uk/itm/Woven-Systems-EFX-1000-10GBE-Switch-x48-Active-Ports-/390176505702?pt=COMP_EN_Hubs&hash=item5ad854ff66)

d000hg
7th September 2011, 18:54
Yes but you don't have a proper business. you don't manufacture anything just write a few programsThat's brilliant coming from someone who's non-business is helping non-businesses to find staff.

AtW
7th September 2011, 18:54
This one? (http://www.ebay.co.uk/itm/Woven-Systems-EFX-1000-10GBE-Switch-x48-Active-Ports-/390176505702?pt=COMP_EN_Hubs&hash=item5ad854ff66)

That's too much, this one is about right -

Netgear ProSafe® 24-port 10 Gigabit Stackable L2.. | Ebuyer.com (http://www.ebuyer.com/247177-netgear-prosafe-24-port-10-gigabit-stackable-l2-managed-switch-xsm7224s-100eus)

conned tractor
7th September 2011, 19:01
That's too much, this one is about right -

Netgear ProSafe® 24-port 10 Gigabit Stackable L2.. | Ebuyer.com (http://www.ebuyer.com/247177-netgear-prosafe-24-port-10-gigabit-stackable-l2-managed-switch-xsm7224s-100eus)

Better value? (http://www.google.co.uk/products/catalog?q=dgs3627&hl=en&um=1&ie=UTF-8&tbm=shop&cid=8088387490118181618&sa=X&ei=G79nTvqPDMSWOqG1yOIL&ved=0CFkQ8wIwAA#)

I'm going to give up now because I ain't really got a clue what I'm looking at. ;)

AtW
7th September 2011, 19:03
Better value? (http://www.google.co.uk/products/catalog?q=dgs3627&hl=en&um=1&ie=UTF-8&tbm=shop&cid=8088387490118181618&sa=X&ei=G79nTvqPDMSWOqG1yOIL&ved=0CFkQ8wIwAA#)

I'm going to give up now because I ain't really got a clue what I'm looking at. ;)

No, that's one of those con "10 GBE" switches that only have a handful of 10 GBE ports to get fast uplink but most of ports on it are 1 Gbit.

Good try anyway.

PAH
8th September 2011, 08:30
Use all that spare cpu capacity to compress the data so it doesn't need 10Gbit links to ship it around the network.

AtW
8th September 2011, 09:56
Use all that spare cpu capacity to compress the data so it doesn't need 10Gbit links to ship it around the network.

What spare capacity? :eyes

PAH
8th September 2011, 10:45
What spare capacity? :eyes


Look for the Turbo button (http://en.wikipedia.org/wiki/Turbo_button). :glasses

doodab
8th September 2011, 11:10
Considering buying replacement switch, looking at £5k minimum cost :eek:

What is wrong with the old one?

AtW
8th September 2011, 11:44
What is wrong with the old one?

Too many cables.

eek
8th September 2011, 12:37
Too many cables.

It's a switch that you connect cables to. How many surplus cables can there be?

Boy I'm glad I moved plan b to the cloud and don't have these hassles any more.

AtW
8th September 2011, 13:09
It's a switch that you connect cables to. How many surplus cables can there be?

We use 48 port switches, that's a lot of cables and a lot of things can go wrong :eyes

doodab
8th September 2011, 13:15
We use 48 port switches, that's a lot of cables and a lot of things can go wrong :eyes

How will a new switch use less cables?

eek
8th September 2011, 13:17
How will a new switch use less cables?

+1. (or more accurately, +however many people have read this thread).

PAH
8th September 2011, 13:27
+1. (or more accurately, +however many people have read this thread).

-1

I assume AtW meant that he has too many cables to fit a 48 port switch, so needs another one. :rolleyes:

AtW
8th September 2011, 13:30
How will a new switch use less cables?

One 10 GBE port with one cable means up to 10 x 1 GBE cables can be removed from equation.

doodab
8th September 2011, 13:33
One 10 GBE port with one cable means up to 10 x 1 GBE cables can be removed from equation.

Aah, you are using teamed gigabit connections at the moment. Don't forget to factor in the cost of 10G adapters for each server as well. That will double the cost.

AtW
8th September 2011, 13:37
Don't forget to factor in the cost of 10G adapters for each server as well. That will double the cost.

10 GBE cards with 2 ports now cost about as much as good quad 1 GBE card, so this is ok.

Full 10 GBE switches on the other hand :mad

minestrone
8th September 2011, 13:39
On the subject of the amazon cloud & performace I moved to one of my database instances to serve a web server and fook me they must have pidgeons moving the data. Not an option and production level.

AtW
8th September 2011, 13:41
Amazon storage has got high latency and can return "busy" error codes apparently which you are meant to retry :puke:

minestrone
8th September 2011, 13:50
Amazon storage has got high latency and can return "busy" error codes apparently which you are meant to retry :puke:

I can except a bit of latency but this is beyond a joke.

http://img20.imageshack.us/img20/4394/chartdo.png (http://imageshack.us/photo/my-images/20/chartdo.png/)

Feck knows what they are doing on their servers and will not be buying from them again.

I take on board some of the pages are a bit crap on the performance stakes but I should not be getting that moving to an off machine mySQL server.

eek
8th September 2011, 13:53
I can except a bit of latency but this is beyond a joke.

http://img20.imageshack.us/img20/4394/chartdo.png (http://imageshack.us/photo/my-images/20/chartdo.png/)

Feck knows what they are doing on their servers and will not be buying from them again.

I take on board some of the pages are a bit crap on the performance stakes but I should not be getting that moving to an off machine mySQL server.

What instance type were you using for the mySQL database?

Its worrying because the plan b backup plan is AWS and I couldn't cope with that type of delay.

minestrone
8th September 2011, 14:03
What instance type were you using for the mySQL database?

Its worrying because the plan b backup plan is AWS and I couldn't cope with that type of delay.

Had a medium sharing the tomcat webserver and mySQL for plan B, moved to 2 small ubuntu webservers for plan B & C and picked up a small shared mySQL instance for both, I know I can get the code to run quicker but it was hitting around the .1 second for page loads which I was OK with, now you can see it is up and down like a hookers knickers.

PAH
8th September 2011, 14:40
You need a private cloud, which 'they' will hire you once you realise their main cloud is no better than normal shared hosting. They keep it slow for that reason, to upsell you. :rolleyes:

AtW
8th September 2011, 15:01
Its worrying because the plan b backup plan is AWS and I couldn't cope with that type of delay.

Just a few posts earlier you were very relived that you use cloud :eyes

Amazon had massive downtime with loss of data spanning over few days, our main competitor was hit hard though they were the lucky ones and were only down for just over 24 hours.

They ain't cheap either - if you have money and factor in costs over 3 years then you are better off leasing/buying yourself, in my calculations Amazon would cost the same over 1 year, so 3 times better ROI easily, more if you buy quality stuff that can last 4-5 years.

eek
8th September 2011, 15:11
Just a few posts earlier you were very relived that you use cloud :eyes

Amazon had massive downtime with loss of data spanning over few days, our main competitor was hit hard though they were the lucky ones and were only down for just over 24 hours.

They ain't cheap either - if you have money and factor in costs over 3 years then you are better off leasing/buying yourself, in my calculations Amazon would cost the same over 1 year, so 3 times better ROI easily, more if you buy quality stuff that can last 4-5 years.


Amazon is the backup plan. Gogrid with actual database servers or Rackspace would be my initial points of call. Azure is only winning as its allowing me to beta test with real life clients for free.

AtW
8th September 2011, 15:48
Amazon is the backup plan. Gogrid with actual database servers or Rackspace would be my initial points of call. Azure is only winning as its allowing me to beta test with real life clients for free.

Rackspace is very expensive.

Ask Milan who he works for, they must be very cheap :grey

eek
8th September 2011, 15:52
Rackspace is very expensive.

Ask Milan who he works for, they must be very cheap :grey

They employ Milan. Need a give you a second reason for looking elsewhere.

minestrone
8th September 2011, 17:08
Just a few posts earlier you were very relived that you use cloud :eyes

Amazon had massive downtime with loss of data spanning over few days, our main competitor was hit hard though they were the lucky ones and were only down for just over 24 hours.

They ain't cheap either - if you have money and factor in costs over 3 years then you are better off leasing/buying yourself, in my calculations Amazon would cost the same over 1 year, so 3 times better ROI easily, more if you buy quality stuff that can last 4-5 years.

True, they are not cheap but they offer a scalable solution which is appealing to smaller start ups on cheaper solutions who want to be able to move plans on an hourly basis. For someone like yourself it is not an option.

They have had some downtime but if you enter into the spirit of the thing the advantage of cloud is having instances running in all their datacenters.

Also you have to buy a seperate mailserver as well which is a pain because nobody trusts mails from their servers.

AtW
8th September 2011, 18:28
they offer a scalable solution

They offer excellent vendor lock in for sure - what a lot of start ups need to do is write efficient code or hire people who can do that, this way they won't be using many resources in the first place: we handled literally millions of queries per day with a handful of cheap boxes, lots of very cheap dedicated servers out there that will be managed by the company that rents them out.

For anything web related its crazy to use Amazon, the only possible valid case I can see in them is being able to run large scale CPU only jobs very rarely, ie: one day a month to run really big job, that's fine, but if you need to run big jobs every day of the month then it's crazy to use Amazon.

They are not fast (latency wise) either for web hosting. Why on earth use them when you can get VPS for £25 per month and it can handle million requests per day if you are good with coding?

MarillionFan
8th September 2011, 18:31
They employ Milan. Need a give you a second reason for looking elsewhere.

What Rackspace???

Wow. That's a lucky escape. I got offered a senior perm job at Rackspace last year. The old tape changer could have been working for me! :eek:

DimPrawn
8th September 2011, 19:15
So this is a thread about what exactly?

Someone who is going to be homeless by Christmas.....and they are worried about a router or sommat.

Cliphead
8th September 2011, 19:17
So this is a thread about what exactly?

Someone who is going to be homeless by Christmas.....and they are worried about a router or sommat.

Everybody needs a router right?

AtW
8th September 2011, 19:19
Someone who is going to be homeless by Christmas.....and they are worried about a router or sommat.

Representative of landlord told me not to worry about signing new contract because it will turn into "continuous tenancy" that is nice and easy and I can give just 1 month notice to them and they can give 2 months notice to me ...

MarillionFan
8th September 2011, 19:24
Representative of landlord told me not to worry about signing new contract because it will turn into "continuous tenancy" that is nice and easy and I can give just 1 month notice to them and they can give 2 months notice to me ...

Oh, just go and buy a place FFS! You know you got it wrong. Now is a good time for you to buy. Now stop being a pratt and go and buy yourself a property. :tantrum:

(I've had a bad day btw and am now drinking - expect some serious abuse in about one hour) :mad:

AtW
8th September 2011, 19:24
This time next year...

eek
8th September 2011, 19:37
This time next year...

I'll be able to pay for some cardboard and refurbish the doorway.

Cliphead
8th September 2011, 19:48
(I've had a bad day btw and am now drinking - expect some serious abuse in about one hour) :mad:

I'm drinking too but I don't get abusive, dickhead.

MarillionFan
8th September 2011, 19:50
I'm drinking too but I don't get abusive, dickhead.

Well I do. Now feck off before I say feck off! :mad

Cliphead
8th September 2011, 19:53
Well I do. Now feck off before I say feck off! :mad

Oops, sense of humour failure....

MarillionFan
8th September 2011, 19:55
Oops, sense of humour failure....

Glad you admit it.

Cliphead
8th September 2011, 19:56
Glad you admit it.

You're definitley drinking.

doodab
8th September 2011, 19:58
I'm drinking.

And silently fuming at the price of beer in the UK. £3 for two 600ml bottles. WTF? I thought the UK was supposed to have a problem with cheap booze.

MarillionFan
8th September 2011, 19:59
I'm drinking.

And silently fuming at the price of beer in the UK. £3 for two 600ml bottles. WTF? I thought the UK was supposed to have a problem with cheap booze.

Exactly! EXACTLY!!!!! :mad

Do you know it's now £3.80 for a pint of stella at my local 'country' pub. How the feck! You can buy a bottle of wine for the same price!!!!:throw:

Cliphead
8th September 2011, 20:03
I'm drinking.

And silently fuming at the price of beer in the UK. £3 for two 600ml bottles. WTF? I thought the UK was supposed to have a problem with cheap booze.


Exactly! EXACTLY!!!!! :mad

Do you know it's now £3.80 for a pint of stella at my local 'country' pub. How the feck! You can buy a bottle of wine for the same price!!!!:throw:

Contractors complaining about the price of booze? I have no idea what it costs per pint, bottle or whatever.

MarillionFan
8th September 2011, 20:05
Contractors complaining about the price of booze? I have no idea what it costs per pint, bottle or whatever.

What's the current price for a gram of coke then? :tongue

doodab
8th September 2011, 20:05
Contractors complaining about the price of booze? I have no idea what it costs per pint, bottle or whatever.

It's €0.85 for a 500ml bottle of Bavaria's finest in Munich. Over £2 for a 660ml bottle of Peroni in my local tesco. That is not good value.

I do recommend Waitrose own brand Bavarian weissbier though, it's the real deal.

Cliphead
8th September 2011, 20:08
What's the current price for a gram of coke then? :tongue

I thought it came in litres or fractions of :confused:

Cliphead
8th September 2011, 20:10
It's €0.85 for a 500ml bottle of Bavaria's finest in Munich. Over £2 for a 660ml bottle of Peroni in my local tesco. That is not good value.

I do recommend Waitrose own brand Bavarian weissbier though, it's the real deal.

Good value isn't a consideration when purchasing booze in the UK.

minestrone
8th September 2011, 20:10
They offer excellent vendor lock in for sure - what a lot of start ups need to do is write efficient code or hire people who can do that, this way they won't be using many resources in the first place: we handled literally millions of queries per day with a handful of cheap boxes, lots of very cheap dedicated servers out there that will be managed by the company that rents them out.

For anything web related its crazy to use Amazon, the only possible valid case I can see in them is being able to run large scale CPU only jobs very rarely, ie: one day a month to run really big job, that's fine, but if you need to run big jobs every day of the month then it's crazy to use Amazon.

They are not fast (latency wise) either for web hosting. Why on earth use them when you can get VPS for £25 per month and it can handle million requests per day if you are good with coding?

While I agree they are not cheap you can see from my stats that they were producing .1 second page returns on an untuned site running a beta JDK 1.7 with the database on the server, that can easily handle 1 million a day on a hosting package of about 50 quid a day. At a very poor CPM rate it is not tough to make that into a profit of some description.

I agree with you on some level that they are crap but not that they are totally crap.

MarillionFan
8th September 2011, 20:19
I thought it came in litres or fractions of :confused:

No. You're getting confused with virgins blood. That normally comes in litres in other parts of the country, of course unless your in Glasgow where it's impossible to get and trying to find three wise men is considered implausible.

Cliphead
8th September 2011, 20:20
No. You're getting confused with virgins blood. That normally comes in litres in other parts of the country, of course unless your in Glasgow where it's impossible to get and trying to find three wise men is considered implausible.

Now that is a true.

d000hg
8th September 2011, 20:38
They offer excellent vendor lock in for sure - what a lot of start ups need to do is write efficient code or hire people who can do that, this way they won't be using many resources in the first place: we handled literally millions of queries per day with a handful of cheap boxes, lots of very cheap dedicated servers out there that will be managed by the company that rents them out.Actually with modern hardware a startup doesn't need efficient code, until they get popular and explode. That's why FaceBook were fine with PHP until they had to write their own "compiled PHP" optimiser and why Twitter was able to run on Roby on Rails initially.

Hacking some crap code on a single dedicated server will let you manage enough users to get a VC sniffing around, then you can splash all their money on fixing it.

Cliphead
8th September 2011, 20:41
Actually with modern hardware a startup doesn't need efficient code, until they get popular and explode. That's why FaceBook were fine with PHP until they had to write their own "compiled PHP" optimiser and why Twitter was able to run on Roby on Rails initially.

Hacking some crap code on a single dedicated server will let you manage enough users to get a VC sniffing around, then you can splash all their money on fixing it.

Never a truer word said.

AtW
8th September 2011, 20:41
Actually with modern hardware a startup doesn't need efficient code, until they get popular and explode. That's why FaceBook were fine with PHP until they had to write their own "compiled PHP" optimiser and why Twitter was able to run on Roby on Rails initially.

Facebook has to build custom datacenters to pay for their original mistake, and Twitter does not work half the time despite hundreds of millions of dollars of investments.

Problem with fixing things after VC money comes in is that your competitors at the same time try to catch up whilst you hack tulipy PHP code. In the end you lose.

That's why a lot of startups use Amazon "Cloud" because they can tick the "Scalability" box when they chase VC money - it is just assumed that just because stuff is in the cloud it will scale well.

d000hg
9th September 2011, 06:43
Facebook has to build custom datacenters to pay for their original mistake, and Twitter does not work half the time despite hundreds of millions of dollars of investments.

Problem with fixing things after VC money comes in is that your competitors at the same time try to catch up whilst you hack tulipy PHP code. In the end you lose.Yes if only those two had listened to your advice, they could be leaders in their fields now.

minestrone
9th September 2011, 07:41
Twitter must have one massive association table in there.

minestrone
9th September 2011, 08:22
Facebook has to build custom datacenters to pay for their original mistake, and Twitter does not work half the time despite hundreds of millions of dollars of investments.


Facebook must get a ridiculous amount of hits serving up ridiculous amounts of data, if you are somehow suggesting facebook and twitter are technology failures then your judgment is way off.

PAH
9th September 2011, 08:53
To see how the big boys do it I love reading the following site from time to time:

High Scalability - High Scalability (http://highscalability.com/)

Are there any other sites offering genuine info on how the big internet players are getting the job done?

Though the best way of avoiding infrastructure headaches is to keep your site traffic small. Works ok if most visitors are paying rather than just browsing. i.e. a subscription based site rather than one dependent on ad revenue.

AtW
9th September 2011, 09:28
Facebook must get a ridiculous amount of hits serving up ridiculous amounts of data, if you are somehow suggesting facebook and twitter are technology failures then your judgment is way off.

Facebook is a success because they had a lot of cash and smart people to deal with it.

Twitter is a technology failure because they are down half the time I try it despite fairly easy to deal with problem (mostly read only DB with very short record sizes).

PAH
9th September 2011, 09:36
Do you have a blog or something where you have detailed your progress with SKA from a technology point of view?

It's interesting to read how an internet based idea survives and becomes a success, as for every success there are thousands of failures.

If you don't have the info in a blog, then a book may be worthwhile.

Any books on how Facebook et al kept going, from a technical point of view, from startup to eventual profit making success? Seems a gap in the sale of knowledge market if there aren't any books covering that in depth.

It can't all be down to 'more cash'. There's got to be some technical lessons in there for amateurs like me to learn from. Easier than learning from ones own mistakes.

AtW
9th September 2011, 09:39
If you don't have the info in a blog, then a book may be worthwhile.

I am having whole chapter dedicated to CUK :mad

doodab
9th September 2011, 09:49
Are there any other sites offering genuine info on how the big internet players are getting the job done?

You get some good articles in Communications of the ACM & Queue from time to time.

PAH
9th September 2011, 09:49
I am having whole chapter dedicated to CUK :mad


That chapter can be summed up with "This page intentionally blank".

AtW
9th September 2011, 09:51
You get some good articles in Communications of the ACM & Queue from time to time.

Aye.

Pretty good mag and you get it free on top of your ACM membership and access to very nice digital library of articles from all over the world.

I wish I had time to read them. :frown

PAH
9th September 2011, 09:51
You get some good articles in Communications of the ACM & Queue from time to time.


Did that appear as a guest publication on Have I Got News For You?

Sounds a bit dry. Is there an audiobook version read by Ken Dodd?

minestrone
9th September 2011, 09:53
Twitter is a technology failure because they are down half the time I try it despite fairly easy to deal with problem (mostly read only DB with very short record sizes).

Can you just tell me what you think 'a half' actually means before we progress with this discussion?

Is it 1/2 or 50% or do you just make up basic maths constructs outwith common mathematical terminology.

Let me give you an example, if I go down the high street to buy a loaf of bread and one bakers has 100 people in the queue and the other has 50 which one has the half sized queue?

doodab
9th September 2011, 09:54
Aye.

Pretty good mag and you get it free on top of your ACM membership and access to very nice digital library of articles from all over the world.

I wish I had time to read them. :frown

You need to eat more fibre.

AtW
9th September 2011, 09:57
Can you just tell me what you think 'a half' actually means before we progress with this discussion?

It's down (overloaded to run search) frequently enough to be noticeable.

It's not 50% of the time, but easily around 10% - no other top site does that, and no other top site has got as simple problem with data (small size never updateable message added to read only DB).

I'd solve it for $10 mln, right after sorting out NHS.

HTH

minestrone
9th September 2011, 09:58
It's down (overloaded to run search) frequently enough to be noticeable.

It's not 50% of the time, but easily around 10% - no other top site does that, and no other top site has got as simple problem with data (small size never updateable message added to read only DB).

I'd solve it for $10 mln, right after sorting out NHS.

HTH

So you think a half is 10%. OK.

AtW
9th September 2011, 10:00
So you think a half is 10%. OK.

10% is very noticeable and can give perception site is down more than actual time it is down.

They are just raising $400 mln out of $800 mln round and perhaps they'll get it fixed.

PAH
9th September 2011, 10:01
The clue is in "It's down half the time I try it".

Maybe just unlucky that it's down half the time he tries it. Should try it more often to get a better success rate. :rolleyes:

d000hg
9th September 2011, 10:02
10% is very noticeable and can give perception site is down more than actual time it is down.If it was 5% would that seem like more or less than 5%? At what magic value does a number appear to be it's actual value?

minestrone
9th September 2011, 10:04
What is this about twitter being a read only DB?

Beyonce rubs her belly and gets 9000 posts a second.

You honesly are a complete idiot claiming that and also that 10% is a half.

Has that Dutch guy given up yet on you?

AtW
9th September 2011, 10:07
If it was 5% would that seem like more or less than 5%? At what magic value does a number appear to be it's actual value?

Here is how it works for top sites.

When was Google last down for you? I can't remember.

When was Twitter failing to run search due to overload? Yesterday.

minestrone
9th September 2011, 10:09
The clue is in "It's down half the time I try it".

Maybe just unlucky that it's down half the time he tries it. Should try it more often to get a better success rate. :rolleyes:

If it is down half the time he uses it he cannot work a browser and or connect to an ISP.

Like that old Doctor joke that the patient thinks his leg arm and head cause him pain when the he touches them but really it is the finger that is broken.

AtW
9th September 2011, 10:14
Beyonce rubs her belly and gets 9000 posts a second.

That's 1.4MB/sec, this can be cached aggressively given that only recent tweets are shown.

Twitter has got 1 bln tweets per week (source (http://thenextweb.com/twitter/2011/03/14/twitter-users-now-sending-1-billion-tweets-per-week/)), that's 140 GB per week ... just over memory capacity of some of our servers that we use.

Technologically Twitter is tulip despite lots of money put in it, their business model does not exist and the only thing they succeeded in is being famous and popular.

d000hg
9th September 2011, 10:26
Here is how it works for top sites.

When was Google last down for you? I can't remember.

When was Twitter failing to run search due to overload? Yesterday.Stop trying to deflect us from your idiocy. You said 10% seems like more than 10%. Would 1% seem like more than 1%?

d000hg
9th September 2011, 10:27
Technologically Twitter is tulip despite lots of money put in it, their business model does not exist and the only thing they succeeded in is being famous and popular.Is that why you hold such a grudge - you're jealous?

minestrone
9th September 2011, 10:28
That's 1.4MB/sec, this can be cached aggressively given that only recent tweets are shown.

Twitter has got 1 bln tweets per week (source (http://thenextweb.com/twitter/2011/03/14/twitter-users-now-sending-1-billion-tweets-per-week/)), that's 140 GB per week ... just over memory capacity of some of our servers that we use.

Technologically Twitter is tulip despite lots of money put in it, their business model does not exist and the only thing they succeeded in is being famous and popular.

So you just told me that it was a read only DB which does 1 billion inserts a week.

This is fantastic.

minestrone
9th September 2011, 10:31
Is that why you hold such a grudge - you're jealous?

Definitely the way it looks to me.

doodab
9th September 2011, 10:34
You said 10% seems like more than 10%. Would 1% seem like more than 1%?

Almost certainly. People are very bad at judging this sort of thing accurately. That is why when looking at performance problems in IT systems it's a good idea to look at 90th percentile response times (i.e. the worst 10%) and not averages. That 10% of requests will be responsible for 99% of user dissatisfaction.

AtW
9th September 2011, 10:39
Stop trying to deflect us from your idiocy. You said 10% seems like more than 10%. Would 1% seem like more than 1%?

If there was a 1% chance that you were executed then I can assure you that you'd feel that you are in mortal danger and about to die.

minestrone
9th September 2011, 10:40
Alexi has always mouthed off about his 1 billion links claim and how that is the highest publicised amount when other have publicised it for years. Greater amounts in fact.

AtW
9th September 2011, 10:40
So you just told me that it was a read only DB which does 1 billion inserts a week.

It's a read only DB after insertion, so no updates necessary, no need for transactions, no need to return data older than 7-10 days, no need to even be up all the time since users don't pay any money.

And there are hundreds of millions of dollars available to buy best technology out there.

AtW
9th September 2011, 10:44
So you just told me that it was a read only DB which does 1 billion inserts a week. This is fantastic.

We have about as much every day and unlike Twitter our data can actually be updated rather than just new insert.

minestrone
9th September 2011, 10:50
It's a read only DB after insertion.


:laugh

So you are saying it is read only database after the 1 billion inserts per week.

I pause for 10 seconds...

...how many inserts have went into the twitter databse in that time?

d000hg
9th September 2011, 10:52
If there was a 1% chance that you were executed then I can assure you that you'd feel that you are in mortal danger and about to die.No I'd feel worried but not that I expected it to happen


Alexi has always mouthed off about his 1 billion links claim and how that is the highest publicised amount when other have publicised it for years. Greater amounts in fact.Maybe those figures are not to be trusted anyway based on this thread. Might have misplaced a couple of zeros.

doodab
9th September 2011, 10:54
A billion links isn't very much. I managed to follow around half that many yesterday, and that was just on youporn.

sasguru
9th September 2011, 10:56
If there was a 1% chance that you were executed then I can assure you that you'd feel that you are in mortal danger and about to die.

Seeing as the risk of your dying from some diseases is higher than this in your thirties, you should be in a permanent state of fear.

You really are a complete moron.

minestrone
9th September 2011, 10:57
Maybe those figures are not to be trusted anyway based on this thread. Might have misplaced a couple of zeros.

I have a 1 year old excel page of about 800 links to a site I own and SKA still thinks it only has one

:laugh

PAH
9th September 2011, 11:13
When was Google last down for you? I can't remember.

When was Twitter failing to run search due to overload? Yesterday.


Failure of the internet's top players is extremely rare:

BBC News - Microsoft online services hit by major failure (http://www.bbc.co.uk/news/technology-14851455)

:laugh

PAH
9th September 2011, 11:18
We have about as much every day and unlike Twitter our data can actually be updated rather than just new insert.


If you're scraping so much every day, how do sites know you're not DoSing them?

Is there a level where you can scrape the info in a timely fashion without them crying foul?

:nerd

minestrone
9th September 2011, 11:24
Is there a level where you can scrape the info in a timely fashion without them crying foul?


The majestic 12 bot hits my sites with such infequency that is never going to be a problem for him.

sasguru
9th September 2011, 11:26
So I still don't understand - does ATW have a credible business model or not?

minestrone
9th September 2011, 11:28
So I still don't understand - does ATW have a credible business model or not?

He thinks a half is 10% and twitter does not work.

I am going to say no.

PAH
9th September 2011, 11:46
The majestic 12 bot hits my sites with such infequency that is never going to be a problem for him.


Maybe it can tell from other links to your sites that it's already captured, that your site hasn't been updated since the last scan.

After all SKA is only interested in links not content.

Which is a shame as he could offer us cheap backups of certain websites (ahem) instead of us all having to download it. About time ISPs had some competition in the internet market.

I've reserved InternetOnAStick.com for when the tech catches up. They can throttle and block the internet as much as they like once I've my own copy. :devil

AtW
9th September 2011, 12:35
If you're scraping so much every day, how do sites know you're not DoSing them?

Well, there are over 100 mln active domains out there, just grabbing 10 links each over 24 hours would make up 1 bln.

d000hg
9th September 2011, 13:34
So I still don't understand - does ATW have a credible business model or not?It depends if you ask in General or not.

minestrone
9th September 2011, 13:50
Well, there are over 100 mln active domains out there, just grabbing 10 links each over 24 hours would make up 1 bln.

You have a system high on updates and low on reads, what makes twitter work is that they actually get people to read their data.

AtW
9th September 2011, 13:55
You have a system high on updates and low on reads, what makes twitter work is that they actually get people to read their data.

Twitter gets a lot of reads which makes things easy because their database is very small: 140 GB over 1 week period max (assuming all tweets use up all length which they don't). Their job is more trivial because they only show recent tweets and rank by recency, in situations when they have 1 mln tweets they just take more recent ones, which means in their full text implementation they can pretty much can drop inverted index data if it exceeds 1000 tweets, this makes it very easy.

minestrone
9th September 2011, 14:17
Twitter gets a lot of reads which makes things easy because their database is very small: 140 GB over 1 week period max (assuming all tweets use up all length which they don't). Their job is more trivial because they only show recent tweets and rank by recency, in situations when they have 1 mln tweets they just take more recent ones, which means in their full text implementation they can pretty much can drop inverted index data if it exceeds 1000 tweets, this makes it very easy.

You seem to think running websites is purely down to DB datasize.

PAH
9th September 2011, 14:29
Well, there are over 100 mln active domains out there, just grabbing 10 links each over 24 hours would make up 1 bln.

That sounds a bit daunting having to keep on top of that lot.

Can you build in some Quality of Content checks then throw out a list of websites worth visiting?

100 million domains yet I only ever visit about 10 of them on a daily basis.

Seems there's a lot of crap out there not worth bothering with. How to find the diamonds in the rough? Google searching is useless, someone needs to invent a better way.

AtW
9th September 2011, 14:32
You seem to think running websites is purely down to DB datasize.

Front end stuff is very easy to run in parallel very cheaply, it's the large scale DB that is a problem for companies like Twitter, Facebook, Google et al.

Real time nature of Twitter certainly made it harder to implement than usual batch processing however inherent advantage in terms of small text size and write once read many times approach make their problem fairly trivial to solve.

It's all really matter of perspective - when you spend your own £50k on stuff like this you'd have to be smart, but when you want to raise hundreds of millions making problem easily solveable will backfire.

AtW
9th September 2011, 14:33
That sounds a bit daunting having to keep on top of that lot.

It's fairly easy so long as you have scalable crawler, the real problem is that handful of very large sites have got hundreds of millions of URLs on them.

doodab
9th September 2011, 14:42
Front end stuff is very easy to run in parallel very cheaply, it's the large scale DB that is a problem for companies like Twitter, Facebook, Google et al.

Real time nature of Twitter certainly made it harder to implement than usual batch processing however inherent advantage in terms of small text size and write once read many times approach make their problem fairly trivial to solve.

It's all really matter of perspective - when you spend your own £50k on stuff like this you'd have to be smart, but when you want to raise hundreds of millions making problem easily solveable will backfire.

Twitter and their ilk can also relax ACID constraints in certain ways compared to a more conventional DB i.e. it doesn't matter if every user sees the latest tweets from every other user at the same time, no one will notice if they are delayed by 500ms or they don't get the exact same ordering twice in a row. I think it's called BASE (as opposed to ACID).

eek
9th September 2011, 14:44
You seem to think running websites is purely down to DB datasize.

I think the problem is that you end up looking at all systems like your own and the solution to your issues become the solution to their issues.

Mind you most issues seem to boil down to one of two areas:-

getting data into database
getting data out of database. The latter is more interesting as once you decide that 100% this millisecond accuracy isn't important you can take a lot of short cuts to speed up handling the data.

AtW
9th September 2011, 14:45
Twitter and their ilk can also relax ACID constraints in certain ways compared to a more conventional DB

Aye.

Twitter does not even need to work 100% of the time!!!

PAH
9th September 2011, 14:46
the real problem is that handful of very large sites have got hundreds of millions of URLs on them.

Offer a cloud service where they can host their sites then you don't need to go crawling them, you'll always be bang up to date.

I wonder if Google or M$ have thought of that yet.

minestrone
9th September 2011, 14:46
Front end stuff is very easy to run in parallel very cheaply.

Bollocks.

If I have 1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second that is easier to run than 1 person accessing 140 million chars in one second.

You talk the biggest pile of crap, truly you seem to know jack shit my simple mathematically challenged friend.

AtW
9th September 2011, 14:54
If I have 1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second that is easier to run than 1 person accessing 140 million chars in one second.

Accessing 140 mln chars in one second would require 1 Gbit connectivity and it's done trivially if you have required bandwidth (and low enough latency).

1 mln accesses to 140 charts over TCP/IP might actually be more difficult problem if lots of separate IPs are involved but in such scenarios having 100 cheap boxes would reduce the problem to 10k accesses each per second which is doable.

doodab
9th September 2011, 14:58
amazon CTO Werner Vogels on eventual consistency (http://queue.acm.org/detail.cfm?id=1466448)

BASE: An Acid Alternative - ACM Queue (http://queue.acm.org/detail.cfm?id=1394128)

Scaling in games and virtual worlds - ACM Queue (http://queue.acm.org/detail.cfm?id=1378716)

minestrone
9th September 2011, 15:02
Accessing 140 mln chars in one second would require 1 Gbit connectivity and it's done trivially if you have required bandwidth (and low enough latency).

1 mln accesses to 140 charts over TCP/IP might actually be more difficult problem if lots of separate IPs are involved but in such scenarios having 100 cheap boxes would reduce the problem to 10k accesses each per second which is doable.

Can I just ask what you think is more problematic for a web server.

"1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second"

"1 person accessing 140 million chars in one second"

eek
9th September 2011, 15:09
Can I just ask what you think is more problematic for a web server.

"1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second"

"1 person accessing 140 million chars in one second"

You forget. ATW has solved his problem by using a hammer so his immediate solution to all problems is now that hammer. And if it doesn't work to buy a bigger hammer.

To be honest that statement is true of most people. They will take a working solution and try and apply it to the next problem that comes along.

Edit to answer the question.

The second one could be a problem based on the size of the network connection.
The first one is a problem for the database but less so if the database keeps recent statements in memory.

Memcache solves the first issue very well. Stackoverflow halved the number of machines they require by caching database results for 3 seconds. I'm sure most popular sites would do the same.

AtW
9th September 2011, 15:11
Can I just ask what you think is more problematic for a web server.

"1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second"

"1 person accessing 140 million chars in one second"

I'd expect the first problem to be more difficult to resolve assuming pessimistic scenario (many IPs, TCP/IP used on many sockets etc)

minestrone
9th September 2011, 15:15
I'd expect the first problem to be more difficult to resolve assuming pessimistic scenario (many IPs, TCP/IP used on many sockets etc)

I am not asking for your solutions I am asking quite clearly...


Can I just ask what you think is more problematic for a web server.

"1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second"

"1 person accessing 140 million chars in one second"

minestrone
9th September 2011, 15:19
Why do I bother, he is clearly useless.

doodab
9th September 2011, 15:19
I am not asking for your solutions I am asking quite clearly...


Can I just ask what you think is more problematic for a web server.

"1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second"

"1 person accessing 140 million chars in one second"

It's a trick question. A web server couldn't do the first one without forwarding the request to a database.

A 1/4 million hits per second for static web content is doable on a single dual xeon box using an in memory cache though and I don't see why that wouldn't scale fairly well to more hardware.

So it would seem like the DB is going to be a bottleneck.

AtW
9th September 2011, 15:21
A 1/4 million hits per second for static web content is doable

Not if you get requests from 1/4 million unique IPs over TCP/IP.

Is it just me who thinks minestrone is an idiot?

AtW
9th September 2011, 15:26
It's a trick question. A web server couldn't do the first one without forwarding the request to a database.

Who said anything about any database in this scenario? We are talking about self contained webserver that has got 140 byte file (possibly cached) and 140 MB file (also possibly cached).

In one case it gets hit with 1 mln requests for 140 bytes and in another 1 requests for 140 MB.

I know I'd prefer the latter scenario and if the user has not got bandwidth to get data file in 1 second then it's problem beyond webserver.

doodab
9th September 2011, 15:27
Not if you get requests from 1/4 million unique IPs.

Why not? You think setup and tear down of the connections would be a problem?

I might have to do some more experiments. That's the trouble with being benched, no access to proper hardware for ******* about.

doodab
9th September 2011, 15:29
Who said anything about any database in this scenario? We are talking about self contained webserver that has got 140 byte file (possibly cached) and 140 MB file (also possibly cached).

It was you who said that scaling front end web stuff was easy, then Soupy was talking about tables and fields. Perhaps he meant it would be a picnic?

AtW
9th September 2011, 15:30
Why not? You think setup and tear down of the connections would be a problem?

I think you'll run out of TCP/IP sockets.

AtW
9th September 2011, 15:31
If I have 1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second that is easier to run than 1 person accessing 140 million chars in one second.

It does not say database here, text field can be in memcache or on disk.

If you keep 140 MB file in a database then you are an idiot.

minestrone
9th September 2011, 15:32
Not if you get requests from 1/4 million unique IPs over TCP/IP.

Is it just me who thinks minestrone is an idiot?

I think you opened up a SKA thread, told us twitter worked only half the time, reduced to 10% when pressed, argued 10% was a half.

doodab
9th September 2011, 15:33
I think you'll run out of TCP/IP sockets.

You only need one on the server side. Do you mean file descriptors?

d000hg
9th September 2011, 15:36
I think you opened up a SKA thread, told us twitter worked only half the time, reduced to 10% when pressed, argued 10% was a half.You can't trust AtW just because he sometimes sounds like he knows about technical issues. After all, as the saying goes, even a stopped clock is right half the time.

AtW
9th September 2011, 15:36
You only need one on the server side. Do you mean file descriptors?

I just had a chat with our network admin and you right, there can be more than 64k internal sockets provided there is enough memory to handle it, however I still maintain 1 mln calls per 1 sec would have more overheads than handling one call sending 140 MB.

minestrone
9th September 2011, 15:39
It does not say database here, text field can be in memcache or on disk.

If you keep 140 MB file in a database then you are an idiot.

Pick out where I said MBs Alexi?

140 chars was my only data description you will find.

Completely unable to understand basic quantities you are.

doodab
9th September 2011, 15:43
1 table with one text field of 140 chars if that gets accessed 1 million times in 1 second


It does not say database here, text field can be in memcache or on disk.

It just seemed like he meant a database table to me. Maybe he didn't. We should ask him.


If you keep 140 MB file in a database then you are an idiot.

That depends. A lot of document management systems do just that. In my last gig I had to move ~15TB of files (10s of millions, some quite large) out of a database to a file system and provide multiple access paths to them. We gained a lot in ease of access but lost out in terms of being able to search the metadata, as FS metadata isn't really optimised for that.

AtW
9th September 2011, 15:45
Pick out where I said MBs Alexi? 140 chars was my only data description you will find. Completely unable to understand basic quantities you are.

Well, char for me in C# is 2 bytes, and in other times including database is 1 byte, if I wanted 2 bytes in DB then I'd expect some other term wchar (?).

Anyhow, I need to get back and do monthly invoices. Go look at something shiny ministrone.

AtW
9th September 2011, 15:46
It just seemed like he meant a database table to me. Maybe he didn't. We should ask him.

Well, let's assume he meant database, in this case I'd keep such data structures in memory and it won't be an issue to query them - there won't be a big delay reading 140 MB.

doodab
9th September 2011, 15:47
I just had a chat with our network admin and you right, there can be more than 64k internal sockets provided there is enough memory to handle it, however I still maintain 1 mln calls per 1 sec would have more overheads than handling one call sending 140 MB.

Yes I agree. But it was you who said that


Front end stuff is very easy to run in parallel very cheaply

I was just pointing out that it's doable.

minestrone
9th September 2011, 15:56
I need to get back and do monthly invoices. Go look at something shiny ministrone.

I will and I can because I have something shiny to look at with lots of shiny things. I own a house, you do not.

You are a Russian financial immigrant who split a job when you got a passport and lived off the tax payer for years.

I have said before you have cost this country more than you can give.

Sue me.

d000hg
9th September 2011, 16:08
I will and I can because I have something shiny to look at with lots of shiny things. I own a house, you do not.A shiny house? Weirdo.


Sue me.Hi Sue. I always figured you were a man.

minestrone
9th September 2011, 16:14
Honestly ATW if you ever want to sue me over my accusation that you are a dole scrounger I am very happy to go through with it, my name and lawyer are always on offer to yourself if I think you are actually capable of it.

AtW
9th September 2011, 16:17
I was just pointing out that it's doable.

I said front end can run in parallel, obviously backend database (if it is used by front end) should be as scalable.

AtW
9th September 2011, 16:18
I will and I can because I have something shiny to look at with lots of shiny things. I own a house, you do not.

You are a Russian financial immigrant who split a job when you got a passport and lived off the tax payer for years.

I have said before you have cost this country more than you can give.

Sue me.

:laugh

I'll dedicate a paragraph to you in my future book, will make sure I clear it with my lawyer first though.

minestrone
9th September 2011, 16:37
Ach, I like like winding you up alexi.