Building a website very soon need a little guidance

[Lenny Forziati]

Just to be clear - A single Alpha Application Server most certainly will NOT handle the stated load in the original post. Neither will a single IIS server or Apache server. No site with that kind of traffic runs on a single server of any type. You are going to require a load balanced cluster, no matter what server platform you choose.

Also as others have stated, you are talking about far more traffic than a T-1 or even T-3 could handle with a reasonable level of responsiveness.

Furthermore, if it is a business critical service, you will want some redundancy, which at a bare minimum means running the server cluster from a well-connected and dedicated Internet data center. If your business really depends on it though, this is something that should be hosted in multiple data centers in distinct geographic regions.

If this endeavor is something that you are committed to, but do not know how to architect a scalable and redundant architecture, Alpha Consulting Services can be engaged to gather requirements and design a suitable infrastructure for you.



As far as user limits and overall performance of the Application Server in general, we do not publish numbers because it is highly dependent on your hardware and Internet connection, as well as how you have built your application, so real-world number will vary greatly. As far as Steve's V10 tests, they are irrelevant in V11 because the server had been significantly enhanced and performance has been improved as a result. Some users have reported more than a 100% performance increase, but this too will vary with your specific application as well as hardware and connection.

Or, you would run out of money scaling Alpha to meet that requirement.

Anyone that needs to build a multi-server cluster for a large application should contact Alpha's sales department for project-specific pricing.

Just a guess, but maybe 3000 users logging in or getting there within a minute. It is the type of app where people will typically be there for more than 10 minutes and people adding records will be on for more than 30 minutes.

User concurrency in a browser-based application is significantly different than the desktop world. During "think time" - a user looking at a screen, inputting data, reviewing a report, etc. - they are not connected to the server in any way and are not counted towards concurrent users from a load perspective. Until they click submit or take some other action that contacts the server, they consume no server resources. Only then do the connect to the server, perform some action (e.g. log in, save data, get more records, generate a report) and then immediately disconnect.

This means that 3000 logged in users is not the same as 3000 concurrent users. The concurrent user count will be just a fraction of the logged in user count, though the fraction depends highly on your specific application and its use cases.

[ssmoore1]

Re: Building a website very soon need a little guidance

Hi Dave,

Sounds like one heck of an interesting project!!

I have done a bit of work on a much much smaller scale doing load balancing and you could use Alpha if you wanted to. But within the scope of the application alpha would just be one component of a very large, insanely complex architecture.

No matter what application servers you are going to go with, you are going to need a huge load balancer to parcel out all the request to the various underlying app servers. So that will be a large cost right there. From there you are going to need multiple app servers (could be alpha, apache, whatever) to keep up with all of the traffic. And the server to host the SQL database is going to be a monster as well, especially since you will be pumping in tons of photos which are big data hogs (compared to regular text data). You will almost guaranteed need to setup SQL replication with master/slave architecture which puts you into a whole new ball park of complexity.

To build a website where 10,000 people can upload a photo at the same time is going to be super expensive - I doubt youtube has that many people uploading a video at the same time.

Are you sure about your user numbers estimates? Because you will be spending a ton of money to keep a website like that up, as well as the need to back up the database, which for that size will require complex back up strategies (different ones depending on SQL engine.)

Personally I think it would be great to build and maintain something like that so I hope you are able to make a go of it, but if the website doesn't make enough money you could be losing tens of thousands of dollars a month keeping it running (and the build will cost you tens of thousands as well if you need to hire help)

Cheers,

Scott

[Rida Alhasan]

Re: Building a website very soon need a little guidance

.
Steve,
Thank you for the prompt reply. So it seems that the reasonable way to test the scenario I described is simply to carry it out, maybe by reducing the number of users to 10% or 20%, keeping the rest of requirements, and then extrapolate to arrive at a "more or less" result.

Regards,
Felix
.

[DaveM]

Re: Building a website very soon need a little guidance

the input records are pretty straight forward with 25 image fields and a max of 60 fields. They will be on the server a maximum of 60 days and be deleted. some will be gone before that. The images will all be transformed and made smaller as they come into the site. They will be about 40k in size.

I have not done the math on most of the data to determine db size yet. I know it will be a lot.

Some of what Lenny said made me feel much more at ease.

[Steve Wood]

Re: Building a website very soon need a little guidance

I think from the other responses you can see why I don't enjoy discussing user capacity.

[DaveM]

Re: Building a website very soon need a little guidance

Understood Steve.

I am working on a solution for this. It seems the large expense is for the host servers. I am not going to try to host it myself. It was one of the questions I was asked and passed on an answer until I searched it out.

You can get space on a host that is very capable of handling it all. It is not cheap, but less than I thought it might be. It is great that I can start it out on a few and upscale it as it is needed.

[Steve Wood]

Re: Building a website very soon need a little guidance

It was one of the questions I was asked...

Ah, this is for someone else, not an application for you. I know you have a very strong business and sales background so don't take this like I am preaching, and I have no idea if this is accurate in your case -- but typically these opportunities come with little or no upfront or concurrent payment to the developer, and your income is to come out of future revenue. Don't know if that is the case here but if so you should ask if payment will be from net or gross revenue and how they plan to pay the $10-50 per user it will likely cost them for their 30,000 users. Though not impossible, it is very unlikely users are going to flock to the application in massive numbers no matter what it is.

[DaveM]

Re: Building a website very soon need a little guidance

OK,

Hosting on cloud = mysql - 1gig mem per 10,000 people at a time - unlimited space or at least what I need - PHP - ajax -

It looks like I can start about 650 a month/less at pure start up and scaled to the 650.00 about 1 year and to a total for the numbers I gave at about 2000.00 a month.

Some of this they suggested and it includes load balancing and however many servers needed.

I am happy about that.

I already have a php generator for mysql and all the other tools I need but one.

I am happy, but know it will not be easy!

[Roderick]

Re: Building a website very soon need a little guidance

This sounds like something that should be on Amazon or Rackspace servers. Contact Zebra too.

I would also consider starting with a smaller group and ramping up in increments if at all possible.

[DaveM]

Re: Building a website very soon need a little guidance

Thanks all for your input. It was very helpful. I had posted this on a few other forums and this one(as always) made the most sense.

Actually, it is my idea and my endeavor. It is so possible to succeed that it scares me. It definitely scared some others. The pre math done on a VERY under achieving score scared some. These were people who could have helped me in the developement and I asked no money from them.

I am going to go it by myself, even if it does take a few months longer. This opportunity will not die for years and I know the business side of the industry as well as how to promote it.

Even though it will generate around 13m per year at about 3 years(conservative), there will be a strong advertising fund to grow it more.

I do not intend to take anything out until then. I don't have to.

The corporate setup is being done, cpa is set, appropriate attorney's are set up, tools are set(mostly), Path to promotion is set, business plan(conservative) is set, money is 90% set, hosting is still between a few. I just have to write code. Db's are 80% done(some work left).

I will let you know when it is online and useable. You can then grade me. That is always helpful too.