As part of my investment from eXscape, I have gotten searchspider.com. A search engine domain that I might think of using in the future.

I will start to put this together when and if I ever get money. I will start a search engine similar to google; searchspider is kind of natural.

We have the search engine already ready to 'crawl' millions of website.

I Need the following equipment.
- A fast quad processor system
- 2 Terabyte+ Space (my first guess would be to use scsi.. as it has low seek time).
- a 10mbit connection.

We would get popular in approximately 6 months.

The sponser (hoping to be a ISP)..
would get ad space with us for 1 year to take advantage of thousands of hits we expect to get each month. In return, they must provide us with funding for or a high capacity server. Ultimately, they would get way more advertising than their funds.

Originally posted by ShareFile
As part of my investment from eXscape, I have gotten searchspider.com. A search engine domain that I might think of using in the future.

I will start to put this together when and if I ever get money. I will start a search engine similar to google; searchspider is kind of natural.

We have the search engine already ready to 'crawl' millions of website.

I Need the following equipment.
- A fast quad processor system
- 2 Terabyte+ Space (my first guess would be to use scsi.. as it has low seek time).
- a 10mbit connection.

We would get popular in approximately 6 months.

The sponser (hoping to be a ISP)..
would get ad space with us for 1 year to take advantage of thousands of hits we expect to get each month. In return, they must provide us with funding for or a high capacity server. Ultimately, they would get way more advertising than their funds.


Just so that you know, the basic cost for such a config is ( i happen to have one very simular to his request, that I did for a client)

2 terra raid about about 26k- 28K
1 quad box with all the goodies about 12 K
10 mbit connection ( non cogent ) 3 k to 6 K

now just to point out some flaws in the design, you do not want a single point of failure in a search engine. so for that 12K quad box, swap it out for, 4 servers clustered for the search engine, 1 server for the load balancer, and take the rest of the money for the cluster consultant. that 10 mips connection should be multihomed with atleast 4 tier 1 carriers ( that's why I placed the high price )

if a host was even to invest in this, I would think that they would want a multi-year contract with some assurances like because they are the VC for all intensive looks of it.
: percentages of monthly revenue
: percentages of all ad space
: 1 board member

Mike

So sorry, I misread your offer a bit.

I feel and see no need to start a competitor to Google as the question "WhY Bother" comes to mind.

But in any event, I do not feel it to be a fruitful endeavour unless you have a technically superior means of doing what Google does.

I don't feel the way patprice feels as you can always start a competition and you never ask the question 'why bother' but always ask the question 'why not'? But better make sure you have a good answer to your question before you jump in. If you know your stuff, then this is the wrong place to get your venture off the ground.

Anyone can get a domain and start an engine - you need lots more to make it successful. Anyone wanting to compete with G, better have some backing. Sorry, but I say you can forget about making it popular in 6 months - even google with such visible partnerships with yahoo, etc., took 3 years to get some visibility.

I agree with the previous post of a single box. Do like Google does and cluster some servers together. Last I knew Google even had some 486's in their cluster farm. This makes for some more maintenance, but cuts down on cost quite a bit.

>Just so that you know, the basic cost for such a config is ( i >happen to have one very simular to his request, that I did for a >client)

>2 terra raid about about 26k- 28K
>1 quad box with all the goodies about 12 K
>10 mbit connection ( non cogent ) 3 k to 6 K

No, your wrong. With proper people, you can have around 1 terabyte of space and a nice quad server for around 10k total.

You would atleast to spend around 1K/month for 6-7 months to see real results on cogent level connection to actually get going with this. Cogent is pretty reliable .. regardless of what anybody says.


>now just to point out some flaws in the design, you do not want >a single point of failure in a search engine. so for that 12K quad >box, swap it out for, 4 servers clustered for the search engine, 1 >server for the load balancer, and take the rest of the money for >the cluster consultant. that 10 mips connection should be >multihomed with atleast 4 tier 1 carriers ( that's why I placed >the high price )

You are partly correct, however, to do this you require a CPU on a single server as well as all the ram in a single server. Clusters are good, however, most of the clusters google uses to crawl.

Google had (before) 1K servers which had about 1-2GB hdd each.

Load balancers are necessary if you would be getting millions of hits a day.. however, its is not necessary in a setup which would only expect thousands in a month.

Regardless of how much google has data (which approximate to around 1 terabyte in total..).. its never enough.. internet is so huge that no one search engine will almost never be able to search it for atleast another couple of decades. Even now, put together all of the systems, internet is alot like a unexplored world.. where about 10% is known.. rest is unknown.

>if a host was even to invest in this, I would think that they >would want a multi-year contract with some assurances like
>because they are the VC for all intensive looks of it.

Probably not. We are really expecting a host that would take in the impressions for one year( prefer a webhost/isp.. of whom services will be known to thousands each month)as well as fund the engine for expansion if that becomes necessary.

The host can easily benefit from say.. in 4-6 months, it would start to get around 900K hits .. (regardless however, this is a long term investment.. there is nothing about a rich quick scheme here)... at the rate of 10$/CPM.. that becomes to 9K just for that month.. so if you should look at the entire year.. the investment is actually has almost 500% return.


: percentages of monthly revenue
: percentages of all ad space
: 1 board member

Yeah.. you should have that.. when the investment is actually in couple of hundred thousands.. other than that, I doubt that would be ever a case with ANY company.. let alone another startup.

The other thing is.. even if you tried to advertise via another company just for your company's campaign, you would end up spending 10K/month anyway . Why not rather spend it on something where you may get double/tripple/quad return ? Most companies have to spend thousands and thousands to buy impressions regardless to get their company known. I think it is probably best idea for a company of that nature to sponsor it as they won't really be wasting money but getting alot in return.

However, our software can crawl upto 2 million urls a day per server. The big draw backs are mostly space.. which can be solved later on I guess.. but perhaps we should put up a demo first with just a couple of millions of pages ?

Also to note, I agree with Gurudev 100%. Most people don't really know the field of search enginer and research involved with them but I guess Gurudev said it best, "Why not ?"