1. Newbie
Join Date
Sep 2006
Location
New Jersey
Posts
17

## Bandwidth Calculation

Hi,
I have a question regarding bandwidth. How many servers can you put on a 10mbit pipe if you offer 2000gig bandwidth per month per server? My calculations are between 12-13 servers at .77mb per sec per server.

Here is how I came up with the figure.
2,000gig is equal to 2,000,000mb.
60seconds x 60minutes x 24hrs x 30 days = 2,592,000.
2,000,000mb divided by 2,592,000 = 0.77mb

10mbit divided by 0.77 is 12.98.
Thats 12-13 servers at 0.77mbit dedicated connection each.

Thats how I came up with 12-13 servers at 2,000 gig bandwidth per server per month.

Is this correct?

Thanx,
Calvin

2. Web Hosting Master
Join Date
Apr 2003
Location
San Jose, CA.
Posts
1,622
In short no.

Simply, because your bandwidth is not going to be spread evenly across each server or over an even time frame.

If you limited each server to .77mbps and each server was sending in/out data 100% of the time then your calculation would be correct... that just isn't likely going to happen tho.

3. Newbie
Join Date
Sep 2006
Location
New Jersey
Posts
17
Hi Lightwave.
No, I wont be restricting each server to .77mbit pipes. They will be on a 100mbit port. How would I go about calculating this then?
How many servers can I put on a 100mbit port if I was to offer 2,000 gigs of bandwidth per server per month? How do I do the calculation then?
Thanks,
Calvin

S

4. Web Hosting Master
Join Date
Dec 2001
Location
Posts
5,946
Unless you're talking about gigabits and not gigabytes, you're mixing mbytes/mbits gbytes/gbits.

1mbps = ~324GB/mo of actual data transfer (if maxed 24x7), so if you intend to split 10mbps between 13 servers, you can expect to push less then 250GB/mo per server sustained max (per direction, in a perfect world, if each server was equally utilizing the bandwidth at all times).

5. Web Hosting Master
Join Date
Sep 2005
Posts
550
8 megabit = 1 megabyte per second
therefore, you can only transfer 12.5 megabytes per second on a 100 megabit connection.
In your case, if you wanted to transfer, 2000 gigabytes per month per server on a 10 megabit connection, you could only provide this to 1 server without overselling.

6. Newbie
Join Date
Sep 2006
Location
New Jersey
Posts
17
Ahhh, I see. So I should not put more then 10 servers on a 100mbit connection.

Thanx for the info guys. I really appreciate it.
Calvin

7. Web Hosting Evangelist
Join Date
Apr 2001
Posts
542
Yikes!

No, the only correct answer is "it depends".

If you sell servers to programmers/developers, they might use very little bandwidth and you could easily put 50 servers on a 100mbit line.

If you sell servers to people offering free services supported by ads (such as free hosting, or bulletin boards) then you will probably find the servers are busy during peak times (lunchtime, dinnertime, evening) and idle other times.

If you can get a good customer mix such that idle times are spread out, then you can still support a lot of servers without a bandwidth issue.

Originally Posted by xoopa
Ahhh, I see. So I should not put more then 10 servers on a 100mbit connection.

Thanx for the info guys. I really appreciate it.
Calvin

8. Junior Guru Wannabe
Join Date
Jan 2006
Posts
32
The exact conversion is 2000 GB/Month of Data Transfer = 6.17 Mbps.

Or to put it in easier to understand terms that means:
2.78 GB/hour
66.64 GB/day

spiv is correct. In hosting there is a general understanding 10% of your customers will use 90% of your capacity. The other 90% will of your customers will use jsut 10% of your capacity.

That comment while interesting really depends business by business. What you really need to do is determine your best and worst case scenarios and then make sure your business model is viable. Can you still successfully provide service to your customers at the price point you were targeting if one customer maxed out your packages? Could you do it if 10 customers maxed them out? What if 50 customers used just 10% of their available capacity?

Hosting and capacity management is a pretty tricky game, but you know your customers best.

9. Newbie
Join Date
Sep 2006
Location
New Jersey
Posts
17
Hi,
Fusionapp, yeah, thats what I was trying to determine. I wanted to make sure in the worst case scenerio, my business model can survive it. Seems pretty tough. The profit margin is minimal if you purchase low commits to start with. Its kinda hard to purchase high commits in the very beginning without a good # of clients.

Thanks again all,
Calvin

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•