Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

need help : urgent


I'm starting a business in web hosting and ASP(email services and HR services).

I need help about the right architecture for my network. I don't want to buy bandwidth from ISP but I prefer to setup my network.. I will have about 10 servers(2 for DNS, 2 for mail, 1 for LDAP, 1 linux web hosting, 1 IIS server, 1 server for applications) and a SAN storage solution.

I need to setup my network : which router and firewall is the right for my architecture, which switch I need to buy, what is the best architecture for this : rouer, firewall, switch,10 servers.

thanks .


Re: need help : urgent

First things first. If you are going into the ASP business, availability of your services should be the first thing that you think about.

If you cannot afford high availability at the start of your business, then I would suggest investing in key areas that enable a HA upgrade when your business grows.

The first element that you will have to deal with is power and cooling. It is very expensive to provide those things redundantly.

The second element is Internet connectivity. Again it is expensive to provide this in a redundant fashion

In my honest opinion, for an embryonic startup it is cheaper (and easier in the long run) to host at a colocation facility that can provide for these first two items. This allows you to focus on moving your business forward instead of futzing with a T1, or your UPS.

Continuing with the them of positioning for High availability I would recommend starting with and ASA 5510 for your firewall. This can be upgraded to a HA configuration at a later date.

Behind that I would put an CSS 11501 (or place your favorite load balancer here) This will allow you to create virtual representation of your web farm, so when you need to scale out horizontally for load or availability it will be transparent to your end users.

Behind that you need to have a switching layer. I would suggest going with a 3750G-24TS. This will allow you to switch at gigabit speeds between all your devices. If you choose to use IP storage this switch will support it. And following the theme of the rest of the devices, the backplane power and supervisor functions can be extended to support high availability at a later date.

Last but not least, you mentioned a SAN. For a small site, I would recomend either a mds9120 or 9216i. The main choice is that if you intend on using ISCSI, the 9216i is a perfect choice. While if you only intend on using fibre channel, the 9120 will be just fine.

When you do grow, these devices can be implimented in a HA multipath configuration.

If you find this helpful, please rate this post.


New Member

Re: need help : urgent

Thanks for your mail.

Please could you valide my this architecture.See the file attached.


Re: need help : urgent

A 3845 would cover all your bandwidth needs.

Though, if you have a redundant ISP connections, then you are only getting uplink redundancy, not hardware redundancy.

One interesting thing in your architecture. You have redundant email, but not redundant web and database servers. That just strikes me as odd.

I would think that you would want to invest in redundancy for internet facing applications,not email which can easily be outsourced for any small startup.

New Member

Re: need help : urgent

Also note that this is all terminating in a single switch. While this would work, I don't think that you will be able to adhere to any SLA's which your clients will require in the event of a hardware failure. I would recommend taking a serious look into your business requirements before moving forward with a physical design like this.


Re: need help : urgent

That is a note in each of my recommendations.

My experience with startups is that they never want to pay for redundancy in the embryonic stage.

Sadly they only want to pay after an extended outage. That is why every suggestion made has the capability of being extended to a HA configuration.