?
Solved

Site replication and real time fail-over

Posted on 2012-08-24
4
Medium Priority
?
1,047 Views
Last Modified: 2012-08-29
I need some advice on network design of a multi-site network (high level ). Here's what we need to do:

We will be hosting 24/7 web based transaction processing site(s) that requires high availability. We need to have the ability to failover between sites (data centers) in the event of a network failure, etc..For example if site A hosts money.domain.com and experiences a service interruption, we need a seamless redirection of traffic for money.domain.com to site B (web and data servers at both locations).

This is all new territory for me, but my initial idea would be the following:

Have 2 physical data centers A & B, with large pipe/tunnel between them, which each hosts replicated web servers and clustered SQL boxes. Then use a Cisco Global Site Selector (GSS) network to manage/direct the traffic to site A or B in the event of an outage.

From a high level does this sound feasible? Any suggestions or things to avoid? Any advice is greatly appreciated.
0
Comment
Question by:ROMAD77
4 Comments
 
LVL 1

Expert Comment

by:lunanat
ID: 38329704
I'm not really an expert in the issue, but I saw your question so I thought I'd pop in.

What you want to look into is multi-homing, and BGP.

For the seamless transition you want, it is my understanding that it must happen at the routing layer... essentially, imagine you have two routers with essentially the same configuration, one at each site.

Your ISP's routers know how to route the last hop to your production network.  What they need is to know that if that last hop is unavailable, it can be reached at "this other" address instead. "This other address" is of course your DR site.

BGP failover can happen very VERY quickly.

It's probably going to be somewhat expensive, depending on your ISP and their willingness to set something like this up.  I know my ISP can do it, for a price.. but it never really was "in the books".
0
 
LVL 6

Assisted Solution

by:Lee_YCP
Lee_YCP earned 1000 total points
ID: 38329783
ROMAD,
Your described goal is exactly the setup we have.  We use eBGP and a back-end T1 for data replication.  If my T1 goes down or I need to move big files from one site to the other, I use a site-to-site VPN because a 50Mb (primary site) to 10Mb (backup site) pipe goes much faster than my backside T-1.

The BGP failover takes about 30 seconds to fail from one site to the other.  Due to our size and minimal requirement for IP addresses, we werent big enough to get a block from ARIN, so we got multiple blocks from our ISP (large Telecom carrier).  We got one block for each site.  To use BGP you have to have an ASN that ties to your IP block, well we didn't.  Our ISP had an ASN for us to use because they are actually the ISP at both our primary and backup locations, which are about 180 miles apart.
0
 
LVL 8

Accepted Solution

by:
gsmartin earned 1000 total points
ID: 38332765
BGP multi-homing is one option.  In my environment I use a WAN Aggregator/ Load Balancer that is also our DNS SOA for both data centers.  This enables us to manage our own DNS and quickly make changes.  Typically, when ISPs manage your DNS changes take approx. 7200 seconds vs in our case 30 seconds.  This allows us toile changes on the fly and have to wait for 2 hours TTL to expire.  Also DNS will intinally advertise both available IP addresses for an active-active configuration.  For example:  money.company.com (66.140.10.1 and 66.140.11.1).  Now, if a server goes down or is taken down for maintenance, DNS (when checking the service availability) will remove the downed IP address from the DNS advertisement; the change only taking 30 seconds for the original DNS entries to expire accross the internet.  And, now all DNS resolution traffic is only resolving the available IP address out of the DR or secondary data center.

The product we use (PowerLink 600 EHQ) is made by a company called Ecessa (www.ecessa.com), which we've been using for over three years.  I prefer their product over F5 and Radware for the price and feature set.  It's a great versital appliance (set up in a HA configuration) providing flexibility to our multi-site data center (HA) configuration for various scenarios beyond DNS.
0
 
LVL 2

Author Closing Comment

by:ROMAD77
ID: 38347071
Thanks, guys. This will get me going in the right direction. Much appreciated.
0

Featured Post

Important Lessons on Recovering from Petya

In their most recent webinar, Skyport Systems explores ways to isolate and protect critical databases to keep the core of your company safe from harm.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

The System Center Operations Manager 2012, known as SCOM, is a part of the Microsoft system center product that provides the user with infrastructure monitoring and application performance monitoring. SCOM monitors:   Windows or UNIX/LinuxNetwo…
Tired of waiting for your show or movie to load?  Are buffering issues a constant problem with your internet connection?  Check this article out to see if these simple adjustments are the solution for you.
After creating this article (http://www.experts-exchange.com/articles/23699/Setup-Mikrotik-routers-with-OSPF.html), I decided to make a video (no audio) to show you how to configure the routers and run some trace routes and pings between the 7 sites…
After creating this article (http://www.experts-exchange.com/articles/23699/Setup-Mikrotik-routers-with-OSPF.html), I decided to make a video (no audio) to show you how to configure the routers and run some trace routes and pings between the 7 sites…

850 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question