Skip to main content

Windows Auto Back up Server

More
14 years 7 months ago #32062 by ardeef
Hello!

There must be a technology or at least, a way to have two servers ( let's say a database servers ) synchronized. Such that if the other server fails, the system ( the database system ) still runs smoothly as long as the second server is ok. It's like RAID but only, in different server.

If it is not possible, what could be the reason?

Please share what's on your mind.
More
14 years 7 months ago #32076 by TheBishop
It is possible. If your database was Oracle there's Oracle dataguard for example that does exactly what you're talking about. If it's just a file/application server you could look at Microsoft clustering or even something home-grown like dual servers with Robocopy to synchronise them and a suitable failure detection mechanism
More
14 years 6 months ago #32285 by ardeef
It's very interesting to know.I am trying to search all these things you've mentioned.BTW, we're using MSSQL server.

Thanks!
More
14 years 6 months ago #32288 by talk2sp
Replied by talk2sp on topic comes in handy for me....
this thread and it's replies comes in handy for me cos i have for long wanted to make this happen i even put up a thread but there was no buzz. so em ardeef thanks for firing another thread on same issue...

More replies please



C0dE - 3

BORN TO BE GREAT

c0de - 3
..........................................................
Take Responsibility! Don't let failures define you
More
14 years 6 months ago #32314 by ikon
nnbnbMS Clustering 2 Types

Cluster Failover or Network Loadbalancer


NLB = 1 Server acts as a Network Load balancer accepting all connections for SQL traffic, according to the avaliability of the SQL servers connections are passed to the Active SQL server( uses Round Robin DNS) sql.mycompany.com points to "A" record x.x.x.x and "A" record x.x.x.x

The server to respond the fastest gets the connection.


Clustering (Requires advanced versions of Windows Server)

2 or more servers setup in cluster all servers are configured with 1 Static IP address and the Cluster itself has an IP address all traffic pointed to Cluster IP, each server has a sate "Online (Active)" "Offline (Failed)"

The online server will have Cluster IP therefore accept traffic.

Cluster can also be configured fro Active Active, a mix of NLB and Cluster.


There are some probelms in each senario especialy with Database Servers, Databases will have different states and some will lock tables when certain administration is being done yet the server will still be online, you dont want traffic being sent to a DB server that is online but has a DB issue or is under administartion, and manually managing this can be a pain, so software be it built into advanced versions of SQL or from third party can be implemented aswell as clustering and NLB to monitor these states and provide and excellent solution.

I use MySQL with Apache i have 2 servers setup identically on linux and use rsync to replicate web directories i use Mysql's Built in Clustering/Replication to replicate the mysql DB's and i use MYSQL Proxy and Mysql Pool Poller.

and excellent guide here and explanation of DB states

dev.mysql.com/tech-resources/articles/fa...rategy-part1.html#t4

Thanks
Time to create page: 0.147 seconds