replicated caching topology, then if one server is down then it will take data from next server

Topics: Developer Forum, Project Management Forum, User Forum
Coordinator
Apr 23, 2009 at 8:23 AM
A question which was emailed to me: If I used replicated caching topology, then if one server is down then it will take data from next server. But the load of the server (in case of replicated caching) may be more than that of distributed caching. But it is faster than distributed cache.

the replicated system works like this: data not found on local node then go to the next configured replicated node. e.g lets say you have 2 datacenters and each of them has 2 server then it would make sense to configure your replicated servers section like this (would be within server config):

Server no. 1 in Data Center A will have this configureation:
<replicatedServers>
    <!-- DATA CENTER 1 -->
    <add key="DATA_CENTER_1_B" ipaddress="192.168.1.3" port="48888" />
    <!-- DATA CENTER 2 -->
    <add key="DATA_CENTER_1_A" ipaddress="192.168.2.2" port="48888" />
    <add key="DATA_CENTER_1_B" ipaddress="192.168.2.3" port="48888" />

</
replicatedServers>

Server no. 2 in Data Center B will have this configureation:
<replicatedServers>
    <!-- DATA CENTER 1 -->
    <add key="DATA_CENTER_1_A" ipaddress="192.168.1.2" port="48888" />
    <!-- DATA CENTER 2 -->
    <add key="DATA_CENTER_1_A" ipaddress="192.168.2.2" port="48888" />
    <add key="DATA_CENTER_1_B" ipaddress="192.168.2.3" port="48888" />

</
replicatedServers>


Server no. 1 in Data Center A will have this configureation:
<replicatedServers>
    <!-- DATA CENTER 1 -->
    <add key="DATA_CENTER_1_A" ipaddress="192.168.1.2" port="48888" />
    <add key="DATA_CENTER_1_B" ipaddress="192.168.1.3" port="48888" />
    <!-- DATA CENTER 2 -->
    <add key="DATA_CENTER_1_B" ipaddress="192.168.2.3" port="48888" />

</
replicatedServers>


Server no. 2 in Data Center B will have this configureation:
<replicatedServers>
    <!-- DATA CENTER 1 -->
    <add key="DATA_CENTER_1_A" ipaddress="192.168.1.2" port="48888" />
    <add key="DATA_CENTER_1_B" ipaddress="192.168.1.3" port="48888" />
    <!-- DATA CENTER 2 -->
    <add key="DATA_CENTER_1_B" ipaddress="192.168.2.3" port="48888" />

</
replicatedServers>


This confguration way gives you the possibility to manage your servers with the smallest network latency. If you would use distributed caching in the same environment then I would suggest to create 2 caching pools. one in data center A and one in data center B which each one contains only 2 instead of 4. Sure you can also use all 4 servers but to be honest i havent seen much use cases where people need more then 4 GB of cache.

Apr 24, 2009 at 10:08 AM
Thanks for your reply. Did you mean that for storing 4 GB of cache in replicated cache we need 4 cache servers?

I have some confusion in the above sample configuration. Could you please explain more? What is data center concept?

Thanks and Regards
Shihab
Coordinator
Apr 24, 2009 at 12:02 PM
Hi Shihab,

i'm sorry if i made more confusion then I helped.

>> Did you mean that for storing 4 GB of cache in replicated cache we need 4 cache servers?
32-bit: max. 3 GB (needs a microsoft patch) otherwise max. 2 GB per server.
64-bit: max. 16 GB per server.

regards,
roni