How to build a DMA Failover cluster in a gepgraphical distant location and IP network, where there can be frequent network breaks between the Main and Standby clusters ? How can Database syncing issue be handled in such design ?
Hi Rajesh,
From the DB architecture options you have two options:
For STaaS you can contact staas@dataminer.services to set up the system for you and all you have to do is configure your DMS to connect towards STaaS.
For Dedicated cluster storage you will need to look into this option. Where you need at least two Cassandra nodes configured to be in two DCs and 3 OpenSearch nodes spread over three locations (on tie-breaker). This means you need at least 5 Linux machines for DB alone to have a geo-redundant DB layer. Then you can start scaling up on both DCs to increase the capacity. There is no easy way to determine how many nodes you would need upfront, but based on an existing system or based on a lot of questions we can make an educated guess.
In STaaS the network between the two sites is under control of the cloud provider and is generally very reliable and in case of any outages on the network it will be handled by the team managing the STaaS systems.
For dedicated clusters there are options to fine-tune the settings to cope as well as possible with network outages, but in general network outages between both sites that should be in sync is not advised and especially not if it is for longer timespans (more than one hour). In case you have an outage that was considerable long the system mostly needs manual attention to get everything back in sync after the network connection is restored between the two sites.