Distributed Data Show Episode 61
Amanda, David, and Adron talk about the challenges associated with moving their KillrVideo reference application to a multi-cloud cluster and the many considerations that come into play when going multi-cloud.
17:22 - “See where providers line up” what is the lowest common denominator
18:55 - Would latency be very different between providers as compared to within a single provider in any given region?
21:15 - “always know what everything is doing”
23:45 - OpsCenter allows homogenous view regardless of cloud provider
24:53 - Ensure to label and identify which providers your data centers and nodes are located in
25:35 - Why would someone span a cluster over multiple providers?
27:32 - Have to be cognizant on where load balancers, name servers, etc... are located
28:38 - Exploring some use cases for multi-cloud clusters
30:02 - Providers become competitors with their clients
30:26 - Moving application and database architecture can take months or years depending on size and complexity
31:06 - Security compliance, intellectual property, tracking, geographic location of data
33:41 - Any database or application needs to take data locality into consideration
34:03 - Replication inherent in Apache Cassandra makes moving data much easier
35:01 - Amanda starts wrapping things up from easy to difficult
36:58 - Things that used to take months or years take minutes today when thinking in terms of cloud compared to procuring hardware and installing software manually